This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make a missing right brace on \N{ fatal
[perl5.git] / regcomp.c
index 409ed24..ce4104a 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -2,7 +2,9 @@
  */
 
 /*
  */
 
 /*
- * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
+ * 'A fair jaw-cracker dwarf-language must be.'            --Samwise Gamgee
+ *
+ *     [p.285 of _The Lord of the Rings_, II/iii: "The Ring Goes South"]
  */
 
 /* This file contains functions for compiling a regular expression.  See
  */
 
 /* This file contains functions for compiling a regular expression.  See
@@ -57,7 +59,8 @@
  ****    Alterations to Henry's code are...
  ****
  ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
  ****    Alterations to Henry's code are...
  ****
  ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
- ****    2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others
+ ****    2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
+ ****    by Larry Wall and others
  ****
  ****    You may distribute under the terms of either the GNU General Public
  ****    License or the Artistic License, as specified in the README file.
  ****
  ****    You may distribute under the terms of either the GNU General Public
  ****    License or the Artistic License, as specified in the README file.
 typedef struct RExC_state_t {
     U32                flags;                  /* are we folding, multilining? */
     char       *precomp;               /* uncompiled string. */
 typedef struct RExC_state_t {
     U32                flags;                  /* are we folding, multilining? */
     char       *precomp;               /* uncompiled string. */
-    regexp     *rx;
+    REGEXP     *rx_sv;                 /* The SV that is the regexp. */
+    regexp     *rx;                    /* perl core regexp structure */
+    regexp_internal    *rxi;           /* internal data for regexp object pprivate field */        
     char       *start;                 /* Start of input for compile */
     char       *end;                   /* End of input for compile */
     char       *parse;                 /* Input-scan pointer. */
     I32                whilem_seen;            /* number of WHILEM in this expr */
     regnode    *emit_start;            /* Start of emitted-code area */
     char       *start;                 /* Start of input for compile */
     char       *end;                   /* End of input for compile */
     char       *parse;                 /* Input-scan pointer. */
     I32                whilem_seen;            /* number of WHILEM in this expr */
     regnode    *emit_start;            /* Start of emitted-code area */
+    regnode    *emit_bound;            /* First regnode outside of the allocated space */
     regnode    *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
     I32                naughty;                /* How bad is this pattern? */
     I32                sawback;                /* Did we see \1, ...? */
     regnode    *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
     I32                naughty;                /* How bad is this pattern? */
     I32                sawback;                /* Did we see \1, ...? */
@@ -122,9 +128,12 @@ typedef struct RExC_state_t {
     regnode    **open_parens;          /* pointers to open parens */
     regnode    **close_parens;         /* pointers to close parens */
     regnode    *opend;                 /* END node in program */
     regnode    **open_parens;          /* pointers to open parens */
     regnode    **close_parens;         /* pointers to close parens */
     regnode    *opend;                 /* END node in program */
-    I32                utf8;
-    HV         *charnames;             /* cache of named sequences */
+    I32                utf8;           /* whether the pattern is utf8 or not */
+    I32                orig_utf8;      /* whether the pattern was originally in utf8 */
+                               /* XXX use this for future optimisation of case
+                                * where pattern must be upgraded to utf8. */
     HV         *paren_names;           /* Paren names */
     HV         *paren_names;           /* Paren names */
+    
     regnode    **recurse;              /* Recurse regops */
     I32                recurse_count;          /* Number of recurse regops */
 #if ADD_TO_REGEXEC
     regnode    **recurse;              /* Recurse regops */
     I32                recurse_count;          /* Number of recurse regops */
 #if ADD_TO_REGEXEC
@@ -134,33 +143,39 @@ typedef struct RExC_state_t {
 #ifdef DEBUGGING
     const char  *lastparse;
     I32         lastnum;
 #ifdef DEBUGGING
     const char  *lastparse;
     I32         lastnum;
+    AV          *paren_name_list;       /* idx -> name */
 #define RExC_lastparse (pRExC_state->lastparse)
 #define RExC_lastnum   (pRExC_state->lastnum)
 #define RExC_lastparse (pRExC_state->lastparse)
 #define RExC_lastnum   (pRExC_state->lastnum)
+#define RExC_paren_name_list    (pRExC_state->paren_name_list)
 #endif
 } RExC_state_t;
 
 #define RExC_flags     (pRExC_state->flags)
 #define RExC_precomp   (pRExC_state->precomp)
 #endif
 } RExC_state_t;
 
 #define RExC_flags     (pRExC_state->flags)
 #define RExC_precomp   (pRExC_state->precomp)
+#define RExC_rx_sv     (pRExC_state->rx_sv)
 #define RExC_rx                (pRExC_state->rx)
 #define RExC_rx                (pRExC_state->rx)
+#define RExC_rxi       (pRExC_state->rxi)
 #define RExC_start     (pRExC_state->start)
 #define RExC_end       (pRExC_state->end)
 #define RExC_parse     (pRExC_state->parse)
 #define RExC_whilem_seen       (pRExC_state->whilem_seen)
 #define RExC_start     (pRExC_state->start)
 #define RExC_end       (pRExC_state->end)
 #define RExC_parse     (pRExC_state->parse)
 #define RExC_whilem_seen       (pRExC_state->whilem_seen)
-#define RExC_offsets   (pRExC_state->rx->offsets) /* I am not like the others */
+#ifdef RE_TRACK_PATTERN_OFFSETS
+#define RExC_offsets   (pRExC_state->rxi->u.offsets) /* I am not like the others */
+#endif
 #define RExC_emit      (pRExC_state->emit)
 #define RExC_emit_start        (pRExC_state->emit_start)
 #define RExC_emit      (pRExC_state->emit)
 #define RExC_emit_start        (pRExC_state->emit_start)
+#define RExC_emit_bound        (pRExC_state->emit_bound)
 #define RExC_naughty   (pRExC_state->naughty)
 #define RExC_sawback   (pRExC_state->sawback)
 #define RExC_seen      (pRExC_state->seen)
 #define RExC_size      (pRExC_state->size)
 #define RExC_npar      (pRExC_state->npar)
 #define RExC_naughty   (pRExC_state->naughty)
 #define RExC_sawback   (pRExC_state->sawback)
 #define RExC_seen      (pRExC_state->seen)
 #define RExC_size      (pRExC_state->size)
 #define RExC_npar      (pRExC_state->npar)
-#define RExC_cpar      (pRExC_state->cpar)
 #define RExC_nestroot   (pRExC_state->nestroot)
 #define RExC_extralen  (pRExC_state->extralen)
 #define RExC_seen_zerolen      (pRExC_state->seen_zerolen)
 #define RExC_seen_evals        (pRExC_state->seen_evals)
 #define RExC_utf8      (pRExC_state->utf8)
 #define RExC_nestroot   (pRExC_state->nestroot)
 #define RExC_extralen  (pRExC_state->extralen)
 #define RExC_seen_zerolen      (pRExC_state->seen_zerolen)
 #define RExC_seen_evals        (pRExC_state->seen_evals)
 #define RExC_utf8      (pRExC_state->utf8)
-#define RExC_charnames  (pRExC_state->charnames)
+#define RExC_orig_utf8 (pRExC_state->orig_utf8)
 #define RExC_open_parens       (pRExC_state->open_parens)
 #define RExC_close_parens      (pRExC_state->close_parens)
 #define RExC_opend     (pRExC_state->opend)
 #define RExC_open_parens       (pRExC_state->open_parens)
 #define RExC_close_parens      (pRExC_state->close_parens)
 #define RExC_opend     (pRExC_state->opend)
@@ -168,6 +183,7 @@ typedef struct RExC_state_t {
 #define RExC_recurse   (pRExC_state->recurse)
 #define RExC_recurse_count     (pRExC_state->recurse_count)
 
 #define RExC_recurse   (pRExC_state->recurse)
 #define RExC_recurse_count     (pRExC_state->recurse_count)
 
+
 #define        ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
 #define        ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
        ((*s) == '{' && regcurly(s)))
 #define        ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
 #define        ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
        ((*s) == '{' && regcurly(s)))
@@ -179,10 +195,11 @@ typedef struct RExC_state_t {
  * Flags to be passed up and down.
  */
 #define        WORST           0       /* Worst case. */
  * Flags to be passed up and down.
  */
 #define        WORST           0       /* Worst case. */
-#define        HASWIDTH        0x1     /* Known to match non-null strings. */
-#define        SIMPLE          0x2     /* Simple enough to be STAR/PLUS operand. */
-#define        SPSTART         0x4     /* Starts with * or +. */
-#define TRYAGAIN       0x8     /* Weeded out a declaration. */
+#define        HASWIDTH        0x01    /* Known to match non-null strings. */
+#define        SIMPLE          0x02    /* Simple enough to be STAR/PLUS operand. */
+#define        SPSTART         0x04    /* Starts with * or +. */
+#define TRYAGAIN       0x08    /* Weeded out a declaration. */
+#define POSTPONED      0x10    /* (?1),(?&name), (??{...}) or similar */
 
 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
 
 
 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
 
@@ -342,8 +359,8 @@ static const scan_data_t zero_scan_data =
 #define SCF_SEEN_ACCEPT         0x8000 
 
 #define UTF (RExC_utf8 != 0)
 #define SCF_SEEN_ACCEPT         0x8000 
 
 #define UTF (RExC_utf8 != 0)
-#define LOC ((RExC_flags & PMf_LOCALE) != 0)
-#define FOLD ((RExC_flags & PMf_FOLD) != 0)
+#define LOC ((RExC_flags & RXf_PMf_LOCALE) != 0)
+#define FOLD ((RExC_flags & RXf_PMf_FOLD) != 0)
 
 #define OOB_UNICODE            12345678
 #define OOB_NAMEDCLASS         -1
 
 #define OOB_UNICODE            12345678
 #define OOB_NAMEDCLASS         -1
@@ -370,21 +387,28 @@ static const scan_data_t zero_scan_data =
  * arg. Show regex, up to a maximum length. If it's too long, chop and add
  * "...".
  */
  * arg. Show regex, up to a maximum length. If it's too long, chop and add
  * "...".
  */
-#define        FAIL(msg) STMT_START {                                          \
+#define _FAIL(code) STMT_START {                                       \
     const char *ellipses = "";                                         \
     IV len = RExC_end - RExC_precomp;                                  \
                                                                        \
     if (!SIZE_ONLY)                                                    \
     const char *ellipses = "";                                         \
     IV len = RExC_end - RExC_precomp;                                  \
                                                                        \
     if (!SIZE_ONLY)                                                    \
-       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                      \
+       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv);                   \
     if (len > RegexLengthToShowInErrorMessages) {                      \
        /* chop 10 shorter than the max, to ensure meaning of "..." */  \
        len = RegexLengthToShowInErrorMessages - 10;                    \
        ellipses = "...";                                               \
     }                                                                  \
     if (len > RegexLengthToShowInErrorMessages) {                      \
        /* chop 10 shorter than the max, to ensure meaning of "..." */  \
        len = RegexLengthToShowInErrorMessages - 10;                    \
        ellipses = "...";                                               \
     }                                                                  \
-    Perl_croak(aTHX_ "%s in regex m/%.*s%s/",                          \
-           msg, (int)len, RExC_precomp, ellipses);                     \
+    code;                                                               \
 } STMT_END
 
 } STMT_END
 
+#define        FAIL(msg) _FAIL(                            \
+    Perl_croak(aTHX_ "%s in regex m/%.*s%s/",      \
+           msg, (int)len, RExC_precomp, ellipses))
+
+#define        FAIL2(msg,arg) _FAIL(                       \
+    Perl_croak(aTHX_ msg " in regex m/%.*s%s/",            \
+           arg, (int)len, RExC_precomp, ellipses))
+
 /*
  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
  */
 /*
  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
  */
@@ -399,7 +423,7 @@ static const scan_data_t zero_scan_data =
  */
 #define        vFAIL(m) STMT_START {                           \
     if (!SIZE_ONLY)                                    \
  */
 #define        vFAIL(m) STMT_START {                           \
     if (!SIZE_ONLY)                                    \
-       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);      \
+       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv);   \
     Simple_vFAIL(m);                                   \
 } STMT_END
 
     Simple_vFAIL(m);                                   \
 } STMT_END
 
@@ -417,7 +441,7 @@ static const scan_data_t zero_scan_data =
  */
 #define        vFAIL2(m,a1) STMT_START {                       \
     if (!SIZE_ONLY)                                    \
  */
 #define        vFAIL2(m,a1) STMT_START {                       \
     if (!SIZE_ONLY)                                    \
-       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);      \
+       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv);   \
     Simple_vFAIL2(m, a1);                              \
 } STMT_END
 
     Simple_vFAIL2(m, a1);                              \
 } STMT_END
 
@@ -436,7 +460,7 @@ static const scan_data_t zero_scan_data =
  */
 #define        vFAIL3(m,a1,a2) STMT_START {                    \
     if (!SIZE_ONLY)                                    \
  */
 #define        vFAIL3(m,a1,a2) STMT_START {                    \
     if (!SIZE_ONLY)                                    \
-       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);      \
+       SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv);   \
     Simple_vFAIL3(m, a1, a2);                          \
 } STMT_END
 
     Simple_vFAIL3(m, a1, a2);                          \
 } STMT_END
 
@@ -449,23 +473,22 @@ static const scan_data_t zero_scan_data =
            (int)offset, RExC_precomp, RExC_precomp + offset);  \
 } STMT_END
 
            (int)offset, RExC_precomp, RExC_precomp + offset);  \
 } STMT_END
 
-#define        vWARN(loc,m) STMT_START {                                       \
+#define        ckWARNreg(loc,m) STMT_START {                                   \
     const IV offset = loc - RExC_precomp;                              \
     const IV offset = loc - RExC_precomp;                              \
-    Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s" REPORT_LOCATION,     \
-           m, (int)offset, RExC_precomp, RExC_precomp + offset);       \
+    Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,     \
+           (int)offset, RExC_precomp, RExC_precomp + offset);          \
 } STMT_END
 
 } STMT_END
 
-#define        vWARNdep(loc,m) STMT_START {                                    \
+#define        ckWARNregdep(loc,m) STMT_START {                                \
     const IV offset = loc - RExC_precomp;                              \
     const IV offset = loc - RExC_precomp;                              \
-    Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP),         \
-           "%s" REPORT_LOCATION,                                       \
-           m, (int)offset, RExC_precomp, RExC_precomp + offset);       \
+    Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP),    \
+           m REPORT_LOCATION,                                          \
+           (int)offset, RExC_precomp, RExC_precomp + offset);          \
 } STMT_END
 
 } STMT_END
 
-
-#define        vWARN2(loc, m, a1) STMT_START {                                 \
+#define        ckWARN2reg(loc, m, a1) STMT_START {                             \
     const IV offset = loc - RExC_precomp;                              \
     const IV offset = loc - RExC_precomp;                              \
-    Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,                \
+    Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,     \
            a1, (int)offset, RExC_precomp, RExC_precomp + offset);      \
 } STMT_END
 
            a1, (int)offset, RExC_precomp, RExC_precomp + offset);      \
 } STMT_END
 
@@ -475,12 +498,24 @@ static const scan_data_t zero_scan_data =
            a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset);  \
 } STMT_END
 
            a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset);  \
 } STMT_END
 
+#define        ckWARN3reg(loc, m, a1, a2) STMT_START {                         \
+    const IV offset = loc - RExC_precomp;                              \
+    Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,     \
+           a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset);  \
+} STMT_END
+
 #define        vWARN4(loc, m, a1, a2, a3) STMT_START {                         \
     const IV offset = loc - RExC_precomp;                              \
     Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,                \
            a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
 } STMT_END
 
 #define        vWARN4(loc, m, a1, a2, a3) STMT_START {                         \
     const IV offset = loc - RExC_precomp;                              \
     Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,                \
            a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
 } STMT_END
 
+#define        ckWARN4reg(loc, m, a1, a2, a3) STMT_START {                     \
+    const IV offset = loc - RExC_precomp;                              \
+    Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,     \
+           a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
+} STMT_END
+
 #define        vWARN5(loc, m, a1, a2, a3, a4) STMT_START {                     \
     const IV offset = loc - RExC_precomp;                              \
     Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,                \
 #define        vWARN5(loc, m, a1, a2, a3, a4) STMT_START {                     \
     const IV offset = loc - RExC_precomp;                              \
     Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION,                \
@@ -499,7 +534,21 @@ static const scan_data_t zero_scan_data =
  * Element 0 holds the number n.
  * Position is 1 indexed.
  */
  * Element 0 holds the number n.
  * Position is 1 indexed.
  */
-
+#ifndef RE_TRACK_PATTERN_OFFSETS
+#define Set_Node_Offset_To_R(node,byte)
+#define Set_Node_Offset(node,byte)
+#define Set_Cur_Node_Offset
+#define Set_Node_Length_To_R(node,len)
+#define Set_Node_Length(node,len)
+#define Set_Node_Cur_Length(node)
+#define Node_Offset(n) 
+#define Node_Length(n) 
+#define Set_Node_Offset_Length(node,offset,len)
+#define ProgLen(ri) ri->u.proglen
+#define SetProgLen(ri,x) ri->u.proglen = x
+#else
+#define ProgLen(ri) ri->u.offsets[0]
+#define SetProgLen(ri,x) ri->u.offsets[0] = x
 #define Set_Node_Offset_To_R(node,byte) STMT_START {                   \
     if (! SIZE_ONLY) {                                                 \
        MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n",         \
 #define Set_Node_Offset_To_R(node,byte) STMT_START {                   \
     if (! SIZE_ONLY) {                                                 \
        MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n",         \
@@ -542,23 +591,24 @@ static const scan_data_t zero_scan_data =
     Set_Node_Offset_To_R((node)-RExC_emit_start, (offset));    \
     Set_Node_Length_To_R((node)-RExC_emit_start, (len));       \
 } STMT_END
     Set_Node_Offset_To_R((node)-RExC_emit_start, (offset));    \
     Set_Node_Length_To_R((node)-RExC_emit_start, (len));       \
 } STMT_END
-
+#endif
 
 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
 #define EXPERIMENTAL_INPLACESCAN
 
 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
 #define EXPERIMENTAL_INPLACESCAN
-#endif
+#endif /*RE_TRACK_PATTERN_OFFSETS*/
 
 
-#define DEBUG_STUDYDATA(data,depth)                                  \
-DEBUG_OPTIMISE_MORE_r(if(data){                                           \
+#define DEBUG_STUDYDATA(str,data,depth)                              \
+DEBUG_OPTIMISE_MORE_r(if(data){                                      \
     PerlIO_printf(Perl_debug_log,                                    \
     PerlIO_printf(Perl_debug_log,                                    \
-        "%*s"/* Len:%"IVdf"/%"IVdf" */"Pos:%"IVdf"/%"IVdf           \
-        " Flags: %"IVdf" Whilem_c: %"IVdf" Lcp: %"IVdf" ",           \
+        "%*s" str "Pos:%"IVdf"/%"IVdf                                \
+        " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s",       \
         (int)(depth)*2, "",                                          \
         (IV)((data)->pos_min),                                       \
         (IV)((data)->pos_delta),                                     \
         (int)(depth)*2, "",                                          \
         (IV)((data)->pos_min),                                       \
         (IV)((data)->pos_delta),                                     \
-        (IV)((data)->flags),                                         \
+        (UV)((data)->flags),                                         \
         (IV)((data)->whilem_c),                                      \
         (IV)((data)->whilem_c),                                      \
-        (IV)((data)->last_closep ? *((data)->last_closep) : -1)      \
+        (IV)((data)->last_closep ? *((data)->last_closep) : -1),     \
+        is_inf ? "INF " : ""                                         \
     );                                                               \
     if ((data)->last_found)                                          \
         PerlIO_printf(Perl_debug_log,                                \
     );                                                               \
     if ((data)->last_found)                                          \
         PerlIO_printf(Perl_debug_log,                                \
@@ -588,12 +638,14 @@ static void clear_re(pTHX_ void *r);
    floating substrings if needed. */
 
 STATIC void
    floating substrings if needed. */
 
 STATIC void
-S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp)
+S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp, int is_inf)
 {
     const STRLEN l = CHR_SVLEN(data->last_found);
     const STRLEN old_l = CHR_SVLEN(*data->longest);
     GET_RE_DEBUG_FLAGS_DECL;
 
 {
     const STRLEN l = CHR_SVLEN(data->last_found);
     const STRLEN old_l = CHR_SVLEN(*data->longest);
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_SCAN_COMMIT;
+
     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
        SvSetMagicSV(*data->longest, data->last_found);
        if (*data->longest == data->longest_fixed) {
     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
        SvSetMagicSV(*data->longest, data->last_found);
        if (*data->longest == data->longest_fixed) {
@@ -606,12 +658,12 @@ S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *min
            data->minlen_fixed=minlenp; 
            data->lookbehind_fixed=0;
        }
            data->minlen_fixed=minlenp; 
            data->lookbehind_fixed=0;
        }
-       else {
+       else { /* *data->longest == data->longest_float */
            data->offset_float_min = l ? data->last_start_min : data->pos_min;
            data->offset_float_max = (l
                                      ? data->last_start_max
                                      : data->pos_min + data->pos_delta);
            data->offset_float_min = l ? data->last_start_min : data->pos_min;
            data->offset_float_max = (l
                                      ? data->last_start_max
                                      : data->pos_min + data->pos_delta);
-           if ((U32)data->offset_float_max > (U32)I32_MAX)
+           if (is_inf || (U32)data->offset_float_max > (U32)I32_MAX)
                data->offset_float_max = I32_MAX;
            if (data->flags & SF_BEFORE_EOL)
                data->flags
                data->offset_float_max = I32_MAX;
            if (data->flags & SF_BEFORE_EOL)
                data->flags
@@ -633,13 +685,15 @@ S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *min
     }
     data->last_end = -1;
     data->flags &= ~SF_BEFORE_EOL;
     }
     data->last_end = -1;
     data->flags &= ~SF_BEFORE_EOL;
-    DEBUG_STUDYDATA(data,0);
+    DEBUG_STUDYDATA("commit: ",data,0);
 }
 
 /* Can match anything (initialization) */
 STATIC void
 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
 }
 
 /* Can match anything (initialization) */
 STATIC void
 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
+    PERL_ARGS_ASSERT_CL_ANYTHING;
+
     ANYOF_CLASS_ZERO(cl);
     ANYOF_BITMAP_SETALL(cl);
     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
     ANYOF_CLASS_ZERO(cl);
     ANYOF_BITMAP_SETALL(cl);
     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
@@ -653,6 +707,8 @@ S_cl_is_anything(const struct regnode_charclass_class *cl)
 {
     int value;
 
 {
     int value;
 
+    PERL_ARGS_ASSERT_CL_IS_ANYTHING;
+
     for (value = 0; value <= ANYOF_MAX; value += 2)
        if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
            return 1;
     for (value = 0; value <= ANYOF_MAX; value += 2)
        if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
            return 1;
@@ -667,6 +723,8 @@ S_cl_is_anything(const struct regnode_charclass_class *cl)
 STATIC void
 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
 STATIC void
 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
+    PERL_ARGS_ASSERT_CL_INIT;
+
     Zero(cl, 1, struct regnode_charclass_class);
     cl->type = ANYOF;
     cl_anything(pRExC_state, cl);
     Zero(cl, 1, struct regnode_charclass_class);
     cl->type = ANYOF;
     cl_anything(pRExC_state, cl);
@@ -675,6 +733,8 @@ S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 STATIC void
 S_cl_init_zero(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
 STATIC void
 S_cl_init_zero(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
 {
+    PERL_ARGS_ASSERT_CL_INIT_ZERO;
+
     Zero(cl, 1, struct regnode_charclass_class);
     cl->type = ANYOF;
     cl_anything(pRExC_state, cl);
     Zero(cl, 1, struct regnode_charclass_class);
     cl->type = ANYOF;
     cl_anything(pRExC_state, cl);
@@ -688,6 +748,7 @@ STATIC void
 S_cl_and(struct regnode_charclass_class *cl,
        const struct regnode_charclass_class *and_with)
 {
 S_cl_and(struct regnode_charclass_class *cl,
        const struct regnode_charclass_class *and_with)
 {
+    PERL_ARGS_ASSERT_CL_AND;
 
     assert(and_with->type == ANYOF);
     if (!(and_with->flags & ANYOF_CLASS)
 
     assert(and_with->type == ANYOF);
     if (!(and_with->flags & ANYOF_CLASS)
@@ -726,6 +787,8 @@ S_cl_and(struct regnode_charclass_class *cl,
 STATIC void
 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
 {
 STATIC void
 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
 {
+    PERL_ARGS_ASSERT_CL_OR;
+
     if (or_with->flags & ANYOF_INVERT) {
        /* We do not use
         * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
     if (or_with->flags & ANYOF_INVERT) {
        /* We do not use
         * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
@@ -789,9 +852,9 @@ S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, con
 
 #ifdef DEBUGGING
 /*
 
 #ifdef DEBUGGING
 /*
-   dump_trie(trie)
-   dump_trie_interim_list(trie,next_alloc)
-   dump_trie_interim_table(trie,next_alloc)
+   dump_trie(trie,widecharmap,revcharmap)
+   dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
+   dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
 
    These routines dump out a trie in a somewhat readable format.
    The _interim_ variants are used for debugging the interim
 
    These routines dump out a trie in a somewhat readable format.
    The _interim_ variants are used for debugging the interim
@@ -804,26 +867,27 @@ S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, con
 */
 
 /*
 */
 
 /*
-  dump_trie(trie)
   Dumps the final compressed table form of the trie to Perl_debug_log.
   Used for debugging make_trie().
 */
  
 STATIC void
   Dumps the final compressed table form of the trie to Perl_debug_log.
   Used for debugging make_trie().
 */
  
 STATIC void
-S_dump_trie(pTHX_ const struct _reg_trie_data *trie,U32 depth)
+S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
+           AV *revcharmap, U32 depth)
 {
     U32 state;
     SV *sv=sv_newmortal();
 {
     U32 state;
     SV *sv=sv_newmortal();
-    int colwidth= trie->widecharmap ? 6 : 4;
+    int colwidth= widecharmap ? 6 : 4;
     GET_RE_DEBUG_FLAGS_DECL;
 
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_DUMP_TRIE;
 
     PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
         (int)depth * 2 + 2,"",
         "Match","Base","Ofs" );
 
     for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
 
     PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
         (int)depth * 2 + 2,"",
         "Match","Base","Ofs" );
 
     for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
-       SV ** const tmp = av_fetch( trie->revcharmap, state, 0);
+       SV ** const tmp = av_fetch( revcharmap, state, 0);
         if ( tmp ) {
             PerlIO_printf( Perl_debug_log, "%*s", 
                 colwidth,
         if ( tmp ) {
             PerlIO_printf( Perl_debug_log, "%*s", 
                 colwidth,
@@ -885,19 +949,23 @@ S_dump_trie(pTHX_ const struct _reg_trie_data *trie,U32 depth)
     }
 }    
 /*
     }
 }    
 /*
-  dump_trie_interim_list(trie,next_alloc)
   Dumps a fully constructed but uncompressed trie in list form.
   List tries normally only are used for construction when the number of 
   possible chars (trie->uniquecharcount) is very high.
   Used for debugging make_trie().
 */
 STATIC void
   Dumps a fully constructed but uncompressed trie in list form.
   List tries normally only are used for construction when the number of 
   possible chars (trie->uniquecharcount) is very high.
   Used for debugging make_trie().
 */
 STATIC void
-S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie, U32 next_alloc,U32 depth)
+S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
+                        HV *widecharmap, AV *revcharmap, U32 next_alloc,
+                        U32 depth)
 {
     U32 state;
     SV *sv=sv_newmortal();
 {
     U32 state;
     SV *sv=sv_newmortal();
-    int colwidth= trie->widecharmap ? 6 : 4;
+    int colwidth= widecharmap ? 6 : 4;
     GET_RE_DEBUG_FLAGS_DECL;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_LIST;
+
     /* print out the table precompression.  */
     PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
         (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
     /* print out the table precompression.  */
     PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
         (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
@@ -916,7 +984,7 @@ S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie, U32 next_alloc
             );
         }
         for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
             );
         }
         for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
-           SV ** const tmp = av_fetch( trie->revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
+           SV ** const tmp = av_fetch( revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
            if ( tmp ) {
                 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
                     colwidth,
            if ( tmp ) {
                 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
                     colwidth,
@@ -938,20 +1006,23 @@ S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie, U32 next_alloc
 }    
 
 /*
 }    
 
 /*
-  dump_trie_interim_table(trie,next_alloc)
   Dumps a fully constructed but uncompressed trie in table form.
   This is the normal DFA style state transition table, with a few 
   twists to facilitate compression later. 
   Used for debugging make_trie().
 */
 STATIC void
   Dumps a fully constructed but uncompressed trie in table form.
   This is the normal DFA style state transition table, with a few 
   twists to facilitate compression later. 
   Used for debugging make_trie().
 */
 STATIC void
-S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie, U32 next_alloc, U32 depth)
+S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
+                         HV *widecharmap, AV *revcharmap, U32 next_alloc,
+                         U32 depth)
 {
     U32 state;
     U16 charid;
     SV *sv=sv_newmortal();
 {
     U32 state;
     U16 charid;
     SV *sv=sv_newmortal();
-    int colwidth= trie->widecharmap ? 6 : 4;
+    int colwidth= widecharmap ? 6 : 4;
     GET_RE_DEBUG_FLAGS_DECL;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_TABLE;
     
     /*
        print out the table precompression so that we can do a visual check
     
     /*
        print out the table precompression so that we can do a visual check
@@ -961,7 +1032,7 @@ S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie, U32 next_allo
     PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
 
     for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
     PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
 
     for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
-       SV ** const tmp = av_fetch( trie->revcharmap, charid, 0);
+       SV ** const tmp = av_fetch( revcharmap, charid, 0);
         if ( tmp ) {
             PerlIO_printf( Perl_debug_log, "%*s", 
                 colwidth,
         if ( tmp ) {
             PerlIO_printf( Perl_debug_log, "%*s", 
                 colwidth,
@@ -1125,11 +1196,19 @@ is the recommended Unicode-aware way of saying
 
 #define TRIE_STORE_REVCHAR                                                 \
     STMT_START {                                                           \
 
 #define TRIE_STORE_REVCHAR                                                 \
     STMT_START {                                                           \
-       SV *tmp = newSVpvs("");                                            \
-       if (UTF) SvUTF8_on(tmp);                                           \
-       Perl_sv_catpvf( aTHX_ tmp, "%c", (int)uvc );                       \
-       av_push( TRIE_REVCHARMAP(trie), tmp );                             \
-    } STMT_END
+       if (UTF) {                                                         \
+           SV *zlopp = newSV(2);                                          \
+           unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp);      \
+           unsigned const char *const kapow = uvuni_to_utf8(flrbbbbb, uvc & 0xFF); \
+           SvCUR_set(zlopp, kapow - flrbbbbb);                            \
+           SvPOK_on(zlopp);                                               \
+           SvUTF8_on(zlopp);                                              \
+           av_push(revcharmap, zlopp);                                    \
+       } else {                                                           \
+           char ooooff = (char)uvc;                                               \
+           av_push(revcharmap, newSVpvn(&ooooff, 1));                     \
+       }                                                                  \
+        } STMT_END
 
 #define TRIE_READ_CHAR STMT_START {                                           \
     wordlen++;                                                                \
 
 #define TRIE_READ_CHAR STMT_START {                                           \
     wordlen++;                                                                \
@@ -1184,18 +1263,17 @@ is the recommended Unicode-aware way of saying
         /* store the word for dumping */                        \
         SV* tmp;                                                \
         if (OP(noper) != NOTHING)                               \
         /* store the word for dumping */                        \
         SV* tmp;                                                \
         if (OP(noper) != NOTHING)                               \
-            tmp = newSVpvn(STRING(noper), STR_LEN(noper));      \
+            tmp = newSVpvn_utf8(STRING(noper), STR_LEN(noper), UTF);   \
         else                                                    \
         else                                                    \
-            tmp = newSVpvn( "", 0 );                            \
-        if ( UTF ) SvUTF8_on( tmp );                            \
-        av_push( trie->words, tmp );                            \
+            tmp = newSVpvn_utf8( "", 0, UTF );                 \
+        av_push( trie_words, tmp );                             \
     });                                                         \
                                                                 \
     curword++;                                                  \
                                                                 \
     if ( noper_next < tail ) {                                  \
         if (!trie->jump)                                        \
     });                                                         \
                                                                 \
     curword++;                                                  \
                                                                 \
     if ( noper_next < tail ) {                                  \
         if (!trie->jump)                                        \
-            Newxz( trie->jump, word_count + 1, U16);            \
+            trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, sizeof(U16) ); \
         trie->jump[curword] = (U16)(noper_next - convert);      \
         if (!jumper)                                            \
             jumper = noper_next;                                \
         trie->jump[curword] = (U16)(noper_next - convert);      \
         if (!jumper)                                            \
             jumper = noper_next;                                \
@@ -1209,7 +1287,8 @@ is the recommended Unicode-aware way of saying
         /* we only allocate the nextword buffer when there    */\
         /* a dupe, so first time we have to do the allocation */\
         if (!trie->nextword)                                    \
         /* we only allocate the nextword buffer when there    */\
         /* a dupe, so first time we have to do the allocation */\
         if (!trie->nextword)                                    \
-            Newxz( trie->nextword, word_count + 1, U16);        \
+            trie->nextword = (U16 *)                                   \
+               PerlMemShared_calloc( word_count + 1, sizeof(U16));     \
         while ( trie->nextword[dupe] )                          \
             dupe= trie->nextword[dupe];                         \
         trie->nextword[dupe]= curword;                          \
         while ( trie->nextword[dupe] )                          \
             dupe= trie->nextword[dupe];                         \
         trie->nextword[dupe]= curword;                          \
@@ -1239,6 +1318,8 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
     dVAR;
     /* first pass, loop through and scan words */
     reg_trie_data *trie;
     dVAR;
     /* first pass, loop through and scan words */
     reg_trie_data *trie;
+    HV *widecharmap = NULL;
+    AV *revcharmap = newAV();
     regnode *cur;
     const U32 uniflags = UTF8_ALLOW_DEFAULT;
     STRLEN len = 0;
     regnode *cur;
     const U32 uniflags = UTF8_ALLOW_DEFAULT;
     STRLEN len = 0;
@@ -1257,32 +1338,35 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                          )
                      );
 
                          )
                      );
 
-    const U32 data_slot = add_data( pRExC_state, 1, "t" );
-    SV *re_trie_maxbuff;
-#ifndef DEBUGGING
-    /* these are only used during construction but are useful during
-     * debugging so we store them in the struct when debugging.
+#ifdef DEBUGGING
+    const U32 data_slot = add_data( pRExC_state, 4, "tuuu" );
+    AV *trie_words = NULL;
+    /* along with revcharmap, this only used during construction but both are
+     * useful during debugging so we store them in the struct when debugging.
      */
      */
+#else
+    const U32 data_slot = add_data( pRExC_state, 2, "tu" );
     STRLEN trie_charcount=0;
     STRLEN trie_charcount=0;
-    AV *trie_revcharmap;
 #endif
 #endif
+    SV *re_trie_maxbuff;
     GET_RE_DEBUG_FLAGS_DECL;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_MAKE_TRIE;
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 
-    Newxz( trie, 1, reg_trie_data );
+    trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
     trie->refcount = 1;
     trie->startstate = 1;
     trie->wordcount = word_count;
     trie->refcount = 1;
     trie->startstate = 1;
     trie->wordcount = word_count;
-    RExC_rx->data->data[ data_slot ] = (void*)trie;
-    Newxz( trie->charmap, 256, U16 );
+    RExC_rxi->data->data[ data_slot ] = (void*)trie;
+    trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
     if (!(UTF && folder))
     if (!(UTF && folder))
-        Newxz( trie->bitmap, ANYOF_BITMAP_SIZE, char );
+       trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
     DEBUG_r({
     DEBUG_r({
-        trie->words = newAV();
+        trie_words = newAV();
     });
     });
-    TRIE_REVCHARMAP(trie) = newAV();
 
     re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
     if (!SvIOK(re_trie_maxbuff)) {
 
     re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
     if (!SvIOK(re_trie_maxbuff)) {
@@ -1314,7 +1398,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
        have unique chars.
 
        We use an array of integers to represent the character codes 0..255
        have unique chars.
 
        We use an array of integers to represent the character codes 0..255
-       (trie->charmap) and we use a an HV* to store unicode characters. We use the
+       (trie->charmap) and we use a an HV* to store Unicode characters. We use the
        native representation of the character value as the key and IV's for the
        coded index.
 
        native representation of the character value as the key and IV's for the
        coded index.
 
@@ -1337,16 +1421,17 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
         const U8 *scan = (U8*)NULL;
         U32 wordlen      = 0;         /* required init */
         U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
         const U8 *scan = (U8*)NULL;
         U32 wordlen      = 0;         /* required init */
-        STRLEN chars=0;
+        STRLEN chars = 0;
+        bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the bitmap?*/
 
         if (OP(noper) == NOTHING) {
             trie->minlen= 0;
             continue;
         }
 
         if (OP(noper) == NOTHING) {
             trie->minlen= 0;
             continue;
         }
-        if (trie->bitmap) {
-            TRIE_BITMAP_SET(trie,*uc);
-            if ( folder ) TRIE_BITMAP_SET(trie,folder[ *uc ]);            
-        }
+        if ( set_bit ) /* bitmap only alloced when !(UTF&&Folding) */
+            TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
+                                          regardless of encoding */
+
         for ( ; uc < e ; uc += len ) {
             TRIE_CHARCOUNT(trie)++;
             TRIE_READ_CHAR;
         for ( ; uc < e ; uc += len ) {
             TRIE_CHARCOUNT(trie)++;
             TRIE_READ_CHAR;
@@ -1358,12 +1443,32 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                         trie->charmap[ folder[ uvc ] ] = trie->charmap[ uvc ];
                     TRIE_STORE_REVCHAR;
                 }
                         trie->charmap[ folder[ uvc ] ] = trie->charmap[ uvc ];
                     TRIE_STORE_REVCHAR;
                 }
+                if ( set_bit ) {
+                    /* store the codepoint in the bitmap, and if its ascii
+                       also store its folded equivelent. */
+                    TRIE_BITMAP_SET(trie,uvc);
+
+                   /* store the folded codepoint */
+                   if ( folder ) TRIE_BITMAP_SET(trie,folder[ uvc ]);
+
+                   if ( !UTF ) {
+                       /* store first byte of utf8 representation of
+                          codepoints in the 127 < uvc < 256 range */
+                       if (127 < uvc && uvc < 192) {
+                           TRIE_BITMAP_SET(trie,194);
+                       } else if (191 < uvc ) {
+                           TRIE_BITMAP_SET(trie,195);
+                       /* && uvc < 256 -- we know uvc is < 256 already */
+                       }
+                   }
+                    set_bit = 0; /* We've done our bit :-) */
+                }
             } else {
                 SV** svpp;
             } else {
                 SV** svpp;
-                if ( !trie->widecharmap )
-                    trie->widecharmap = newHV();
+                if ( !widecharmap )
+                    widecharmap = newHV();
 
 
-                svpp = hv_fetch( trie->widecharmap, (char*)&uvc, sizeof( UV ), 1 );
+                svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
 
                 if ( !svpp )
                     Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
 
                 if ( !svpp )
                     Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
@@ -1387,11 +1492,11 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
     DEBUG_TRIE_COMPILE_r(
         PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
                 (int)depth * 2 + 2,"",
     DEBUG_TRIE_COMPILE_r(
         PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
                 (int)depth * 2 + 2,"",
-                ( trie->widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
+                ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
                (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
                (int)trie->minlen, (int)trie->maxlen )
     );
                (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
                (int)trie->minlen, (int)trie->maxlen )
     );
-    Newxz( trie->wordlen, word_count, U32 );
+    trie->wordlen = (U32 *) PerlMemShared_calloc( word_count, sizeof(U32) );
 
     /*
         We now know what we are dealing with in terms of unique chars and
 
     /*
         We now know what we are dealing with in terms of unique chars and
@@ -1433,8 +1538,10 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log, 
             "%*sCompiling trie using list compiler\n",
             (int)depth * 2 + 2, ""));
         DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log, 
             "%*sCompiling trie using list compiler\n",
             (int)depth * 2 + 2, ""));
-
-        Newxz( trie->states, TRIE_CHARCOUNT(trie) + 2, reg_trie_state );
+       
+       trie->states = (reg_trie_state *)
+           PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
+                                 sizeof(reg_trie_state) );
         TRIE_LIST_NEW(1);
         next_alloc = 2;
 
         TRIE_LIST_NEW(1);
         next_alloc = 2;
 
@@ -1458,7 +1565,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                     if ( uvc < 256 ) {
                         charid = trie->charmap[ uvc ];
                    } else {
                     if ( uvc < 256 ) {
                         charid = trie->charmap[ uvc ];
                    } else {
-                        SV** const svpp = hv_fetch( trie->widecharmap, (char*)&uvc, sizeof( UV ), 0);
+                        SV** const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
                         if ( !svpp ) {
                             charid = 0;
                         } else {
                         if ( !svpp ) {
                             charid = 0;
                         } else {
@@ -1498,14 +1605,19 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
 
         /* next alloc is the NEXT state to be allocated */
         trie->statecount = next_alloc; 
 
         /* next alloc is the NEXT state to be allocated */
         trie->statecount = next_alloc; 
-        Renew( trie->states, next_alloc, reg_trie_state );
+        trie->states = (reg_trie_state *)
+           PerlMemShared_realloc( trie->states,
+                                  next_alloc
+                                  * sizeof(reg_trie_state) );
 
         /* and now dump it out before we compress it */
 
         /* and now dump it out before we compress it */
-        DEBUG_TRIE_COMPILE_MORE_r(
-            dump_trie_interim_list(trie,next_alloc,depth+1)
+        DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
+                                                        revcharmap, next_alloc,
+                                                        depth+1)
         );
 
         );
 
-        Newxz( trie->trans, transcount ,reg_trie_trans );
+        trie->trans = (reg_trie_trans *)
+           PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
         {
             U32 state;
             U32 tp = 0;
         {
             U32 state;
             U32 tp = 0;
@@ -1536,7 +1648,10 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                     }
                     if ( transcount < tp + maxid - minid + 1) {
                         transcount *= 2;
                     }
                     if ( transcount < tp + maxid - minid + 1) {
                         transcount *= 2;
-                        Renew( trie->trans, transcount, reg_trie_trans );
+                       trie->trans = (reg_trie_trans *)
+                           PerlMemShared_realloc( trie->trans,
+                                                    transcount
+                                                    * sizeof(reg_trie_trans) );
                         Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
                     }
                     base = trie->uniquecharcount + tp - minid;
                         Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
                     }
                     base = trie->uniquecharcount + tp - minid;
@@ -1616,9 +1731,13 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
             "%*sCompiling trie using table compiler\n",
             (int)depth * 2 + 2, ""));
 
             "%*sCompiling trie using table compiler\n",
             (int)depth * 2 + 2, ""));
 
-        Newxz( trie->trans, ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1,
-              reg_trie_trans );
-        Newxz( trie->states, TRIE_CHARCOUNT(trie) + 2, reg_trie_state );
+       trie->trans = (reg_trie_trans *)
+           PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
+                                 * trie->uniquecharcount + 1,
+                                 sizeof(reg_trie_trans) );
+        trie->states = (reg_trie_state *)
+           PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
+                                 sizeof(reg_trie_state) );
         next_alloc = trie->uniquecharcount + 1;
 
 
         next_alloc = trie->uniquecharcount + 1;
 
 
@@ -1646,7 +1765,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                     if ( uvc < 256 ) {
                         charid = trie->charmap[ uvc ];
                     } else {
                     if ( uvc < 256 ) {
                         charid = trie->charmap[ uvc ];
                     } else {
-                        SV* const * const svpp = hv_fetch( trie->widecharmap, (char*)&uvc, sizeof( UV ), 0);
+                        SV* const * const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
                         charid = svpp ? (U16)SvIV(*svpp) : 0;
                     }
                     if ( charid ) {
                         charid = svpp ? (U16)SvIV(*svpp) : 0;
                     }
                     if ( charid ) {
@@ -1669,9 +1788,9 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         } /* end second pass */
 
         /* and now dump it out before we compress it */
         } /* end second pass */
 
         /* and now dump it out before we compress it */
-        DEBUG_TRIE_COMPILE_MORE_r(
-            dump_trie_interim_table(trie,next_alloc,depth+1)
-        );
+        DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
+                                                         revcharmap,
+                                                         next_alloc, depth+1));
 
         {
         /*
 
         {
         /*
@@ -1775,7 +1894,9 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
             }
         }
         trie->lasttrans = pos + 1;
             }
         }
         trie->lasttrans = pos + 1;
-        Renew( trie->states, laststate, reg_trie_state);
+        trie->states = (reg_trie_state *)
+           PerlMemShared_realloc( trie->states, laststate
+                                  * sizeof(reg_trie_state) );
         DEBUG_TRIE_COMPILE_MORE_r(
                 PerlIO_printf( Perl_debug_log,
                    "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
         DEBUG_TRIE_COMPILE_MORE_r(
                 PerlIO_printf( Perl_debug_log,
                    "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
@@ -1795,12 +1916,12 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                 (UV)trie->lasttrans)
     );
     /* resize the trans array to remove unused space */
                 (UV)trie->lasttrans)
     );
     /* resize the trans array to remove unused space */
-    Renew( trie->trans, trie->lasttrans, reg_trie_trans);
+    trie->trans = (reg_trie_trans *)
+       PerlMemShared_realloc( trie->trans, trie->lasttrans
+                              * sizeof(reg_trie_trans) );
 
     /* and now dump out the compressed format */
 
     /* and now dump out the compressed format */
-    DEBUG_TRIE_COMPILE_r(
-        dump_trie(trie,depth+1)
-    );
+    DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
 
     {   /* Modify the program and insert the new TRIE node*/ 
         U8 nodetype =(U8)(flags & 0xFF);
 
     {   /* Modify the program and insert the new TRIE node*/ 
         U8 nodetype =(U8)(flags & 0xFF);
@@ -1808,9 +1929,12 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         
 #ifdef DEBUGGING
         regnode *optimize = NULL;
         
 #ifdef DEBUGGING
         regnode *optimize = NULL;
+#ifdef RE_TRACK_PATTERN_OFFSETS
+
         U32 mjd_offset = 0;
         U32 mjd_nodelen = 0;
         U32 mjd_offset = 0;
         U32 mjd_nodelen = 0;
-#endif
+#endif /* RE_TRACK_PATTERN_OFFSETS */
+#endif /* DEBUGGING */
         /*
            This means we convert either the first branch or the first Exact,
            depending on whether the thing following (in 'last') is a branch
         /*
            This means we convert either the first branch or the first Exact,
            depending on whether the thing following (in 'last') is a branch
@@ -1823,29 +1947,32 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         if ( first != startbranch || OP( last ) == BRANCH ) {
             /* branch sub-chain */
             NEXT_OFF( first ) = (U16)(last - first);
         if ( first != startbranch || OP( last ) == BRANCH ) {
             /* branch sub-chain */
             NEXT_OFF( first ) = (U16)(last - first);
+#ifdef RE_TRACK_PATTERN_OFFSETS
             DEBUG_r({
                 mjd_offset= Node_Offset((convert));
                 mjd_nodelen= Node_Length((convert));
             });
             DEBUG_r({
                 mjd_offset= Node_Offset((convert));
                 mjd_nodelen= Node_Length((convert));
             });
+#endif
             /* whole branch chain */
             /* whole branch chain */
-        } else {
+        }
+#ifdef RE_TRACK_PATTERN_OFFSETS
+        else {
             DEBUG_r({
                 const  regnode *nop = NEXTOPER( convert );
                 mjd_offset= Node_Offset((nop));
                 mjd_nodelen= Node_Length((nop));
             });
         }
             DEBUG_r({
                 const  regnode *nop = NEXTOPER( convert );
                 mjd_offset= Node_Offset((nop));
                 mjd_nodelen= Node_Length((nop));
             });
         }
-        
         DEBUG_OPTIMISE_r(
             PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
                 (int)depth * 2 + 2, "",
                 (UV)mjd_offset, (UV)mjd_nodelen)
         );
         DEBUG_OPTIMISE_r(
             PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
                 (int)depth * 2 + 2, "",
                 (UV)mjd_offset, (UV)mjd_nodelen)
         );
-
+#endif
         /* But first we check to see if there is a common prefix we can 
            split out as an EXACT and put in front of the TRIE node.  */
         trie->startstate= 1;
         /* But first we check to see if there is a common prefix we can 
            split out as an EXACT and put in front of the TRIE node.  */
         trie->startstate= 1;
-        if ( trie->bitmap && !trie->widecharmap && !trie->jump  ) {
+        if ( trie->bitmap && !widecharmap && !trie->jump  ) {
             U32 state;
             for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
                 U32 ofs = 0;
             U32 state;
             for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
                 U32 ofs = 0;
@@ -1862,7 +1989,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                          trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
                     {
                         if ( ++count > 1 ) {
                          trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
                     {
                         if ( ++count > 1 ) {
-                            SV **tmp = av_fetch( TRIE_REVCHARMAP(trie), ofs, 0);
+                            SV **tmp = av_fetch( revcharmap, ofs, 0);
                            const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
                             if ( state == 1 ) break;
                             if ( count == 2 ) {
                            const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
                             if ( state == 1 ) break;
                             if ( count == 2 ) {
@@ -1873,14 +2000,14 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                                         (int)depth * 2 + 2, "",
                                         (UV)state));
                                if (idx >= 0) {
                                         (int)depth * 2 + 2, "",
                                         (UV)state));
                                if (idx >= 0) {
-                                   SV ** const tmp = av_fetch( TRIE_REVCHARMAP(trie), idx, 0);
+                                   SV ** const tmp = av_fetch( revcharmap, idx, 0);
                                    const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
 
                                     TRIE_BITMAP_SET(trie,*ch);
                                     if ( folder )
                                         TRIE_BITMAP_SET(trie, folder[ *ch ]);
                                     DEBUG_OPTIMISE_r(
                                    const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
 
                                     TRIE_BITMAP_SET(trie,*ch);
                                     if ( folder )
                                         TRIE_BITMAP_SET(trie, folder[ *ch ]);
                                     DEBUG_OPTIMISE_r(
-                                        PerlIO_printf(Perl_debug_log, (char*)ch)
+                                        PerlIO_printf(Perl_debug_log, "%s", (char*)ch)
                                     );
                                }
                            }
                                     );
                                }
                            }
@@ -1893,22 +2020,30 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                    }
                 }
                 if ( count == 1 ) {
                    }
                 }
                 if ( count == 1 ) {
-                    SV **tmp = av_fetch( TRIE_REVCHARMAP(trie), idx, 0);
-                    const char *ch = SvPV_nolen_const( *tmp );
-                    DEBUG_OPTIMISE_r(
+                    SV **tmp = av_fetch( revcharmap, idx, 0);
+                    STRLEN len;
+                    char *ch = SvPV( *tmp, len );
+                    DEBUG_OPTIMISE_r({
+                        SV *sv=sv_newmortal();
                         PerlIO_printf( Perl_debug_log,
                            "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
                             (int)depth * 2 + 2, "",
                         PerlIO_printf( Perl_debug_log,
                            "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
                             (int)depth * 2 + 2, "",
-                            (UV)state, (UV)idx, ch)
-                    );
+                            (UV)state, (UV)idx, 
+                            pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6, 
+                               PL_colors[0], PL_colors[1],
+                               (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
+                               PERL_PV_ESCAPE_FIRSTCHAR 
+                            )
+                        );
+                    });
                     if ( state==1 ) {
                         OP( convert ) = nodetype;
                         str=STRING(convert);
                         STR_LEN(convert)=0;
                     }
                     if ( state==1 ) {
                         OP( convert ) = nodetype;
                         str=STRING(convert);
                         STR_LEN(convert)=0;
                     }
-                    *str++=*ch;
-                    STR_LEN(convert)++;
-
+                    STR_LEN(convert) += len;
+                    while (len--)
+                        *str++ = *ch++;
                } else {
 #ifdef DEBUGGING           
                    if (state>1)
                } else {
 #ifdef DEBUGGING           
                    if (state>1)
@@ -1923,14 +2058,35 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
                 trie->startstate = state;
                 trie->minlen -= (state - 1);
                 trie->maxlen -= (state - 1);
                 trie->startstate = state;
                 trie->minlen -= (state - 1);
                 trie->maxlen -= (state - 1);
-                DEBUG_r({
-                    regnode *fix = convert;
-                    mjd_nodelen++;
-                    Set_Node_Offset_Length(convert, mjd_offset, state - 1);
-                    while( ++fix < n ) {
-                        Set_Node_Offset_Length(fix, 0, 0);
-                    }
-                });
+#ifdef DEBUGGING
+               /* At least the UNICOS C compiler choked on this
+                * being argument to DEBUG_r(), so let's just have
+                * it right here. */
+               if (
+#ifdef PERL_EXT_RE_BUILD
+                   1
+#else
+                   DEBUG_r_TEST
+#endif
+                   ) {
+                   regnode *fix = convert;
+                   U32 word = trie->wordcount;
+                   mjd_nodelen++;
+                   Set_Node_Offset_Length(convert, mjd_offset, state - 1);
+                   while( ++fix < n ) {
+                       Set_Node_Offset_Length(fix, 0, 0);
+                   }
+                   while (word--) {
+                       SV ** const tmp = av_fetch( trie_words, word, 0 );
+                       if (tmp) {
+                           if ( STR_LEN(convert) <= SvCUR(*tmp) )
+                               sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
+                           else
+                               sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
+                       }
+                   }
+               }
+#endif
                 if (trie->maxlen) {
                     convert = n;
                } else {
                 if (trie->maxlen) {
                     convert = n;
                } else {
@@ -1956,7 +2112,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
             {
                 OP( convert ) = TRIEC;
                 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
             {
                 OP( convert ) = TRIEC;
                 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
-                Safefree(trie->bitmap);
+                PerlMemShared_free(trie->bitmap);
                 trie->bitmap= NULL;
             } else 
                 OP( convert ) = TRIE;
                 trie->bitmap= NULL;
             } else 
                 OP( convert ) = TRIE;
@@ -1974,6 +2130,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
         /* needed for dumping*/
         DEBUG_r(if (optimize) {
             regnode *opt = convert;
         /* needed for dumping*/
         DEBUG_r(if (optimize) {
             regnode *opt = convert;
+
             while ( ++opt < optimize) {
                 Set_Node_Offset_Length(opt,0,0);
             }
             while ( ++opt < optimize) {
                 Set_Node_Offset_Length(opt,0,0);
             }
@@ -1990,8 +2147,12 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *firs
             Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
         });
     } /* end node insert */
             Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
         });
     } /* end node insert */
-#ifndef DEBUGGING
-    SvREFCNT_dec(TRIE_REVCHARMAP(trie));
+    RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
+#ifdef DEBUGGING
+    RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
+    RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
+#else
+    SvREFCNT_dec(revcharmap);
 #endif
     return trie->jump 
            ? MADE_JUMP_TRIE 
 #endif
     return trie->jump 
            ? MADE_JUMP_TRIE 
@@ -2021,7 +2182,8 @@ S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source,  regnode
    try 'g' and succeed, prodceding to match 'cdgu'.
  */
  /* add a fail transition */
    try 'g' and succeed, prodceding to match 'cdgu'.
  */
  /* add a fail transition */
-    reg_trie_data *trie=(reg_trie_data *)RExC_rx->data->data[ARG(source)];
+    const U32 trie_offset = ARG(source);
+    reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
     U32 *q;
     const U32 ucharcount = trie->uniquecharcount;
     const U32 numstates = trie->statecount;
     U32 *q;
     const U32 ucharcount = trie->uniquecharcount;
     const U32 numstates = trie->statecount;
@@ -2034,19 +2196,21 @@ S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source,  regnode
     reg_ac_data *aho;
     const U32 data_slot = add_data( pRExC_state, 1, "T" );
     GET_RE_DEBUG_FLAGS_DECL;
     reg_ac_data *aho;
     const U32 data_slot = add_data( pRExC_state, 1, "T" );
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_MAKE_TRIE_FAILTABLE;
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 
 
     ARG_SET( stclass, data_slot );
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 
 
     ARG_SET( stclass, data_slot );
-    Newxz( aho, 1, reg_ac_data );
-    RExC_rx->data->data[ data_slot ] = (void*)aho;
-    aho->trie=trie;
-    aho->states=(reg_trie_state *)savepvn((const char*)trie->states,
-        numstates * sizeof(reg_trie_state));
+    aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
+    RExC_rxi->data->data[ data_slot ] = (void*)aho;
+    aho->trie=trie_offset;
+    aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
+    Copy( trie->states, aho->states, numstates, reg_trie_state );
     Newxz( q, numstates, U32);
     Newxz( q, numstates, U32);
-    Newxz( aho->fail, numstates, U32 );
+    aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
     aho->refcount = 1;
     fail = aho->fail;
     /* initialize fail[0..1] to be 1 so that we always have
     aho->refcount = 1;
     fail = aho->fail;
     /* initialize fail[0..1] to be 1 so that we always have
@@ -2150,6 +2314,8 @@ S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan, I32 *min, U32 flags
 #else
     PERL_UNUSED_ARG(depth);
 #endif
 #else
     PERL_UNUSED_ARG(depth);
 #endif
+
+    PERL_ARGS_ASSERT_JOIN_EXACT;
 #ifndef EXPERIMENTAL_INPLACESCAN
     PERL_UNUSED_ARG(flags);
     PERL_UNUSED_ARG(val);
 #ifndef EXPERIMENTAL_INPLACESCAN
     PERL_UNUSED_ARG(flags);
     PERL_UNUSED_ARG(val);
@@ -2300,6 +2466,37 @@ typedef struct scan_frame {
     I32 stop; /* what stopparen do we use */
 } scan_frame;
 
     I32 stop; /* what stopparen do we use */
 } scan_frame;
 
+
+#define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
+
+#define CASE_SYNST_FNC(nAmE)                                       \
+case nAmE:                                                         \
+    if (flags & SCF_DO_STCLASS_AND) {                              \
+           for (value = 0; value < 256; value++)                  \
+               if (!is_ ## nAmE ## _cp(value))                       \
+                   ANYOF_BITMAP_CLEAR(data->start_class, value);  \
+    }                                                              \
+    else {                                                         \
+           for (value = 0; value < 256; value++)                  \
+               if (is_ ## nAmE ## _cp(value))                        \
+                   ANYOF_BITMAP_SET(data->start_class, value);    \
+    }                                                              \
+    break;                                                         \
+case N ## nAmE:                                                    \
+    if (flags & SCF_DO_STCLASS_AND) {                              \
+           for (value = 0; value < 256; value++)                   \
+               if (is_ ## nAmE ## _cp(value))                         \
+                   ANYOF_BITMAP_CLEAR(data->start_class, value);   \
+    }                                                               \
+    else {                                                          \
+           for (value = 0; value < 256; value++)                   \
+               if (!is_ ## nAmE ## _cp(value))                        \
+                   ANYOF_BITMAP_SET(data->start_class, value);     \
+    }                                                               \
+    break
+
+
+
 STATIC I32
 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                         I32 *minlenp, I32 *deltap,
 STATIC I32
 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                         I32 *minlenp, I32 *deltap,
@@ -2329,9 +2526,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
     regnode *first_non_open = scan;
     I32 stopmin = I32_MAX;
     scan_frame *frame = NULL;
     regnode *first_non_open = scan;
     I32 stopmin = I32_MAX;
     scan_frame *frame = NULL;
-
     GET_RE_DEBUG_FLAGS_DECL;
 
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_STUDY_CHUNK;
+
 #ifdef DEBUGGING
     StructCopy(&zero_scan_data, &data_fake, scan_data_t);
 #endif
 #ifdef DEBUGGING
     StructCopy(&zero_scan_data, &data_fake, scan_data_t);
 #endif
@@ -2345,7 +2543,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
   fake_study_recurse:
     while ( scan && OP(scan) != END && scan < last ){
        /* Peephole optimizer: */
   fake_study_recurse:
     while ( scan && OP(scan) != END && scan < last ){
        /* Peephole optimizer: */
-       DEBUG_STUDYDATA(data,depth);
+       DEBUG_STUDYDATA("Peep:", data,depth);
        DEBUG_PEEP("Peep",scan,depth);
         JOIN_EXACT(scan,&min,0);
 
        DEBUG_PEEP("Peep",scan,depth);
         JOIN_EXACT(scan,&min,0);
 
@@ -2391,7 +2589,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                regnode * const startbranch=scan;
                
                if (flags & SCF_DO_SUBSTR)
                regnode * const startbranch=scan;
                
                if (flags & SCF_DO_SUBSTR)
-                   scan_commit(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
+                   SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
                if (flags & SCF_DO_STCLASS)
                    cl_init_zero(pRExC_state, &accum);
 
                if (flags & SCF_DO_STCLASS)
                    cl_init_zero(pRExC_state, &accum);
 
@@ -2407,6 +2605,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    }
                    else
                        data_fake.last_closep = &fake;
                    }
                    else
                        data_fake.last_closep = &fake;
+
+                   data_fake.pos_delta = delta;
                    next = regnext(scan);
                    scan = NEXTOPER(scan);
                    if (code != BRANCH)
                    next = regnext(scan);
                    scan = NEXTOPER(scan);
                    if (code != BRANCH)
@@ -2415,7 +2615,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                        cl_init(pRExC_state, &this_class);
                        data_fake.start_class = &this_class;
                        f = SCF_DO_STCLASS_AND;
                        cl_init(pRExC_state, &this_class);
                        data_fake.start_class = &this_class;
                        f = SCF_DO_STCLASS_AND;
-                   }           
+                   }
                    if (flags & SCF_WHILEM_VISITED_POS)
                        f |= SCF_WHILEM_VISITED_POS;
 
                    if (flags & SCF_WHILEM_VISITED_POS)
                        f |= SCF_WHILEM_VISITED_POS;
 
@@ -2630,7 +2830,20 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                                     last = cur;
                                 }
                             } else {
                                     last = cur;
                                 }
                             } else {
-                                if ( last ) {
+/* 
+    Currently we do not believe that the trie logic can
+    handle case insensitive matching properly when the
+    pattern is not unicode (thus forcing unicode semantics).
+
+    If/when this is fixed the following define can be swapped
+    in below to fully enable trie logic.
+
+#define TRIE_TYPE_IS_SAFE 1
+
+*/
+#define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT)
+
+                                if ( last && TRIE_TYPE_IS_SAFE ) {
                                     make_trie( pRExC_state, 
                                             startbranch, first, cur, tail, count, 
                                             optype, depth+1 );
                                     make_trie( pRExC_state, 
                                             startbranch, first, cur, tail, count, 
                                             optype, depth+1 );
@@ -2658,7 +2871,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                               "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
 
                         });
                               "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
 
                         });
-                        if ( last ) {
+                        
+                        if ( last && TRIE_TYPE_IS_SAFE ) {
                             made= make_trie( pRExC_state, startbranch, first, scan, tail, count, optype, depth+1 );
 #ifdef TRIE_STUDY_OPT  
                             if ( ((made == MADE_EXACT_TRIE && 
                             made= make_trie( pRExC_state, startbranch, first, scan, tail, count, optype, depth+1 );
 #ifdef TRIE_STUDY_OPT  
                             if ( ((made == MADE_EXACT_TRIE && 
@@ -2699,7 +2913,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     end   = RExC_close_parens[paren-1];
                 } else {
                     paren = 0;
                     end   = RExC_close_parens[paren-1];
                 } else {
                     paren = 0;
-                    start = RExC_rx->program + 1;
+                    start = RExC_rxi->program + 1;
                     end   = RExC_opend;
                 }
                 if (!recursed) {
                     end   = RExC_opend;
                 }
                 if (!recursed) {
@@ -2711,7 +2925,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     Newx(newframe,1,scan_frame);
                 } else {
                     if (flags & SCF_DO_SUBSTR) {
                     Newx(newframe,1,scan_frame);
                 } else {
                     if (flags & SCF_DO_SUBSTR) {
-                        scan_commit(pRExC_state,data,minlenp);
+                        SCAN_COMMIT(pRExC_state,data,minlenp);
                         data->longest = &(data->longest_float);
                     }
                     is_inf = is_inf_internal = 1;
                         data->longest = &(data->longest_float);
                     }
                     is_inf = is_inf_internal = 1;
@@ -2813,7 +3027,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
            /* Search for fixed substrings supports EXACT only. */
            if (flags & SCF_DO_SUBSTR) {
                assert(data);
            /* Search for fixed substrings supports EXACT only. */
            if (flags & SCF_DO_SUBSTR) {
                assert(data);
-               scan_commit(pRExC_state, data, minlenp);
+               SCAN_COMMIT(pRExC_state, data, minlenp);
            }
            if (UTF) {
                const U8 * const s = (U8 *)STRING(scan);
            }
            if (UTF) {
                const U8 * const s = (U8 *)STRING(scan);
@@ -2892,7 +3106,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                is_inf = is_inf_internal = 1;
                scan = regnext(scan);
                if (flags & SCF_DO_SUBSTR) {
                is_inf = is_inf_internal = 1;
                scan = regnext(scan);
                if (flags & SCF_DO_SUBSTR) {
-                   scan_commit(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
+                   SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
                    data->longest = &(data->longest_float);
                }
                goto optimize_curly_tail;
                    data->longest = &(data->longest_float);
                }
                goto optimize_curly_tail;
@@ -2915,7 +3129,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                next_is_eval = (OP(scan) == EVAL);
              do_curly:
                if (flags & SCF_DO_SUBSTR) {
                next_is_eval = (OP(scan) == EVAL);
              do_curly:
                if (flags & SCF_DO_SUBSTR) {
-                   if (mincount == 0) scan_commit(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
+                   if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
                    pos_before = data->pos_min;
                }
                if (data) {
                    pos_before = data->pos_min;
                }
                if (data) {
@@ -2979,11 +3193,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    (next_is_eval || !(mincount == 0 && maxcount == 1))
                    && (minnext == 0) && (deltanext == 0)
                    && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
                    (next_is_eval || !(mincount == 0 && maxcount == 1))
                    && (minnext == 0) && (deltanext == 0)
                    && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
-                   && maxcount <= REG_INFTY/3 /* Complement check for big count */
-                   && ckWARN(WARN_REGEXP))
+                   && maxcount <= REG_INFTY/3) /* Complement check for big count */
                {
                {
-                   vWARN(RExC_parse,
-                         "Quantifier unexpected on zero-length expression");
+                   ckWARNreg(RExC_parse,
+                             "Quantifier unexpected on zero-length expression");
                }
 
                min += minnext * mincount;
                }
 
                min += minnext * mincount;
@@ -3148,9 +3361,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                        
                        l -= old;
                        /* Get the added string: */
                        
                        l -= old;
                        /* Get the added string: */
-                       last_str = newSVpvn(s  + old, l);
-                       if (UTF)
-                           SvUTF8_on(last_str);
+                       last_str = newSVpvn_utf8(s  + old, l, UTF);
                        if (deltanext == 0 && pos_before == b) {
                            /* What was added is a constant string */
                            if (mincount > 1) {
                        if (deltanext == 0 && pos_before == b) {
                            /* What was added is a constant string */
                            if (mincount > 1) {
@@ -3168,7 +3379,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                                        SvUTF8(sv) && SvMAGICAL(sv) ?
                                        mg_find(sv, PERL_MAGIC_utf8) : NULL;
                                    if (mg && mg->mg_len >= 0)
                                        SvUTF8(sv) && SvMAGICAL(sv) ?
                                        mg_find(sv, PERL_MAGIC_utf8) : NULL;
                                    if (mg && mg->mg_len >= 0)
-                                       mg->mg_len += CHR_SVLEN(last_str);
+                                       mg->mg_len += CHR_SVLEN(last_str) - l;
                                }
                                data->last_end += l * (mincount - 1);
                            }
                                }
                                data->last_end += l * (mincount - 1);
                            }
@@ -3186,7 +3397,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    if (mincount != maxcount) {
                         /* Cannot extend fixed substrings found inside
                            the group.  */
                    if (mincount != maxcount) {
                         /* Cannot extend fixed substrings found inside
                            the group.  */
-                       scan_commit(pRExC_state,data,minlenp);
+                       SCAN_COMMIT(pRExC_state,data,minlenp);
                        if (mincount && last_str) {
                            SV * const sv = data->last_found;
                            MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
                        if (mincount && last_str) {
                            SV * const sv = data->last_found;
                            MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
@@ -3218,7 +3429,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                continue;
            default:                    /* REF and CLUMP only? */
                if (flags & SCF_DO_SUBSTR) {
                continue;
            default:                    /* REF and CLUMP only? */
                if (flags & SCF_DO_SUBSTR) {
-                   scan_commit(pRExC_state,data,minlenp);      /* Cannot expect anything... */
+                   SCAN_COMMIT(pRExC_state,data,minlenp);      /* Cannot expect anything... */
                    data->longest = &(data->longest_float);
                }
                is_inf = is_inf_internal = 1;
                    data->longest = &(data->longest_float);
                }
                is_inf = is_inf_internal = 1;
@@ -3228,11 +3439,51 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                break;
            }
        }
                break;
            }
        }
+       else if (OP(scan) == LNBREAK) {
+           if (flags & SCF_DO_STCLASS) {
+               int value = 0;
+               data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
+               if (flags & SCF_DO_STCLASS_AND) {
+                    for (value = 0; value < 256; value++)
+                        if (!is_VERTWS_cp(value))
+                            ANYOF_BITMAP_CLEAR(data->start_class, value);  
+                }                                                              
+                else {                                                         
+                    for (value = 0; value < 256; value++)
+                        if (is_VERTWS_cp(value))
+                            ANYOF_BITMAP_SET(data->start_class, value);           
+                }                                                              
+                if (flags & SCF_DO_STCLASS_OR)
+                   cl_and(data->start_class, and_withp);
+               flags &= ~SCF_DO_STCLASS;
+            }
+           min += 1;
+           delta += 1;
+            if (flags & SCF_DO_SUBSTR) {
+               SCAN_COMMIT(pRExC_state,data,minlenp);  /* Cannot expect anything... */
+               data->pos_min += 1;
+               data->pos_delta += 1;
+               data->longest = &(data->longest_float);
+           }
+           
+       }
+       else if (OP(scan) == FOLDCHAR) {
+           int d = ARG(scan)==0xDF ? 1 : 2;
+           flags &= ~SCF_DO_STCLASS;
+            min += 1;
+            delta += d;
+            if (flags & SCF_DO_SUBSTR) {
+               SCAN_COMMIT(pRExC_state,data,minlenp);  /* Cannot expect anything... */
+               data->pos_min += 1;
+               data->pos_delta += d;
+               data->longest = &(data->longest_float);
+           }
+       }
        else if (strchr((const char*)PL_simple,OP(scan))) {
            int value = 0;
 
            if (flags & SCF_DO_SUBSTR) {
        else if (strchr((const char*)PL_simple,OP(scan))) {
            int value = 0;
 
            if (flags & SCF_DO_SUBSTR) {
-               scan_commit(pRExC_state,data,minlenp);
+               SCAN_COMMIT(pRExC_state,data,minlenp);
                data->pos_min++;
            }
            min++;
                data->pos_min++;
            }
            min++;
@@ -3422,6 +3673,9 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                        }
                    }
                    break;
                        }
                    }
                    break;
+               CASE_SYNST_FNC(VERTWS);
+               CASE_SYNST_FNC(HORIZWS);
+               
                }
                if (flags & SCF_DO_STCLASS_OR)
                    cl_and(data->start_class, and_withp);
                }
                if (flags & SCF_DO_STCLASS_OR)
                    cl_and(data->start_class, and_withp);
@@ -3456,6 +3710,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                }
                 else
                     data_fake.last_closep = &fake;
                }
                 else
                     data_fake.last_closep = &fake;
+               data_fake.pos_delta = delta;
                 if ( flags & SCF_DO_STCLASS && !scan->flags
                      && OP(scan) == IFMATCH ) { /* Lookahead */
                     cl_init(pRExC_state, &intrnl);
                 if ( flags & SCF_DO_STCLASS && !scan->flags
                      && OP(scan) == IFMATCH ) { /* Lookahead */
                     cl_init(pRExC_state, &intrnl);
@@ -3470,10 +3725,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     last, &data_fake, stopparen, recursed, NULL, f, depth+1);
                 if (scan->flags) {
                     if (deltanext) {
                     last, &data_fake, stopparen, recursed, NULL, f, depth+1);
                 if (scan->flags) {
                     if (deltanext) {
-                        vFAIL("Variable length lookbehind not implemented");
+                       FAIL("Variable length lookbehind not implemented");
                     }
                     else if (minnext > (I32)U8_MAX) {
                     }
                     else if (minnext > (I32)U8_MAX) {
-                        vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
+                       FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
                     }
                     scan->flags = (U8)minnext;
                 }
                     }
                     scan->flags = (U8)minnext;
                 }
@@ -3485,11 +3740,22 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     data->whilem_c = data_fake.whilem_c;
                 }
                 if (f & SCF_DO_STCLASS_AND) {
                     data->whilem_c = data_fake.whilem_c;
                 }
                 if (f & SCF_DO_STCLASS_AND) {
-                    const int was = (data->start_class->flags & ANYOF_EOS);
-
-                    cl_and(data->start_class, &intrnl);
-                    if (was)
-                        data->start_class->flags |= ANYOF_EOS;
+                   if (flags & SCF_DO_STCLASS_OR) {
+                       /* OR before, AND after: ideally we would recurse with
+                        * data_fake to get the AND applied by study of the
+                        * remainder of the pattern, and then derecurse;
+                        * *** HACK *** for now just treat as "no information".
+                        * See [perl #56690].
+                        */
+                       cl_init(pRExC_state, data->start_class);
+                   }  else {
+                       /* AND before and after: combine and continue */
+                       const int was = (data->start_class->flags & ANYOF_EOS);
+
+                       cl_and(data->start_class, &intrnl);
+                       if (was)
+                           data->start_class->flags |= ANYOF_EOS;
+                   }
                 }
            }
 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
                 }
            }
 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
@@ -3520,13 +3786,14 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     if ((flags & SCF_DO_SUBSTR) && data->last_found) {
                         f |= SCF_DO_SUBSTR;
                         if (scan->flags) 
                     if ((flags & SCF_DO_SUBSTR) && data->last_found) {
                         f |= SCF_DO_SUBSTR;
                         if (scan->flags) 
-                            scan_commit(pRExC_state, &data_fake,minlenp);
+                            SCAN_COMMIT(pRExC_state, &data_fake,minlenp);
                         data_fake.last_found=newSVsv(data->last_found);
                     }
                 }
                 else
                     data_fake.last_closep = &fake;
                 data_fake.flags = 0;
                         data_fake.last_found=newSVsv(data->last_found);
                     }
                 }
                 else
                     data_fake.last_closep = &fake;
                 data_fake.flags = 0;
+               data_fake.pos_delta = delta;
                 if (is_inf)
                    data_fake.flags |= SF_IS_INF;
                 if ( flags & SCF_DO_STCLASS && !scan->flags
                 if (is_inf)
                    data_fake.flags |= SF_IS_INF;
                 if ( flags & SCF_DO_STCLASS && !scan->flags
@@ -3544,10 +3811,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     last, &data_fake, stopparen, recursed, NULL, f,depth+1);
                 if (scan->flags) {
                     if (deltanext) {
                     last, &data_fake, stopparen, recursed, NULL, f,depth+1);
                 if (scan->flags) {
                     if (deltanext) {
-                        vFAIL("Variable length lookbehind not implemented");
+                       FAIL("Variable length lookbehind not implemented");
                     }
                     else if (*minnextp > (I32)U8_MAX) {
                     }
                     else if (*minnextp > (I32)U8_MAX) {
-                        vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
+                       FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
                     }
                     scan->flags = (U8)*minnextp;
                 }
                     }
                     scan->flags = (U8)*minnextp;
                 }
@@ -3570,7 +3837,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
                         if (RExC_rx->minlen<*minnextp)
                             RExC_rx->minlen=*minnextp;
                     if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
                         if (RExC_rx->minlen<*minnextp)
                             RExC_rx->minlen=*minnextp;
-                        scan_commit(pRExC_state, &data_fake, minnextp);
+                        SCAN_COMMIT(pRExC_state, &data_fake, minnextp);
                         SvREFCNT_dec(data_fake.last_found);
                         
                         if ( data_fake.minlen_fixed != minlenp ) 
                         SvREFCNT_dec(data_fake.last_found);
                         
                         if ( data_fake.minlen_fixed != minlenp ) 
@@ -3616,7 +3883,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
        }
        else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
            if (flags & SCF_DO_SUBSTR) {
        }
        else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
            if (flags & SCF_DO_SUBSTR) {
-               scan_commit(pRExC_state,data,minlenp);
+               SCAN_COMMIT(pRExC_state,data,minlenp);
                flags &= ~SCF_DO_SUBSTR;
            }
            if (data && OP(scan)==ACCEPT) {
                flags &= ~SCF_DO_SUBSTR;
            }
            if (data && OP(scan)==ACCEPT) {
@@ -3628,7 +3895,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
        else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
        {
                if (flags & SCF_DO_SUBSTR) {
        else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
        {
                if (flags & SCF_DO_SUBSTR) {
-                   scan_commit(pRExC_state,data,minlenp);
+                   SCAN_COMMIT(pRExC_state,data,minlenp);
                    data->longest = &(data->longest_float);
                }
                is_inf = is_inf_internal = 1;
                    data->longest = &(data->longest_float);
                }
                is_inf = is_inf_internal = 1;
@@ -3636,6 +3903,19 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    cl_anything(pRExC_state, data->start_class);
                flags &= ~SCF_DO_STCLASS;
        }
                    cl_anything(pRExC_state, data->start_class);
                flags &= ~SCF_DO_STCLASS;
        }
+       else if (OP(scan) == GPOS) {
+           if (!(RExC_rx->extflags & RXf_GPOS_FLOAT) &&
+               !(delta || is_inf || (data && data->pos_delta))) 
+           {
+               if (!(RExC_rx->extflags & RXf_ANCH) && (flags & SCF_DO_SUBSTR))
+                   RExC_rx->extflags |= RXf_ANCH_GPOS;
+               if (RExC_rx->gofs < (U32)min)
+                   RExC_rx->gofs = min;
+            } else {
+                RExC_rx->extflags |= RXf_GPOS_FLOAT;
+                RExC_rx->gofs = 0;
+            }      
+       }
 #ifdef TRIE_STUDY_OPT
 #ifdef FULL_TRIE_STUDY
         else if (PL_regkind[OP(scan)] == TRIE) {
 #ifdef TRIE_STUDY_OPT
 #ifdef FULL_TRIE_STUDY
         else if (PL_regkind[OP(scan)] == TRIE) {
@@ -3644,12 +3924,12 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                check there too. */
             regnode *trie_node= scan;
             regnode *tail= regnext(scan);
                check there too. */
             regnode *trie_node= scan;
             regnode *tail= regnext(scan);
-            reg_trie_data *trie = (reg_trie_data*)RExC_rx->data->data[ ARG(scan) ];
+            reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
             I32 max1 = 0, min1 = I32_MAX;
             struct regnode_charclass_class accum;
 
             if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
             I32 max1 = 0, min1 = I32_MAX;
             struct regnode_charclass_class accum;
 
             if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
-                scan_commit(pRExC_state, data,minlenp); /* Cannot merge strings after this. */
+                SCAN_COMMIT(pRExC_state, data,minlenp); /* Cannot merge strings after this. */
             if (flags & SCF_DO_STCLASS)
                 cl_init_zero(pRExC_state, &accum);
                 
             if (flags & SCF_DO_STCLASS)
                 cl_init_zero(pRExC_state, &accum);
                 
@@ -3672,7 +3952,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                     }
                     else
                         data_fake.last_closep = &fake;
                     }
                     else
                         data_fake.last_closep = &fake;
-                        
+                   data_fake.pos_delta = delta;
                     if (flags & SCF_DO_STCLASS) {
                         cl_init(pRExC_state, &this_class);
                         data_fake.start_class = &this_class;
                     if (flags & SCF_DO_STCLASS) {
                         cl_init(pRExC_state, &this_class);
                         data_fake.start_class = &this_class;
@@ -3759,14 +4039,14 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
         }
 #else
        else if (PL_regkind[OP(scan)] == TRIE) {
         }
 #else
        else if (PL_regkind[OP(scan)] == TRIE) {
-           reg_trie_data *trie = (reg_trie_data*)RExC_rx->data->data[ ARG(scan) ];
+           reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
            U8*bang=NULL;
            
            min += trie->minlen;
            delta += (trie->maxlen - trie->minlen);
            flags &= ~SCF_DO_STCLASS; /* xxx */
             if (flags & SCF_DO_SUBSTR) {
            U8*bang=NULL;
            
            min += trie->minlen;
            delta += (trie->maxlen - trie->minlen);
            flags &= ~SCF_DO_STCLASS; /* xxx */
             if (flags & SCF_DO_SUBSTR) {
-               scan_commit(pRExC_state,data,minlenp);  /* Cannot expect anything... */
+               SCAN_COMMIT(pRExC_state,data,minlenp);  /* Cannot expect anything... */
                data->pos_min += trie->minlen;
                data->pos_delta += (trie->maxlen - trie->minlen);
                if (trie->maxlen != trie->minlen)
                data->pos_min += trie->minlen;
                data->pos_delta += (trie->maxlen - trie->minlen);
                if (trie->maxlen != trie->minlen)
@@ -3777,6 +4057,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
        }
 #endif /* old or new */
 #endif /* TRIE_STUDY_OPT */    
        }
 #endif /* old or new */
 #endif /* TRIE_STUDY_OPT */    
+
        /* Else: zero-length, ignore. */
        scan = regnext(scan);
     }
        /* Else: zero-length, ignore. */
        scan = regnext(scan);
     }
@@ -3790,6 +4071,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
 
   finish:
     assert(!frame);
 
   finish:
     assert(!frame);
+    DEBUG_STUDYDATA("pre-fin:",data,depth);
 
     *scanp = scan;
     *deltap = is_inf_internal ? I32_MAX : delta;
 
     *scanp = scan;
     *deltap = is_inf_internal ? I32_MAX : delta;
@@ -3810,32 +4092,31 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
     if (flags & SCF_TRIE_RESTUDY)
         data->flags |=         SCF_TRIE_RESTUDY;
     
     if (flags & SCF_TRIE_RESTUDY)
         data->flags |=         SCF_TRIE_RESTUDY;
     
-    DEBUG_STUDYDATA(data,depth);
+    DEBUG_STUDYDATA("post-fin:",data,depth);
     
     return min < stopmin ? min : stopmin;
 }
 
     
     return min < stopmin ? min : stopmin;
 }
 
-STATIC I32
-S_add_data(RExC_state_t *pRExC_state, I32 n, const char *s)
+STATIC U32
+S_add_data(RExC_state_t *pRExC_state, U32 n, const char *s)
 {
 {
-    if (RExC_rx->data) {
-       const U32 count = RExC_rx->data->count;
-       Renewc(RExC_rx->data,
-              sizeof(*RExC_rx->data) + sizeof(void*) * (count + n - 1),
-              char, struct reg_data);
-       Renew(RExC_rx->data->what, count + n, U8);
-       RExC_rx->data->count += n;
-    }
-    else {
-       Newxc(RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
-            char, struct reg_data);
-       Newx(RExC_rx->data->what, n, U8);
-       RExC_rx->data->count = n;
-    }
-    Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
-    return RExC_rx->data->count - n;
+    U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
+
+    PERL_ARGS_ASSERT_ADD_DATA;
+
+    Renewc(RExC_rxi->data,
+          sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
+          char, struct reg_data);
+    if(count)
+       Renew(RExC_rxi->data->what, count + n, U8);
+    else
+       Newx(RExC_rxi->data->what, n, U8);
+    RExC_rxi->data->count = count + n;
+    Copy(s, RExC_rxi->data->what + count, n, U8);
+    return count;
 }
 
 }
 
+/*XXX: todo make this not included in a non debugging perl */
 #ifndef PERL_IN_XSUB_RE
 void
 Perl_reginitcolors(pTHX)
 #ifndef PERL_IN_XSUB_RE
 void
 Perl_reginitcolors(pTHX)
@@ -3899,38 +4180,45 @@ Perl_reginitcolors(pTHX)
 extern const struct regexp_engine my_reg_engine;
 #define RE_ENGINE_PTR &my_reg_engine
 #endif
 extern const struct regexp_engine my_reg_engine;
 #define RE_ENGINE_PTR &my_reg_engine
 #endif
-/* these make a few things look better, to avoid indentation */
-#define BEGIN_BLOCK {
-#define END_BLOCK }
-regexp *
-Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
+
+#ifndef PERL_IN_XSUB_RE 
+REGEXP *
+Perl_pregcomp(pTHX_ SV * const pattern, const U32 flags)
 {
     dVAR;
 {
     dVAR;
-    GET_RE_DEBUG_FLAGS_DECL;
-    DEBUG_r(if (!PL_colorset) reginitcolors());
-#ifndef PERL_IN_XSUB_RE
-    BEGIN_BLOCK
+    HV * const table = GvHV(PL_hintgv);
+
+    PERL_ARGS_ASSERT_PREGCOMP;
+
     /* Dispatch a request to compile a regexp to correct 
        regexp engine. */
     /* Dispatch a request to compile a regexp to correct 
        regexp engine. */
-    HV * const table = GvHV(PL_hintgv);
     if (table) {
         SV **ptr= hv_fetchs(table, "regcomp", FALSE);
     if (table) {
         SV **ptr= hv_fetchs(table, "regcomp", FALSE);
+        GET_RE_DEBUG_FLAGS_DECL;
         if (ptr && SvIOK(*ptr) && SvIV(*ptr)) {
             const regexp_engine *eng=INT2PTR(regexp_engine*,SvIV(*ptr));
             DEBUG_COMPILE_r({
                 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
                     SvIV(*ptr));
             });            
         if (ptr && SvIOK(*ptr) && SvIV(*ptr)) {
             const regexp_engine *eng=INT2PTR(regexp_engine*,SvIV(*ptr));
             DEBUG_COMPILE_r({
                 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
                     SvIV(*ptr));
             });            
-            return CALLREGCOMP_ENG(eng, exp, xend, pm);
+            return CALLREGCOMP_ENG(eng, pattern, flags);
         } 
     }
         } 
     }
-    END_BLOCK
+    return Perl_re_compile(aTHX_ pattern, flags);
+}
 #endif
 #endif
-    BEGIN_BLOCK    
-    register regexp *r;
+
+REGEXP *
+Perl_re_compile(pTHX_ SV * const pattern, U32 pm_flags)
+{
+    dVAR;
+    REGEXP *rx;
+    struct regexp *r;
+    register regexp_internal *ri;
+    STRLEN plen;
+    char  *exp = SvPV(pattern, plen);
+    char* xend = exp + plen;
     regnode *scan;
     regnode *scan;
-    regnode *first;
     I32 flags;
     I32 minlen = 0;
     I32 sawplus = 0;
     I32 flags;
     I32 minlen = 0;
     I32 sawplus = 0;
@@ -3942,20 +4230,25 @@ Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
     int restudied= 0;
     RExC_state_t copyRExC_state;
 #endif    
     int restudied= 0;
     RExC_state_t copyRExC_state;
 #endif    
-    if (exp == NULL)
-       FAIL("NULL regexp argument");
+    GET_RE_DEBUG_FLAGS_DECL;
 
 
-    RExC_utf8 = pm->op_pmdynflags & PMdf_CMP_UTF8;
+    PERL_ARGS_ASSERT_RE_COMPILE;
+
+    DEBUG_r(if (!PL_colorset) reginitcolors());
+
+    RExC_utf8 = RExC_orig_utf8 = SvUTF8(pattern);
 
 
-    RExC_precomp = exp;
     DEBUG_COMPILE_r({
         SV *dsv= sv_newmortal();
         RE_PV_QUOTED_DECL(s, RExC_utf8,
     DEBUG_COMPILE_r({
         SV *dsv= sv_newmortal();
         RE_PV_QUOTED_DECL(s, RExC_utf8,
-            dsv, RExC_precomp, (xend - exp), 60);
+            dsv, exp, plen, 60);
         PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
                       PL_colors[4],PL_colors[5],s);
     });
         PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
                       PL_colors[4],PL_colors[5],s);
     });
-    RExC_flags = pm->op_pmflags;
+
+redo_first_pass:
+    RExC_precomp = exp;
+    RExC_flags = pm_flags;
     RExC_sawback = 0;
 
     RExC_seen = 0;
     RExC_sawback = 0;
 
     RExC_seen = 0;
@@ -3969,16 +4262,17 @@ Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
     RExC_end = xend;
     RExC_naughty = 0;
     RExC_npar = 1;
     RExC_end = xend;
     RExC_naughty = 0;
     RExC_npar = 1;
-    RExC_cpar = 1;
     RExC_nestroot = 0;
     RExC_size = 0L;
     RExC_emit = &PL_regdummy;
     RExC_whilem_seen = 0;
     RExC_nestroot = 0;
     RExC_size = 0L;
     RExC_emit = &PL_regdummy;
     RExC_whilem_seen = 0;
-    RExC_charnames = NULL;
     RExC_open_parens = NULL;
     RExC_close_parens = NULL;
     RExC_opend = NULL;
     RExC_paren_names = NULL;
     RExC_open_parens = NULL;
     RExC_close_parens = NULL;
     RExC_opend = NULL;
     RExC_paren_names = NULL;
+#ifdef DEBUGGING
+    RExC_paren_name_list = NULL;
+#endif
     RExC_recurse = NULL;
     RExC_recurse_count = 0;
 
     RExC_recurse = NULL;
     RExC_recurse_count = 0;
 
@@ -3991,6 +4285,25 @@ Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
        RExC_precomp = NULL;
        return(NULL);
     }
        RExC_precomp = NULL;
        return(NULL);
     }
+    if (RExC_utf8 && !RExC_orig_utf8) {
+        /* It's possible to write a regexp in ascii that represents Unicode
+        codepoints outside of the byte range, such as via \x{100}. If we
+        detect such a sequence we have to convert the entire pattern to utf8
+        and then recompile, as our sizing calculation will have been based
+        on 1 byte == 1 character, but we will need to use utf8 to encode
+        at least some part of the pattern, and therefore must convert the whole
+        thing.
+        XXX: somehow figure out how to make this less expensive...
+        -- dmq */
+        STRLEN len = plen;
+        DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
+           "UTF8 mismatch! Converting to utf8 for resizing and compile\n"));
+        exp = (char*)Perl_bytes_to_utf8(aTHX_ (U8*)exp, &len);
+        xend = exp + len;
+        RExC_orig_utf8 = RExC_utf8;
+        SAVEFREEPV(exp);
+        goto redo_first_pass;
+    }
     DEBUG_PARSE_r({
         PerlIO_printf(Perl_debug_log, 
             "Required size %"IVdf" nodes\n"
     DEBUG_PARSE_r({
         PerlIO_printf(Perl_debug_log, 
             "Required size %"IVdf" nodes\n"
@@ -4008,40 +4321,77 @@ Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
     if (RExC_whilem_seen > 15)
        RExC_whilem_seen = 15;
 
     if (RExC_whilem_seen > 15)
        RExC_whilem_seen = 15;
 
-#ifdef DEBUGGING
-    /* Make room for a sentinel value at the end of the program */
-    RExC_size++;
-#endif
-
     /* Allocate space and zero-initialize. Note, the two step process 
        of zeroing when in debug mode, thus anything assigned has to 
        happen after that */
     /* Allocate space and zero-initialize. Note, the two step process 
        of zeroing when in debug mode, thus anything assigned has to 
        happen after that */
-    Newxc(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
-        char, regexp);
-    if (r == NULL)
+    rx = (REGEXP*) newSV_type(SVt_REGEXP);
+    r = (struct regexp*)SvANY(rx);
+    Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
+        char, regexp_internal);
+    if ( r == NULL || ri == NULL )
        FAIL("Regexp out of space");
 #ifdef DEBUGGING
     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
        FAIL("Regexp out of space");
 #ifdef DEBUGGING
     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
-    Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
+    Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode), char);
+#else 
+    /* bulk initialize base fields with 0. */
+    Zero(ri, sizeof(regexp_internal), char);        
 #endif
 #endif
-    /* initialization begins here */
+
+    /* non-zero initialization begins here */
+    RXi_SET( r, ri );
     r->engine= RE_ENGINE_PTR;
     r->engine= RE_ENGINE_PTR;
-    r->refcnt = 1;
-    r->prelen = xend - exp;
-    r->precomp = savepvn(RExC_precomp, r->prelen);
-    r->subbeg = NULL;
-#ifdef PERL_OLD_COPY_ON_WRITE
-    r->saved_copy = NULL;
-#endif
-    r->reganch = pm->op_pmflags & PMf_COMPILETIME;
-    r->nparens = RExC_npar - 1;        /* set early to validate backrefs */
-    r->lastparen = 0;                  /* mg.c reads this.  */
+    r->extflags = pm_flags;
+    {
+        bool has_p     = ((r->extflags & RXf_PMf_KEEPCOPY) == RXf_PMf_KEEPCOPY);
+       bool has_minus = ((r->extflags & RXf_PMf_STD_PMMOD) != RXf_PMf_STD_PMMOD);
+       bool has_runon = ((RExC_seen & REG_SEEN_RUN_ON_COMMENT)==REG_SEEN_RUN_ON_COMMENT);
+       U16 reganch = (U16)((r->extflags & RXf_PMf_STD_PMMOD)
+                           >> RXf_PMf_STD_PMMOD_SHIFT);
+       const char *fptr = STD_PAT_MODS;        /*"msix"*/
+       char *p;
+       const STRLEN wraplen = plen + has_minus + has_p + has_runon
+            + (sizeof(STD_PAT_MODS) - 1)
+            + (sizeof("(?:)") - 1);
+
+       p = sv_grow(MUTABLE_SV(rx), wraplen + 1);
+       SvCUR_set(rx, wraplen);
+       SvPOK_on(rx);
+       SvFLAGS(rx) |= SvUTF8(pattern);
+        *p++='('; *p++='?';
+        if (has_p)
+            *p++ = KEEPCOPY_PAT_MOD; /*'p'*/
+        {
+            char *r = p + (sizeof(STD_PAT_MODS) - 1) + has_minus - 1;
+            char *colon = r + 1;
+            char ch;
+
+            while((ch = *fptr++)) {
+                if(reganch & 1)
+                    *p++ = ch;
+                else
+                    *r-- = ch;
+                reganch >>= 1;
+            }
+            if(has_minus) {
+                *r = '-';
+                p = colon;
+            }
+        }
+
+        *p++ = ':';
+        Copy(RExC_precomp, p, plen, char);
+       assert ((RX_WRAPPED(rx) - p) < 16);
+       r->pre_prefix = p - RX_WRAPPED(rx);
+        p += plen;
+        if (has_runon)
+            *p++ = '\n';
+        *p++ = ')';
+        *p = 0;
+    }
 
 
-    r->substrs = 0;                    /* Useful during FAIL. */
-    r->startp = 0;                     /* Useful during FAIL. */
-    r->endp = 0;                       
-    r->swap = NULL; 
-    r->paren_names = 0;
+    r->intflags = 0;
+    r->nparens = RExC_npar - 1;        /* set early to validate backrefs */
     
     if (RExC_seen & REG_SEEN_RECURSE) {
         Newxz(RExC_open_parens, RExC_npar,regnode *);
     
     if (RExC_seen & REG_SEEN_RECURSE) {
         Newxz(RExC_open_parens, RExC_npar,regnode *);
@@ -4051,38 +4401,35 @@ Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
     }
 
     /* Useful during FAIL. */
     }
 
     /* Useful during FAIL. */
-    Newxz(r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
-    if (r->offsets) {
-       r->offsets[0] = RExC_size;
-    }
+#ifdef RE_TRACK_PATTERN_OFFSETS
+    Newxz(ri->u.offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
     DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
                           "%s %"UVuf" bytes for offset annotations.\n",
     DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
                           "%s %"UVuf" bytes for offset annotations.\n",
-                          r->offsets ? "Got" : "Couldn't get",
+                          ri->u.offsets ? "Got" : "Couldn't get",
                           (UV)((2*RExC_size+1) * sizeof(U32))));
                           (UV)((2*RExC_size+1) * sizeof(U32))));
-
+#endif
+    SetProgLen(ri,RExC_size);
+    RExC_rx_sv = rx;
     RExC_rx = r;
     RExC_rx = r;
+    RExC_rxi = ri;
 
     /* Second pass: emit code. */
 
     /* Second pass: emit code. */
-    RExC_flags = pm->op_pmflags;       /* don't let top level (?i) bleed */
+    RExC_flags = pm_flags;     /* don't let top level (?i) bleed */
     RExC_parse = exp;
     RExC_end = xend;
     RExC_naughty = 0;
     RExC_npar = 1;
     RExC_parse = exp;
     RExC_end = xend;
     RExC_naughty = 0;
     RExC_npar = 1;
-    RExC_cpar = 1;
-    RExC_emit_start = r->program;
-    RExC_emit = r->program;
-#ifdef DEBUGGING
-    /* put a sentinal on the end of the program so we can check for
-       overwrites */
-    r->program[RExC_size].type = 255;
-#endif
+    RExC_emit_start = ri->program;
+    RExC_emit = ri->program;
+    RExC_emit_bound = ri->program + RExC_size + 1;
+
     /* Store the count of eval-groups for security checks: */
     /* Store the count of eval-groups for security checks: */
-    RExC_emit->next_off = (RExC_seen_evals > (I32)U16_MAX) ? U16_MAX : (U16)RExC_seen_evals;
+    RExC_rx->seen_evals = RExC_seen_evals;
     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
-    r->data = 0;
-    if (reg(pRExC_state, 0, &flags,1) == NULL)
+    if (reg(pRExC_state, 0, &flags,1) == NULL) {
+       ReREFCNT_dec(rx);   
        return(NULL);
        return(NULL);
-
+    }
     /* XXXX To minimize changes to RE engine we always allocate
        3-units-long substrs field. */
     Newx(r->substrs, 1, struct reg_substr_data);
     /* XXXX To minimize changes to RE engine we always allocate
        3-units-long substrs field. */
     Newx(r->substrs, 1, struct reg_substr_data);
@@ -4096,7 +4443,10 @@ reStudy:
     Zero(r->substrs, 1, struct reg_substr_data);
 
 #ifdef TRIE_STUDY_OPT
     Zero(r->substrs, 1, struct reg_substr_data);
 
 #ifdef TRIE_STUDY_OPT
-    if ( restudied ) {
+    if (!restudied) {
+        StructCopy(&zero_scan_data, &data, scan_data_t);
+        copyRExC_state = RExC_state;
+    } else {
         U32 seen=RExC_seen;
         DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
         
         U32 seen=RExC_seen;
         DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
         
@@ -4111,23 +4461,21 @@ reStudy:
            SvREFCNT_dec(data.last_found);
        }
        StructCopy(&zero_scan_data, &data, scan_data_t);
            SvREFCNT_dec(data.last_found);
        }
        StructCopy(&zero_scan_data, &data, scan_data_t);
-    } else {
-        StructCopy(&zero_scan_data, &data, scan_data_t);
-        copyRExC_state = RExC_state;
     }
 #else
     StructCopy(&zero_scan_data, &data, scan_data_t);
 #endif    
 
     /* Dig out information for optimizations. */
     }
 #else
     StructCopy(&zero_scan_data, &data, scan_data_t);
 #endif    
 
     /* Dig out information for optimizations. */
-    r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
-    pm->op_pmflags = RExC_flags;
+    r->extflags = RExC_flags; /* was pm_op */
+    /*dmq: removed as part of de-PMOP: pm->op_pmflags = RExC_flags; */
     if (UTF)
     if (UTF)
-        r->reganch |= ROPT_UTF8;       /* Unicode in it? */
-    r->regstclass = NULL;
+       SvUTF8_on(rx);  /* Unicode in it? */
+    ri->regstclass = NULL;
     if (RExC_naughty >= 10)    /* Probably an expensive pattern. */
     if (RExC_naughty >= 10)    /* Probably an expensive pattern. */
-       r->reganch |= ROPT_NAUGHTY;
-    scan = r->program + 1;             /* First BRANCH. */
+       r->intflags |= PREGf_NAUGHTY;
+    scan = ri->program + 1;            /* First BRANCH. */
 
     /* testing for BRANCH here tells us whether there is "must appear"
        data in the pattern. If there is then we can use it for optimisations */
 
     /* testing for BRANCH here tells us whether there is "must appear"
        data in the pattern. If there is then we can use it for optimisations */
@@ -4137,29 +4485,43 @@ reStudy:
        struct regnode_charclass_class ch_class; /* pointed to by data */
        int stclass_flag;
        I32 last_close = 0; /* pointed to by data */
        struct regnode_charclass_class ch_class; /* pointed to by data */
        int stclass_flag;
        I32 last_close = 0; /* pointed to by data */
-
-       first = scan;
-       /* Skip introductions and multiplicators >= 1. */
+        regnode *first= scan;
+        regnode *first_next= regnext(first);
+       
+       /*
+        * Skip introductions and multiplicators >= 1
+        * so that we can extract the 'meat' of the pattern that must 
+        * match in the large if() sequence following.
+        * NOTE that EXACT is NOT covered here, as it is normally
+        * picked up by the optimiser separately. 
+        *
+        * This is unfortunate as the optimiser isnt handling lookahead
+        * properly currently.
+        *
+        */
        while ((OP(first) == OPEN && (sawopen = 1)) ||
               /* An OR of *one* alternative - should not happen now. */
        while ((OP(first) == OPEN && (sawopen = 1)) ||
               /* An OR of *one* alternative - should not happen now. */
-           (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
+           (OP(first) == BRANCH && OP(first_next) != BRANCH) ||
            /* for now we can't handle lookbehind IFMATCH*/
            (OP(first) == IFMATCH && !first->flags) || 
            (OP(first) == PLUS) ||
            (OP(first) == MINMOD) ||
               /* An {n,m} with n>0 */
            /* for now we can't handle lookbehind IFMATCH*/
            (OP(first) == IFMATCH && !first->flags) || 
            (OP(first) == PLUS) ||
            (OP(first) == MINMOD) ||
               /* An {n,m} with n>0 */
-           (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ) 
+           (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ||
+           (OP(first) == NOTHING && PL_regkind[OP(first_next)] != END ))
        {
        {
-               
+               /* 
+                * the only op that could be a regnode is PLUS, all the rest
+                * will be regnode_1 or regnode_2.
+                *
+                */
                if (OP(first) == PLUS)
                    sawplus = 1;
                else
                    first += regarglen[OP(first)];
                if (OP(first) == PLUS)
                    sawplus = 1;
                else
                    first += regarglen[OP(first)];
-               if (OP(first) == IFMATCH) {
-                   first = NEXTOPER(first);
-                   first += EXTRA_STEP_2ARGS;
-               } else  /* XXX possible optimisation for /(?=)/  */
-                   first = NEXTOPER(first);
+               
+               first = NEXTOPER(first);
+               first_next= regnext(first);
        }
 
        /* Starting-point info. */
        }
 
        /* Starting-point info. */
@@ -4170,67 +4532,68 @@ reStudy:
            if (OP(first) == EXACT)
                NOOP;   /* Empty, get anchored substr later. */
            else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
            if (OP(first) == EXACT)
                NOOP;   /* Empty, get anchored substr later. */
            else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
-               r->regstclass = first;
+               ri->regstclass = first;
        }
 #ifdef TRIE_STCLASS    
        else if (PL_regkind[OP(first)] == TRIE &&
        }
 #ifdef TRIE_STCLASS    
        else if (PL_regkind[OP(first)] == TRIE &&
-               ((reg_trie_data *)r->data->data[ ARG(first) ])->minlen>0) 
+               ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0) 
        {
            regnode *trie_op;
            /* this can happen only on restudy */
            if ( OP(first) == TRIE ) {
        {
            regnode *trie_op;
            /* this can happen only on restudy */
            if ( OP(first) == TRIE ) {
-                struct regnode_1 *trieop;
-                Newxz(trieop,1,struct regnode_1);
+                struct regnode_1 *trieop = (struct regnode_1 *)
+                   PerlMemShared_calloc(1, sizeof(struct regnode_1));
                 StructCopy(first,trieop,struct regnode_1);
                 trie_op=(regnode *)trieop;
             } else {
                 StructCopy(first,trieop,struct regnode_1);
                 trie_op=(regnode *)trieop;
             } else {
-                struct regnode_charclass *trieop;
-                Newxz(trieop,1,struct regnode_charclass);
+                struct regnode_charclass *trieop = (struct regnode_charclass *)
+                   PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
                 StructCopy(first,trieop,struct regnode_charclass);
                 trie_op=(regnode *)trieop;
             }
             OP(trie_op)+=2;
             make_trie_failtable(pRExC_state, (regnode *)first, trie_op, 0);
                 StructCopy(first,trieop,struct regnode_charclass);
                 trie_op=(regnode *)trieop;
             }
             OP(trie_op)+=2;
             make_trie_failtable(pRExC_state, (regnode *)first, trie_op, 0);
-           r->regstclass = trie_op;
+           ri->regstclass = trie_op;
        }
 #endif 
        else if (strchr((const char*)PL_simple,OP(first)))
        }
 #endif 
        else if (strchr((const char*)PL_simple,OP(first)))
-           r->regstclass = first;
+           ri->regstclass = first;
        else if (PL_regkind[OP(first)] == BOUND ||
                 PL_regkind[OP(first)] == NBOUND)
        else if (PL_regkind[OP(first)] == BOUND ||
                 PL_regkind[OP(first)] == NBOUND)
-           r->regstclass = first;
+           ri->regstclass = first;
        else if (PL_regkind[OP(first)] == BOL) {
        else if (PL_regkind[OP(first)] == BOL) {
-           r->reganch |= (OP(first) == MBOL
-                          ? ROPT_ANCH_MBOL
+           r->extflags |= (OP(first) == MBOL
+                          ? RXf_ANCH_MBOL
                           : (OP(first) == SBOL
                           : (OP(first) == SBOL
-                             ? ROPT_ANCH_SBOL
-                             : ROPT_ANCH_BOL));
+                             ? RXf_ANCH_SBOL
+                             : RXf_ANCH_BOL));
            first = NEXTOPER(first);
            goto again;
        }
        else if (OP(first) == GPOS) {
            first = NEXTOPER(first);
            goto again;
        }
        else if (OP(first) == GPOS) {
-           r->reganch |= ROPT_ANCH_GPOS;
+           r->extflags |= RXf_ANCH_GPOS;
            first = NEXTOPER(first);
            goto again;
        }
        else if ((!sawopen || !RExC_sawback) &&
            (OP(first) == STAR &&
            PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
            first = NEXTOPER(first);
            goto again;
        }
        else if ((!sawopen || !RExC_sawback) &&
            (OP(first) == STAR &&
            PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
-           !(r->reganch & ROPT_ANCH) )
+           !(r->extflags & RXf_ANCH) && !(RExC_seen & REG_SEEN_EVAL))
        {
            /* turn .* into ^.* with an implied $*=1 */
            const int type =
                (OP(NEXTOPER(first)) == REG_ANY)
        {
            /* turn .* into ^.* with an implied $*=1 */
            const int type =
                (OP(NEXTOPER(first)) == REG_ANY)
-                   ? ROPT_ANCH_MBOL
-                   : ROPT_ANCH_SBOL;
-           r->reganch |= type | ROPT_IMPLICIT;
+                   ? RXf_ANCH_MBOL
+                   : RXf_ANCH_SBOL;
+           r->extflags |= type;
+           r->intflags |= PREGf_IMPLICIT;
            first = NEXTOPER(first);
            goto again;
        }
        if (sawplus && (!sawopen || !RExC_sawback)
            && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
            /* x+ must match at the 1st pos of run of x's */
            first = NEXTOPER(first);
            goto again;
        }
        if (sawplus && (!sawopen || !RExC_sawback)
            && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
            /* x+ must match at the 1st pos of run of x's */
-           r->reganch |= ROPT_SKIP;
+           r->intflags |= PREGf_SKIP;
 
        /* Scan is after the zeroth branch, first is atomic matcher. */
 #ifdef TRIE_STUDY_OPT
 
        /* Scan is after the zeroth branch, first is atomic matcher. */
 #ifdef TRIE_STUDY_OPT
@@ -4264,7 +4627,7 @@ reStudy:
        data.last_found = newSVpvs("");
        data.longest = &(data.longest_fixed);
        first = scan;
        data.last_found = newSVpvs("");
        data.longest = &(data.longest_fixed);
        first = scan;
-       if (!r->regstclass) {
+       if (!ri->regstclass) {
            cl_init(pRExC_state, &ch_class);
            data.start_class = &ch_class;
            stclass_flag = SCF_DO_STCLASS_AND;
            cl_init(pRExC_state, &ch_class);
            data.start_class = &ch_class;
            stclass_flag = SCF_DO_STCLASS_AND;
@@ -4283,9 +4646,10 @@ reStudy:
        if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
             && data.last_start_min == 0 && data.last_end > 0
             && !RExC_seen_zerolen
        if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
             && data.last_start_min == 0 && data.last_end > 0
             && !RExC_seen_zerolen
-            && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
-           r->reganch |= ROPT_CHECK_ALL;
-       scan_commit(pRExC_state, &data,&minlen);
+            && !(RExC_seen & REG_SEEN_VERBARG)
+            && (!(RExC_seen & REG_SEEN_GPOS) || (r->extflags & RXf_ANCH_GPOS)))
+           r->extflags |= RXf_CHECK_ALL;
+       scan_commit(pRExC_state, &data,&minlen,0);
        SvREFCNT_dec(data.last_found);
 
         /* Note that code very similar to this but for anchored string 
        SvREFCNT_dec(data.last_found);
 
         /* Note that code very similar to this but for anchored string 
@@ -4296,7 +4660,7 @@ reStudy:
        if (longest_float_length
            || (data.flags & SF_FL_BEFORE_EOL
                && (!(data.flags & SF_FL_BEFORE_MEOL)
        if (longest_float_length
            || (data.flags & SF_FL_BEFORE_EOL
                && (!(data.flags & SF_FL_BEFORE_MEOL)
-                   || (RExC_flags & PMf_MULTILINE)))) 
+                   || (RExC_flags & RXf_PMf_MULTILINE)))) 
         {
             I32 t,ml;
 
         {
             I32 t,ml;
 
@@ -4330,7 +4694,7 @@ reStudy:
            
            t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
                       && (!(data.flags & SF_FL_BEFORE_MEOL)
            
            t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
                       && (!(data.flags & SF_FL_BEFORE_MEOL)
-                          || (RExC_flags & PMf_MULTILINE)));
+                          || (RExC_flags & RXf_PMf_MULTILINE)));
            fbm_compile(data.longest_float, t ? FBMcf_TAIL : 0);
        }
        else {
            fbm_compile(data.longest_float, t ? FBMcf_TAIL : 0);
        }
        else {
@@ -4348,7 +4712,7 @@ reStudy:
        if (longest_fixed_length
            || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
                && (!(data.flags & SF_FIX_BEFORE_MEOL)
        if (longest_fixed_length
            || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
                && (!(data.flags & SF_FIX_BEFORE_MEOL)
-                   || (RExC_flags & PMf_MULTILINE)))) 
+                   || (RExC_flags & RXf_PMf_MULTILINE)))) 
         {
             I32 t,ml;
 
         {
             I32 t,ml;
 
@@ -4374,7 +4738,7 @@ reStudy:
 
            t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
 
            t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
-                    || (RExC_flags & PMf_MULTILINE)));
+                    || (RExC_flags & RXf_PMf_MULTILINE)));
            fbm_compile(data.longest_fixed, t ? FBMcf_TAIL : 0);
        }
        else {
            fbm_compile(data.longest_fixed, t ? FBMcf_TAIL : 0);
        }
        else {
@@ -4382,23 +4746,23 @@ reStudy:
            SvREFCNT_dec(data.longest_fixed);
            longest_fixed_length = 0;
        }
            SvREFCNT_dec(data.longest_fixed);
            longest_fixed_length = 0;
        }
-       if (r->regstclass
-           && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
-           r->regstclass = NULL;
+       if (ri->regstclass
+           && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
+           ri->regstclass = NULL;
        if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
            && stclass_flag
            && !(data.start_class->flags & ANYOF_EOS)
            && !cl_is_anything(data.start_class))
        {
        if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
            && stclass_flag
            && !(data.start_class->flags & ANYOF_EOS)
            && !cl_is_anything(data.start_class))
        {
-           const I32 n = add_data(pRExC_state, 1, "f");
+           const U32 n = add_data(pRExC_state, 1, "f");
 
 
-           Newx(RExC_rx->data->data[n], 1,
+           Newx(RExC_rxi->data->data[n], 1,
                struct regnode_charclass_class);
            StructCopy(data.start_class,
                struct regnode_charclass_class);
            StructCopy(data.start_class,
-                      (struct regnode_charclass_class*)RExC_rx->data->data[n],
+                      (struct regnode_charclass_class*)RExC_rxi->data->data[n],
                       struct regnode_charclass_class);
                       struct regnode_charclass_class);
-           r->regstclass = (regnode*)RExC_rx->data->data[n];
-           r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
+           ri->regstclass = (regnode*)RExC_rxi->data->data[n];
+           r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
            DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
                      regprop(r, sv, (regnode*)data.start_class);
                      PerlIO_printf(Perl_debug_log,
            DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
                      regprop(r, sv, (regnode*)data.start_class);
                      PerlIO_printf(Perl_debug_log,
@@ -4412,8 +4776,8 @@ reStudy:
            r->check_substr = r->anchored_substr;
            r->check_utf8 = r->anchored_utf8;
            r->check_offset_min = r->check_offset_max = r->anchored_offset;
            r->check_substr = r->anchored_substr;
            r->check_utf8 = r->anchored_utf8;
            r->check_offset_min = r->check_offset_max = r->anchored_offset;
-           if (r->reganch & ROPT_ANCH_SINGLE)
-               r->reganch |= ROPT_NOSCAN;
+           if (r->extflags & RXf_ANCH_SINGLE)
+               r->extflags |= RXf_NOSCAN;
        }
        else {
            r->check_end_shift = r->float_end_shift;
        }
        else {
            r->check_end_shift = r->float_end_shift;
@@ -4424,10 +4788,10 @@ reStudy:
        }
        /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
           This should be changed ASAP!  */
        }
        /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
           This should be changed ASAP!  */
-       if ((r->check_substr || r->check_utf8) && !(r->reganch & ROPT_ANCH_GPOS)) {
-           r->reganch |= RE_USE_INTUIT;
+       if ((r->check_substr || r->check_utf8) && !(r->extflags & RXf_ANCH_GPOS)) {
+           r->extflags |= RXf_USE_INTUIT;
            if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
            if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
-               r->reganch |= RE_INTUIT_TAIL;
+               r->extflags |= RXf_INTUIT_TAIL;
        }
        /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
        if ( (STRLEN)minlen < longest_float_length )
        }
        /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
        if ( (STRLEN)minlen < longest_float_length )
@@ -4444,7 +4808,7 @@ reStudy:
        
        DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
 
        
        DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
 
-       scan = r->program + 1;
+       scan = ri->program + 1;
        cl_init(pRExC_state, &ch_class);
        data.start_class = &ch_class;
        data.last_closep = &last_close;
        cl_init(pRExC_state, &ch_class);
        data.start_class = &ch_class;
        data.last_closep = &last_close;
@@ -4460,15 +4824,15 @@ reStudy:
        if (!(data.start_class->flags & ANYOF_EOS)
            && !cl_is_anything(data.start_class))
        {
        if (!(data.start_class->flags & ANYOF_EOS)
            && !cl_is_anything(data.start_class))
        {
-           const I32 n = add_data(pRExC_state, 1, "f");
+           const U32 n = add_data(pRExC_state, 1, "f");
 
 
-           Newx(RExC_rx->data->data[n], 1,
+           Newx(RExC_rxi->data->data[n], 1,
                struct regnode_charclass_class);
            StructCopy(data.start_class,
                struct regnode_charclass_class);
            StructCopy(data.start_class,
-                      (struct regnode_charclass_class*)RExC_rx->data->data[n],
+                      (struct regnode_charclass_class*)RExC_rxi->data->data[n],
                       struct regnode_charclass_class);
                       struct regnode_charclass_class);
-           r->regstclass = (regnode*)RExC_rx->data->data[n];
-           r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
+           ri->regstclass = (regnode*)RExC_rxi->data->data[n];
+           r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
            DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
                      regprop(r, sv, (regnode*)data.start_class);
                      PerlIO_printf(Perl_debug_log,
            DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
                      regprop(r, sv, (regnode*)data.start_class);
                      PerlIO_printf(Perl_debug_log,
@@ -4481,133 +4845,521 @@ reStudy:
        the "real" pattern. */
     DEBUG_OPTIMISE_r({
        PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf"\n",
        the "real" pattern. */
     DEBUG_OPTIMISE_r({
        PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf"\n",
-           minlen, r->minlen);
+                     (IV)minlen, (IV)r->minlen);
     });
     r->minlenret = minlen;
     if (r->minlen < minlen) 
         r->minlen = minlen;
     
     if (RExC_seen & REG_SEEN_GPOS)
     });
     r->minlenret = minlen;
     if (r->minlen < minlen) 
         r->minlen = minlen;
     
     if (RExC_seen & REG_SEEN_GPOS)
-       r->reganch |= ROPT_GPOS_SEEN;
+       r->extflags |= RXf_GPOS_SEEN;
     if (RExC_seen & REG_SEEN_LOOKBEHIND)
     if (RExC_seen & REG_SEEN_LOOKBEHIND)
-       r->reganch |= ROPT_LOOKBEHIND_SEEN;
+       r->extflags |= RXf_LOOKBEHIND_SEEN;
     if (RExC_seen & REG_SEEN_EVAL)
     if (RExC_seen & REG_SEEN_EVAL)
-       r->reganch |= ROPT_EVAL_SEEN;
+       r->extflags |= RXf_EVAL_SEEN;
     if (RExC_seen & REG_SEEN_CANY)
     if (RExC_seen & REG_SEEN_CANY)
-       r->reganch |= ROPT_CANY_SEEN;
+       r->extflags |= RXf_CANY_SEEN;
     if (RExC_seen & REG_SEEN_VERBARG)
     if (RExC_seen & REG_SEEN_VERBARG)
-       r->reganch |= ROPT_VERBARG_SEEN;
+       r->intflags |= PREGf_VERBARG_SEEN;
     if (RExC_seen & REG_SEEN_CUTGROUP)
     if (RExC_seen & REG_SEEN_CUTGROUP)
-       r->reganch |= ROPT_CUTGROUP_SEEN;
+       r->intflags |= PREGf_CUTGROUP_SEEN;
     if (RExC_paren_names)
     if (RExC_paren_names)
-        r->paren_names = (HV*)SvREFCNT_inc(RExC_paren_names);
+        RXp_PAREN_NAMES(r) = MUTABLE_HV(SvREFCNT_inc(RExC_paren_names));
     else
     else
-        r->paren_names = NULL;
-               
+        RXp_PAREN_NAMES(r) = NULL;
+
+#ifdef STUPID_PATTERN_CHECKS            
+    if (RX_PRELEN(rx) == 0)
+        r->extflags |= RXf_NULL;
+    if (r->extflags & RXf_SPLIT && RX_PRELEN(rx) == 1 && RX_PRECOMP(rx)[0] == ' ')
+        /* XXX: this should happen BEFORE we compile */
+        r->extflags |= (RXf_SKIPWHITE|RXf_WHITE); 
+    else if (RX_PRELEN(rx) == 3 && memEQ("\\s+", RX_PRECOMP(rx), 3))
+        r->extflags |= RXf_WHITE;
+    else if (RX_PRELEN(rx) == 1 && RXp_PRECOMP(rx)[0] == '^')
+        r->extflags |= RXf_START_ONLY;
+#else
+    if (r->extflags & RXf_SPLIT && RX_PRELEN(rx) == 1 && RX_PRECOMP(rx)[0] == ' ')
+            /* XXX: this should happen BEFORE we compile */
+            r->extflags |= (RXf_SKIPWHITE|RXf_WHITE); 
+    else {
+        regnode *first = ri->program + 1;
+        U8 fop = OP(first);
+        U8 nop = OP(NEXTOPER(first));
+        
+        if (PL_regkind[fop] == NOTHING && nop == END)
+            r->extflags |= RXf_NULL;
+        else if (PL_regkind[fop] == BOL && nop == END)
+            r->extflags |= RXf_START_ONLY;
+        else if (fop == PLUS && nop ==SPACE && OP(regnext(first))==END)
+            r->extflags |= RXf_WHITE;    
+    }
+#endif
+#ifdef DEBUGGING
+    if (RExC_paren_names) {
+        ri->name_list_idx = add_data( pRExC_state, 1, "p" );
+        ri->data->data[ri->name_list_idx] = (void*)SvREFCNT_inc(RExC_paren_name_list);
+    } else
+#endif
+        ri->name_list_idx = 0;
+
     if (RExC_recurse_count) {
         for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
             const regnode *scan = RExC_recurse[RExC_recurse_count-1];
             ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
         }
     }
     if (RExC_recurse_count) {
         for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
             const regnode *scan = RExC_recurse[RExC_recurse_count-1];
             ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
         }
     }
-    Newxz(r->startp, RExC_npar, I32);
-    Newxz(r->endp, RExC_npar, I32);
+    Newxz(r->offs, RExC_npar, regexp_paren_pair);
     /* assume we don't need to swap parens around before we match */
 
     DEBUG_DUMP_r({
         PerlIO_printf(Perl_debug_log,"Final program:\n");
         regdump(r);
     });
     /* assume we don't need to swap parens around before we match */
 
     DEBUG_DUMP_r({
         PerlIO_printf(Perl_debug_log,"Final program:\n");
         regdump(r);
     });
-    DEBUG_OFFSETS_r(if (r->offsets) {
-        const U32 len = r->offsets[0];
+#ifdef RE_TRACK_PATTERN_OFFSETS
+    DEBUG_OFFSETS_r(if (ri->u.offsets) {
+        const U32 len = ri->u.offsets[0];
         U32 i;
         GET_RE_DEBUG_FLAGS_DECL;
         U32 i;
         GET_RE_DEBUG_FLAGS_DECL;
-        PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
+        PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)ri->u.offsets[0]);
         for (i = 1; i <= len; i++) {
         for (i = 1; i <= len; i++) {
-            if (r->offsets[i*2-1] || r->offsets[i*2])
+            if (ri->u.offsets[i*2-1] || ri->u.offsets[i*2])
                 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
                 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
-                (UV)i, (UV)r->offsets[i*2-1], (UV)r->offsets[i*2]);
+                (UV)i, (UV)ri->u.offsets[i*2-1], (UV)ri->u.offsets[i*2]);
             }
         PerlIO_printf(Perl_debug_log, "\n");
     });
             }
         PerlIO_printf(Perl_debug_log, "\n");
     });
-    return(r);
-    END_BLOCK    
+#endif
+    return rx;
 }
 
 }
 
-#undef CORE_ONLY_BLOCK
-#undef END_BLOCK
 #undef RE_ENGINE_PTR
 
 #undef RE_ENGINE_PTR
 
-#ifndef PERL_IN_XSUB_RE
+
 SV*
 SV*
-Perl_reg_named_buff_sv(pTHX_ SV* namesv)
+Perl_reg_named_buff(pTHX_ REGEXP * const rx, SV * const key, SV * const value,
+                    const U32 flags)
 {
 {
-    I32 parno = 0; /* no match */
-    if (PL_curpm) {
-        const REGEXP * const rx = PM_GETRE(PL_curpm);
-        if (rx && rx->paren_names) {            
-            HE *he_str = hv_fetch_ent( rx->paren_names, namesv, 0, 0 );
-            if (he_str) {
-                IV i;
-                SV* sv_dat=HeVAL(he_str);
-                I32 *nums=(I32*)SvPVX(sv_dat);
-                for ( i=0; i<SvIVX(sv_dat); i++ ) {
-                    if ((I32)(rx->lastparen) >= nums[i] &&
-                        rx->endp[nums[i]] != -1) 
-                    {
-                        parno = nums[i];
-                        break;
-                    }
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF;
+
+    PERL_UNUSED_ARG(value);
+
+    if (flags & RXapif_FETCH) {
+        return reg_named_buff_fetch(rx, key, flags);
+    } else if (flags & (RXapif_STORE | RXapif_DELETE | RXapif_CLEAR)) {
+        Perl_croak(aTHX_ "%s", PL_no_modify);
+        return NULL;
+    } else if (flags & RXapif_EXISTS) {
+        return reg_named_buff_exists(rx, key, flags)
+            ? &PL_sv_yes
+            : &PL_sv_no;
+    } else if (flags & RXapif_REGNAMES) {
+        return reg_named_buff_all(rx, flags);
+    } else if (flags & (RXapif_SCALAR | RXapif_REGNAMES_COUNT)) {
+        return reg_named_buff_scalar(rx, flags);
+    } else {
+        Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff", (int)flags);
+        return NULL;
+    }
+}
+
+SV*
+Perl_reg_named_buff_iter(pTHX_ REGEXP * const rx, const SV * const lastkey,
+                         const U32 flags)
+{
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_ITER;
+    PERL_UNUSED_ARG(lastkey);
+
+    if (flags & RXapif_FIRSTKEY)
+        return reg_named_buff_firstkey(rx, flags);
+    else if (flags & RXapif_NEXTKEY)
+        return reg_named_buff_nextkey(rx, flags);
+    else {
+        Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_iter", (int)flags);
+        return NULL;
+    }
+}
+
+SV*
+Perl_reg_named_buff_fetch(pTHX_ REGEXP * const r, SV * const namesv,
+                         const U32 flags)
+{
+    AV *retarray = NULL;
+    SV *ret;
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_FETCH;
+
+    if (flags & RXapif_ALL)
+        retarray=newAV();
+
+    if (rx && RXp_PAREN_NAMES(rx)) {
+        HE *he_str = hv_fetch_ent( RXp_PAREN_NAMES(rx), namesv, 0, 0 );
+        if (he_str) {
+            IV i;
+            SV* sv_dat=HeVAL(he_str);
+            I32 *nums=(I32*)SvPVX(sv_dat);
+            for ( i=0; i<SvIVX(sv_dat); i++ ) {
+                if ((I32)(rx->nparens) >= nums[i]
+                    && rx->offs[nums[i]].start != -1
+                    && rx->offs[nums[i]].end != -1)
+                {
+                    ret = newSVpvs("");
+                    CALLREG_NUMBUF_FETCH(r,nums[i],ret);
+                    if (!retarray)
+                        return ret;
+                } else {
+                    ret = newSVsv(&PL_sv_undef);
                 }
                 }
+                if (retarray)
+                    av_push(retarray, ret);
             }
             }
+            if (retarray)
+                return newRV_noinc(MUTABLE_SV(retarray));
         }
     }
         }
     }
-    if ( !parno ) {
-        return 0;
+    return NULL;
+}
+
+bool
+Perl_reg_named_buff_exists(pTHX_ REGEXP * const r, SV * const key,
+                           const U32 flags)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_EXISTS;
+
+    if (rx && RXp_PAREN_NAMES(rx)) {
+        if (flags & RXapif_ALL) {
+            return hv_exists_ent(RXp_PAREN_NAMES(rx), key, 0);
+        } else {
+           SV *sv = CALLREG_NAMED_BUFF_FETCH(r, key, flags);
+            if (sv) {
+               SvREFCNT_dec(sv);
+                return TRUE;
+            } else {
+                return FALSE;
+            }
+        }
     } else {
     } else {
-        GV *gv_paren;
-        SV *sv= sv_newmortal();
-        Perl_sv_setpvf(aTHX_ sv, "%"IVdf,(IV)parno);
-        gv_paren= Perl_gv_fetchsv(aTHX_ sv, GV_ADD, SVt_PVGV);
-        return GvSVn(gv_paren);
+        return FALSE;
     }
 }
     }
 }
-#endif
 
 
-/* Scans the name of a named buffer from the pattern.
- * If flags is REG_RSN_RETURN_NULL returns null.
- * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
- * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
- * to the parsed name as looked up in the RExC_paren_names hash.
- * If there is an error throws a vFAIL().. type exception.
- */
+SV*
+Perl_reg_named_buff_firstkey(pTHX_ REGEXP * const r, const U32 flags)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
 
 
-#define REG_RSN_RETURN_NULL    0
-#define REG_RSN_RETURN_NAME    1
-#define REG_RSN_RETURN_DATA    2
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_FIRSTKEY;
 
 
-STATIC SV*
-S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags) {
-    char *name_start = RExC_parse;
-    if ( UTF ) {
-       STRLEN numlen;
-        while( isIDFIRST_uni(utf8n_to_uvchr((U8*)RExC_parse,
-            RExC_end - RExC_parse, &numlen, UTF8_ALLOW_DEFAULT)))
-        {
-                RExC_parse += numlen;
-        }
+    if ( rx && RXp_PAREN_NAMES(rx) ) {
+       (void)hv_iterinit(RXp_PAREN_NAMES(rx));
+
+       return CALLREG_NAMED_BUFF_NEXTKEY(r, NULL, flags & ~RXapif_FIRSTKEY);
     } else {
     } else {
-        while( isIDFIRST(*RExC_parse) )
-           RExC_parse++;
+       return FALSE;
     }
     }
-    if ( flags ) {
-        SV* sv_name = sv_2mortal(Perl_newSVpvn(aTHX_ name_start,
-            (int)(RExC_parse - name_start)));
-       if (UTF)
-            SvUTF8_on(sv_name);
-        if ( flags == REG_RSN_RETURN_NAME)
-            return sv_name;
-        else if (flags==REG_RSN_RETURN_DATA) {
+}
+
+SV*
+Perl_reg_named_buff_nextkey(pTHX_ REGEXP * const r, const U32 flags)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+    GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_NEXTKEY;
+
+    if (rx && RXp_PAREN_NAMES(rx)) {
+        HV *hv = RXp_PAREN_NAMES(rx);
+        HE *temphe;
+        while ( (temphe = hv_iternext_flags(hv,0)) ) {
+            IV i;
+            IV parno = 0;
+            SV* sv_dat = HeVAL(temphe);
+            I32 *nums = (I32*)SvPVX(sv_dat);
+            for ( i = 0; i < SvIVX(sv_dat); i++ ) {
+                if ((I32)(rx->lastparen) >= nums[i] &&
+                    rx->offs[nums[i]].start != -1 &&
+                    rx->offs[nums[i]].end != -1)
+                {
+                    parno = nums[i];
+                    break;
+                }
+            }
+            if (parno || flags & RXapif_ALL) {
+               return newSVhek(HeKEY_hek(temphe));
+            }
+        }
+    }
+    return NULL;
+}
+
+SV*
+Perl_reg_named_buff_scalar(pTHX_ REGEXP * const r, const U32 flags)
+{
+    SV *ret;
+    AV *av;
+    I32 length;
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_SCALAR;
+
+    if (rx && RXp_PAREN_NAMES(rx)) {
+        if (flags & (RXapif_ALL | RXapif_REGNAMES_COUNT)) {
+            return newSViv(HvTOTALKEYS(RXp_PAREN_NAMES(rx)));
+        } else if (flags & RXapif_ONE) {
+            ret = CALLREG_NAMED_BUFF_ALL(r, (flags | RXapif_REGNAMES));
+            av = MUTABLE_AV(SvRV(ret));
+            length = av_len(av);
+           SvREFCNT_dec(ret);
+            return newSViv(length + 1);
+        } else {
+            Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar", (int)flags);
+            return NULL;
+        }
+    }
+    return &PL_sv_undef;
+}
+
+SV*
+Perl_reg_named_buff_all(pTHX_ REGEXP * const r, const U32 flags)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+    AV *av = newAV();
+
+    PERL_ARGS_ASSERT_REG_NAMED_BUFF_ALL;
+
+    if (rx && RXp_PAREN_NAMES(rx)) {
+        HV *hv= RXp_PAREN_NAMES(rx);
+        HE *temphe;
+        (void)hv_iterinit(hv);
+        while ( (temphe = hv_iternext_flags(hv,0)) ) {
+            IV i;
+            IV parno = 0;
+            SV* sv_dat = HeVAL(temphe);
+            I32 *nums = (I32*)SvPVX(sv_dat);
+            for ( i = 0; i < SvIVX(sv_dat); i++ ) {
+                if ((I32)(rx->lastparen) >= nums[i] &&
+                    rx->offs[nums[i]].start != -1 &&
+                    rx->offs[nums[i]].end != -1)
+                {
+                    parno = nums[i];
+                    break;
+                }
+            }
+            if (parno || flags & RXapif_ALL) {
+                av_push(av, newSVhek(HeKEY_hek(temphe)));
+            }
+        }
+    }
+
+    return newRV_noinc(MUTABLE_SV(av));
+}
+
+void
+Perl_reg_numbered_buff_fetch(pTHX_ REGEXP * const r, const I32 paren,
+                            SV * const sv)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+    char *s = NULL;
+    I32 i = 0;
+    I32 s1, t1;
+
+    PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_FETCH;
+        
+    if (!rx->subbeg) {
+        sv_setsv(sv,&PL_sv_undef);
+        return;
+    } 
+    else               
+    if (paren == RX_BUFF_IDX_PREMATCH && rx->offs[0].start != -1) {
+        /* $` */
+       i = rx->offs[0].start;
+       s = rx->subbeg;
+    }
+    else 
+    if (paren == RX_BUFF_IDX_POSTMATCH && rx->offs[0].end != -1) {
+        /* $' */
+       s = rx->subbeg + rx->offs[0].end;
+       i = rx->sublen - rx->offs[0].end;
+    } 
+    else
+    if ( 0 <= paren && paren <= (I32)rx->nparens &&
+        (s1 = rx->offs[paren].start) != -1 &&
+        (t1 = rx->offs[paren].end) != -1)
+    {
+        /* $& $1 ... */
+        i = t1 - s1;
+        s = rx->subbeg + s1;
+    } else {
+        sv_setsv(sv,&PL_sv_undef);
+        return;
+    }          
+    assert(rx->sublen >= (s - rx->subbeg) + i );
+    if (i >= 0) {
+        const int oldtainted = PL_tainted;
+        TAINT_NOT;
+        sv_setpvn(sv, s, i);
+        PL_tainted = oldtainted;
+        if ( (rx->extflags & RXf_CANY_SEEN)
+            ? (RXp_MATCH_UTF8(rx)
+                        && (!i || is_utf8_string((U8*)s, i)))
+            : (RXp_MATCH_UTF8(rx)) )
+        {
+            SvUTF8_on(sv);
+        }
+        else
+            SvUTF8_off(sv);
+        if (PL_tainting) {
+            if (RXp_MATCH_TAINTED(rx)) {
+                if (SvTYPE(sv) >= SVt_PVMG) {
+                    MAGIC* const mg = SvMAGIC(sv);
+                    MAGIC* mgt;
+                    PL_tainted = 1;
+                    SvMAGIC_set(sv, mg->mg_moremagic);
+                    SvTAINT(sv);
+                    if ((mgt = SvMAGIC(sv))) {
+                        mg->mg_moremagic = mgt;
+                        SvMAGIC_set(sv, mg);
+                    }
+                } else {
+                    PL_tainted = 1;
+                    SvTAINT(sv);
+                }
+            } else 
+                SvTAINTED_off(sv);
+        }
+    } else {
+        sv_setsv(sv,&PL_sv_undef);
+        return;
+    }
+}
+
+void
+Perl_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
+                                                        SV const * const value)
+{
+    PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_STORE;
+
+    PERL_UNUSED_ARG(rx);
+    PERL_UNUSED_ARG(paren);
+    PERL_UNUSED_ARG(value);
+
+    if (!PL_localizing)
+        Perl_croak(aTHX_ "%s", PL_no_modify);
+}
+
+I32
+Perl_reg_numbered_buff_length(pTHX_ REGEXP * const r, const SV * const sv,
+                              const I32 paren)
+{
+    struct regexp *const rx = (struct regexp *)SvANY(r);
+    I32 i;
+    I32 s1, t1;
+
+    PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_LENGTH;
+
+    /* Some of this code was originally in C<Perl_magic_len> in F<mg.c> */
+       switch (paren) {
+      /* $` / ${^PREMATCH} */
+      case RX_BUFF_IDX_PREMATCH:
+        if (rx->offs[0].start != -1) {
+                       i = rx->offs[0].start;
+                       if (i > 0) {
+                               s1 = 0;
+                               t1 = i;
+                               goto getlen;
+                       }
+           }
+        return 0;
+      /* $' / ${^POSTMATCH} */
+      case RX_BUFF_IDX_POSTMATCH:
+           if (rx->offs[0].end != -1) {
+                       i = rx->sublen - rx->offs[0].end;
+                       if (i > 0) {
+                               s1 = rx->offs[0].end;
+                               t1 = rx->sublen;
+                               goto getlen;
+                       }
+           }
+        return 0;
+      /* $& / ${^MATCH}, $1, $2, ... */
+      default:
+           if (paren <= (I32)rx->nparens &&
+            (s1 = rx->offs[paren].start) != -1 &&
+            (t1 = rx->offs[paren].end) != -1)
+           {
+            i = t1 - s1;
+            goto getlen;
+        } else {
+            if (ckWARN(WARN_UNINITIALIZED))
+                report_uninit((const SV *)sv);
+            return 0;
+        }
+    }
+  getlen:
+    if (i > 0 && RXp_MATCH_UTF8(rx)) {
+        const char * const s = rx->subbeg + s1;
+        const U8 *ep;
+        STRLEN el;
+
+        i = t1 - s1;
+        if (is_utf8_string_loclen((U8*)s, i, &ep, &el))
+                       i = el;
+    }
+    return i;
+}
+
+SV*
+Perl_reg_qr_package(pTHX_ REGEXP * const rx)
+{
+    PERL_ARGS_ASSERT_REG_QR_PACKAGE;
+       PERL_UNUSED_ARG(rx);
+       if (0)
+           return NULL;
+       else
+           return newSVpvs("Regexp");
+}
+
+/* Scans the name of a named buffer from the pattern.
+ * If flags is REG_RSN_RETURN_NULL returns null.
+ * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
+ * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
+ * to the parsed name as looked up in the RExC_paren_names hash.
+ * If there is an error throws a vFAIL().. type exception.
+ */
+
+#define REG_RSN_RETURN_NULL    0
+#define REG_RSN_RETURN_NAME    1
+#define REG_RSN_RETURN_DATA    2
+
+STATIC SV*
+S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags)
+{
+    char *name_start = RExC_parse;
+
+    PERL_ARGS_ASSERT_REG_SCAN_NAME;
+
+    if (isIDFIRST_lazy_if(RExC_parse, UTF)) {
+        /* skip IDFIRST by using do...while */
+       if (UTF)
+           do {
+               RExC_parse += UTF8SKIP(RExC_parse);
+           } while (isALNUM_utf8((U8*)RExC_parse));
+       else
+           do {
+               RExC_parse++;
+           } while (isALNUM(*RExC_parse));
+    }
+
+    if ( flags ) {
+        SV* sv_name
+           = newSVpvn_flags(name_start, (int)(RExC_parse - name_start),
+                            SVs_TEMP | (UTF ? SVf_UTF8 : 0));
+        if ( flags == REG_RSN_RETURN_NAME)
+            return sv_name;
+        else if (flags==REG_RSN_RETURN_DATA) {
             HE *he_str = NULL;
             SV *sv_dat = NULL;
             if ( ! sv_name )      /* should not happen*/
             HE *he_str = NULL;
             SV *sv_dat = NULL;
             if ( ! sv_name )      /* should not happen*/
@@ -4648,7 +5400,7 @@ S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags) {
         PerlIO_printf(Perl_debug_log,"%16s","");                \
                                                                 \
     if (SIZE_ONLY)                                              \
         PerlIO_printf(Perl_debug_log,"%16s","");                \
                                                                 \
     if (SIZE_ONLY)                                              \
-       num=RExC_size;                                           \
+       num = RExC_size + 1;                                     \
     else                                                        \
        num=REG_NODE_NUM(RExC_emit);                             \
     if (RExC_lastnum!=num)                                      \
     else                                                        \
        num=REG_NODE_NUM(RExC_emit);                             \
     if (RExC_lastnum!=num)                                      \
@@ -4689,10 +5441,6 @@ S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags) {
 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
 #endif
 
 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
 #endif
 
-/* this idea is borrowed from STR_WITH_LEN in handy.h */
-#define CHECK_WORD(s,v,l)  \
-    (((sizeof(s)-1)==(l)) && (strnEQ(start_verb, (s ""), (sizeof(s)-1))))
-
 STATIC regnode *
 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
 STATIC regnode *
 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
@@ -4704,9 +5452,11 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     register regnode *ender = NULL;
     register I32 parno = 0;
     I32 flags;
     register regnode *ender = NULL;
     register I32 parno = 0;
     I32 flags;
-    const I32 oregflags = RExC_flags;
+    U32 oregflags = RExC_flags;
     bool have_branch = 0;
     bool is_open = 0;
     bool have_branch = 0;
     bool is_open = 0;
+    I32 freeze_paren = 0;
+    I32 after_freeze = 0;
 
     /* for (?g), (?gc), and (?o) warnings; warning
        about (?c) will warn about (?g) -- japhy    */
 
     /* for (?g), (?gc), and (?o) warnings; warning
        about (?c) will warn about (?g) -- japhy    */
@@ -4721,8 +5471,9 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     char * const oregcomp_parse = RExC_parse;
 
     GET_RE_DEBUG_FLAGS_DECL;
     char * const oregcomp_parse = RExC_parse;
 
     GET_RE_DEBUG_FLAGS_DECL;
-    DEBUG_PARSE("reg ");
 
 
+    PERL_ARGS_ASSERT_REG;
+    DEBUG_PARSE("reg ");
 
     *flagp = 0;                                /* Tentatively. */
 
 
     *flagp = 0;                                /* Tentatively. */
 
@@ -4760,39 +5511,39 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            
            switch ( *start_verb ) {
             case 'A':  /* (*ACCEPT) */
            
            switch ( *start_verb ) {
             case 'A':  /* (*ACCEPT) */
-                if ( CHECK_WORD("ACCEPT",start_verb,verb_len) ) {
+                if ( memEQs(start_verb,verb_len,"ACCEPT") ) {
                    op = ACCEPT;
                    internal_argval = RExC_nestroot;
                }
                break;
             case 'C':  /* (*COMMIT) */
                    op = ACCEPT;
                    internal_argval = RExC_nestroot;
                }
                break;
             case 'C':  /* (*COMMIT) */
-                if ( CHECK_WORD("COMMIT",start_verb,verb_len) )
+                if ( memEQs(start_verb,verb_len,"COMMIT") )
                     op = COMMIT;
                 break;
             case 'F':  /* (*FAIL) */
                     op = COMMIT;
                 break;
             case 'F':  /* (*FAIL) */
-                if ( verb_len==1 || CHECK_WORD("FAIL",start_verb,verb_len) ) {
+                if ( verb_len==1 || memEQs(start_verb,verb_len,"FAIL") ) {
                    op = OPFAIL;
                    argok = 0;
                }
                break;
             case ':':  /* (*:NAME) */
            case 'M':  /* (*MARK:NAME) */
                    op = OPFAIL;
                    argok = 0;
                }
                break;
             case ':':  /* (*:NAME) */
            case 'M':  /* (*MARK:NAME) */
-               if ( verb_len==0 || CHECK_WORD("MARK",start_verb,verb_len) ) {
+               if ( verb_len==0 || memEQs(start_verb,verb_len,"MARK") ) {
                     op = MARKPOINT;
                     argok = -1;
                 }
                 break;
             case 'P':  /* (*PRUNE) */
                     op = MARKPOINT;
                     argok = -1;
                 }
                 break;
             case 'P':  /* (*PRUNE) */
-                if ( CHECK_WORD("PRUNE",start_verb,verb_len) )
+                if ( memEQs(start_verb,verb_len,"PRUNE") )
                     op = PRUNE;
                 break;
             case 'S':   /* (*SKIP) */  
                     op = PRUNE;
                 break;
             case 'S':   /* (*SKIP) */  
-                if ( CHECK_WORD("SKIP",start_verb,verb_len) ) 
+                if ( memEQs(start_verb,verb_len,"SKIP") ) 
                     op = SKIP;
                 break;
             case 'T':  /* (*THEN) */
                 /* [19:06] <TimToady> :: is then */
                     op = SKIP;
                 break;
             case 'T':  /* (*THEN) */
                 /* [19:06] <TimToady> :: is then */
-                if ( CHECK_WORD("THEN",start_verb,verb_len) ) {
+                if ( memEQs(start_verb,verb_len,"THEN") ) {
                     op = CUTGROUP;
                     RExC_seen |= REG_SEEN_CUTGROUP;
                 }
                     op = CUTGROUP;
                     RExC_seen |= REG_SEEN_CUTGROUP;
                 }
@@ -4816,7 +5567,7 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                         if (start_arg) {
                             SV *sv = newSVpvn( start_arg, RExC_parse - start_arg);
                             ARG(ret) = add_data( pRExC_state, 1, "S" );
                         if (start_arg) {
                             SV *sv = newSVpvn( start_arg, RExC_parse - start_arg);
                             ARG(ret) = add_data( pRExC_state, 1, "S" );
-                            RExC_rx->data->data[ARG(ret)]=(void*)sv;
+                            RExC_rxi->data->data[ARG(ret)]=(void*)sv;
                             ret->flags = 0;
                         } else {
                             ret->flags = 1; 
                             ret->flags = 0;
                         } else {
                             ret->flags = 1; 
@@ -4835,8 +5586,6 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            return ret;
         } else 
        if (*RExC_parse == '?') { /* (?...) */
            return ret;
         } else 
        if (*RExC_parse == '?') { /* (?...) */
-           U32 posflags = 0, negflags = 0;
-           U32 *flagsp = &posflags;
            bool is_logical = 0;
            const char * const seqstart = RExC_parse;
 
            bool is_logical = 0;
            const char * const seqstart = RExC_parse;
 
@@ -4845,10 +5594,48 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            ret = NULL;                 /* For look-ahead/behind. */
            switch (paren) {
 
            ret = NULL;                 /* For look-ahead/behind. */
            switch (paren) {
 
-           case '<':           /* (?<...) */
+           case 'P':   /* (?P...) variants for those used to PCRE/Python */
+               paren = *RExC_parse++;
+               if ( paren == '<')         /* (?P<...>) named capture */
+                   goto named_capture;
+                else if (paren == '>') {   /* (?P>name) named recursion */
+                    goto named_recursion;
+                }
+                else if (paren == '=') {   /* (?P=...)  named backref */
+                    /* this pretty much dupes the code for \k<NAME> in regatom(), if
+                       you change this make sure you change that */
+                    char* name_start = RExC_parse;
+                   U32 num = 0;
+                    SV *sv_dat = reg_scan_name(pRExC_state,
+                        SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
+                    if (RExC_parse == name_start || *RExC_parse != ')')
+                        vFAIL2("Sequence %.3s... not terminated",parse_start);
+
+                    if (!SIZE_ONLY) {
+                        num = add_data( pRExC_state, 1, "S" );
+                        RExC_rxi->data->data[num]=(void*)sv_dat;
+                        SvREFCNT_inc_simple_void(sv_dat);
+                    }
+                    RExC_sawback = 1;
+                    ret = reganode(pRExC_state,
+                          (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
+                          num);
+                    *flagp |= HASWIDTH;
+
+                    Set_Node_Offset(ret, parse_start+1);
+                    Set_Node_Cur_Length(ret); /* MJD */
+
+                    nextchar(pRExC_state);
+                    return ret;
+                }
+                RExC_parse++;
+               vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
+               /*NOTREACHED*/
+            case '<':           /* (?<...) */
                if (*RExC_parse == '!')
                    paren = ',';
                else if (*RExC_parse != '=') 
                if (*RExC_parse == '!')
                    paren = ',';
                else if (*RExC_parse != '=') 
+              named_capture:
                {               /* (?<...>) */
                    char *name_start;
                    SV *svname;
                {               /* (?<...>) */
                    char *name_start;
                    SV *svname;
@@ -4859,8 +5646,11 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                        SIZE_ONLY ?  /* reverse test from the others */
                        REG_RSN_RETURN_NAME : 
                        REG_RSN_RETURN_NULL);
                        SIZE_ONLY ?  /* reverse test from the others */
                        REG_RSN_RETURN_NAME : 
                        REG_RSN_RETURN_NULL);
-                   if (RExC_parse == name_start)
-                       goto unknown;
+                   if (RExC_parse == name_start) {
+                       RExC_parse++;
+                       vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
+                       /*NOTREACHED*/
+                    }
                    if (*RExC_parse != paren)
                        vFAIL2("Sequence (?%c... not terminated",
                            paren=='>' ? '<' : paren);
                    if (*RExC_parse != paren)
                        vFAIL2("Sequence (?%c... not terminated",
                            paren=='>' ? '<' : paren);
@@ -4872,7 +5662,11 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                                 "panic: reg_scan_name returned NULL");
                         if (!RExC_paren_names) {
                             RExC_paren_names= newHV();
                                 "panic: reg_scan_name returned NULL");
                         if (!RExC_paren_names) {
                             RExC_paren_names= newHV();
-                            sv_2mortal((SV*)RExC_paren_names);
+                            sv_2mortal(MUTABLE_SV(RExC_paren_names));
+#ifdef DEBUGGING
+                            RExC_paren_name_list= newAV();
+                            sv_2mortal(MUTABLE_SV(RExC_paren_name_list));
+#endif
                         }
                         he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
                         if ( he_str )
                         }
                         he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
                         if ( he_str )
@@ -4882,17 +5676,36 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                             Perl_croak(aTHX_
                                 "panic: paren_name hash element allocation failed");
                         } else if ( SvPOK(sv_dat) ) {
                             Perl_croak(aTHX_
                                 "panic: paren_name hash element allocation failed");
                         } else if ( SvPOK(sv_dat) ) {
-                            IV count=SvIV(sv_dat);
-                            I32 *pv=(I32*)SvGROW(sv_dat,SvCUR(sv_dat)+sizeof(I32)+1);
-                            SvCUR_set(sv_dat,SvCUR(sv_dat)+sizeof(I32));
-                            pv[count]=RExC_npar;
-                            SvIVX(sv_dat)++;
+                            /* (?|...) can mean we have dupes so scan to check
+                               its already been stored. Maybe a flag indicating
+                               we are inside such a construct would be useful,
+                               but the arrays are likely to be quite small, so
+                               for now we punt -- dmq */
+                            IV count = SvIV(sv_dat);
+                            I32 *pv = (I32*)SvPVX(sv_dat);
+                            IV i;
+                            for ( i = 0 ; i < count ; i++ ) {
+                                if ( pv[i] == RExC_npar ) {
+                                    count = 0;
+                                    break;
+                                }
+                            }
+                            if ( count ) {
+                                pv = (I32*)SvGROW(sv_dat, SvCUR(sv_dat) + sizeof(I32)+1);
+                                SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
+                                pv[count] = RExC_npar;
+                                SvIV_set(sv_dat, SvIVX(sv_dat) + 1);
+                            }
                         } else {
                             (void)SvUPGRADE(sv_dat,SVt_PVNV);
                             sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
                             SvIOK_on(sv_dat);
                         } else {
                             (void)SvUPGRADE(sv_dat,SVt_PVNV);
                             sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
                             SvIOK_on(sv_dat);
-                            SvIVX(sv_dat)= 1;
+                            SvIV_set(sv_dat, 1);
                         }
                         }
+#ifdef DEBUGGING
+                        if (!av_store(RExC_paren_name_list, RExC_npar, SvREFCNT_inc(svname)))
+                            SvREFCNT_dec(svname);
+#endif
 
                         /*sv_dump(sv_dat);*/
                     }
 
                         /*sv_dump(sv_dat);*/
                     }
@@ -4903,6 +5716,8 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                 RExC_seen |= REG_SEEN_LOOKBEHIND;
                RExC_parse++;
            case '=':           /* (?=...) */
                 RExC_seen |= REG_SEEN_LOOKBEHIND;
                RExC_parse++;
            case '=':           /* (?=...) */
+               RExC_seen_zerolen++;
+                       break;
            case '!':           /* (?!...) */
                RExC_seen_zerolen++;
                if (*RExC_parse == ')') {
            case '!':           /* (?!...) */
                RExC_seen_zerolen++;
                if (*RExC_parse == ')') {
@@ -4910,6 +5725,13 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                    nextchar(pRExC_state);
                    return ret;
                }
                    nextchar(pRExC_state);
                    return ret;
                }
+               break;
+           case '|':           /* (?|...) */
+               /* branch reset, behave like a (?:...) except that
+                  buffers in alternations share the same numbers */
+               paren = ':'; 
+               after_freeze = freeze_paren = RExC_npar;
+               break;
            case ':':           /* (?:...) */
            case '>':           /* (?>...) */
                break;
            case ':':           /* (?:...) */
            case '>':           /* (?>...) */
                break;
@@ -4930,14 +5752,15 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                if (*RExC_parse != ')')
                    FAIL("Sequence (?R) not terminated");
                ret = reg_node(pRExC_state, GOSTART);
                if (*RExC_parse != ')')
                    FAIL("Sequence (?R) not terminated");
                ret = reg_node(pRExC_state, GOSTART);
+               *flagp |= POSTPONED;
                nextchar(pRExC_state);
                return ret;
                /*notreached*/
             { /* named and numeric backreferences */
                 I32 num;
                nextchar(pRExC_state);
                return ret;
                /*notreached*/
             { /* named and numeric backreferences */
                 I32 num;
-                char * parse_start;
             case '&':            /* (?&NAME) */
                 parse_start = RExC_parse - 1;
             case '&':            /* (?&NAME) */
                 parse_start = RExC_parse - 1;
+              named_recursion:
                 {
                    SV *sv_dat = reg_scan_name(pRExC_state,
                        SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
                 {
                    SV *sv_dat = reg_scan_name(pRExC_state,
                        SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
@@ -5010,24 +5833,26 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
                Set_Node_Offset(ret, parse_start); /* MJD */
 
                 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
                Set_Node_Offset(ret, parse_start); /* MJD */
 
+                *flagp |= POSTPONED;
                 nextchar(pRExC_state);
                 return ret;
             } /* named and numeric backreferences */
             /* NOT REACHED */
 
                 nextchar(pRExC_state);
                 return ret;
             } /* named and numeric backreferences */
             /* NOT REACHED */
 
-           case 'p':           /* (?p...) */
-               if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
-                   vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
-               /* FALL THROUGH*/
            case '?':           /* (??...) */
                is_logical = 1;
            case '?':           /* (??...) */
                is_logical = 1;
-               if (*RExC_parse != '{')
-                   goto unknown;
+               if (*RExC_parse != '{') {
+                   RExC_parse++;
+                   vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
+                   /*NOTREACHED*/
+               }
+               *flagp |= POSTPONED;
                paren = *RExC_parse++;
                /* FALL THROUGH */
            case '{':           /* (?{...}) */
            {
                paren = *RExC_parse++;
                /* FALL THROUGH */
            case '{':           /* (?{...}) */
            {
-               I32 count = 1, n = 0;
+               I32 count = 1;
+               U32 n = 0;
                char c;
                char *s = RExC_parse;
 
                char c;
                char *s = RExC_parse;
 
@@ -5062,9 +5887,9 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                    LEAVE;
 
                    n = add_data(pRExC_state, 3, "nop");
                    LEAVE;
 
                    n = add_data(pRExC_state, 3, "nop");
-                   RExC_rx->data->data[n] = (void*)rop;
-                   RExC_rx->data->data[n+1] = (void*)sop;
-                   RExC_rx->data->data[n+2] = (void*)pad;
+                   RExC_rxi->data->data[n] = (void*)rop;
+                   RExC_rxi->data->data[n+1] = (void*)sop;
+                   RExC_rxi->data->data[n+2] = (void*)pad;
                    SvREFCNT_dec(sv);
                }
                else {                                          /* First pass */
                    SvREFCNT_dec(sv);
                }
                else {                                          /* First pass */
@@ -5116,7 +5941,7 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                {
                    char ch = RExC_parse[0] == '<' ? '>' : '\'';
                    char *name_start= RExC_parse++;
                {
                    char ch = RExC_parse[0] == '<' ? '>' : '\'';
                    char *name_start= RExC_parse++;
-                   I32 num = 0;
+                   U32 num = 0;
                    SV *sv_dat=reg_scan_name(pRExC_state,
                        SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
                    if (RExC_parse == name_start || *RExC_parse != ch)
                    SV *sv_dat=reg_scan_name(pRExC_state,
                        SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
                    if (RExC_parse == name_start || *RExC_parse != ch)
@@ -5125,8 +5950,8 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                     RExC_parse++;
                    if (!SIZE_ONLY) {
                         num = add_data( pRExC_state, 1, "S" );
                     RExC_parse++;
                    if (!SIZE_ONLY) {
                         num = add_data( pRExC_state, 1, "S" );
-                        RExC_rx->data->data[num]=(void*)sv_dat;
-                        SvREFCNT_inc(sv_dat);
+                        RExC_rxi->data->data[num]=(void*)sv_dat;
+                        SvREFCNT_inc_simple_void(sv_dat);
                     }
                     ret = reganode(pRExC_state,NGROUPP,num);
                     goto insert_if_check_paren;
                     }
                     ret = reganode(pRExC_state,NGROUPP,num);
                     goto insert_if_check_paren;
@@ -5204,6 +6029,9 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                    }
                    else
                         REGTAIL(pRExC_state, ret, ender);
                    }
                    else
                         REGTAIL(pRExC_state, ret, ender);
+                    RExC_size++; /* XXX WHY do we need this?!!
+                                    For large programs it seems to be required
+                                    but I can't figure out why. -- dmq*/
                    return ret;
                }
                else {
                    return ret;
                }
                else {
@@ -5215,13 +6043,20 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                 vFAIL("Sequence (? incomplete");
                 break;
            default:
                 vFAIL("Sequence (? incomplete");
                 break;
            default:
-               --RExC_parse;
-             parse_flags:      /* (?i) */
-               while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
+               --RExC_parse;
+               parse_flags:      /* (?i) */  
+           {
+                U32 posflags = 0, negflags = 0;
+               U32 *flagsp = &posflags;
+
+               while (*RExC_parse) {
+                   /* && strchr("iogcmsx", *RExC_parse) */
                    /* (?g), (?gc) and (?o) are useless here
                       and must be globally applied -- japhy */
                    /* (?g), (?gc) and (?o) are useless here
                       and must be globally applied -- japhy */
-
-                   if (*RExC_parse == 'o' || *RExC_parse == 'g') {
+                    switch (*RExC_parse) {
+                   CASE_STD_PMMOD_FLAGS_PARSE_SET(flagsp);
+                    case ONCE_PAT_MOD: /* 'o' */
+                    case GLOBAL_PAT_MOD: /* 'g' */
                        if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
                            const I32 wflagbit = *RExC_parse == 'o' ? WASTED_O : WASTED_G;
                            if (! (wastedflags & wflagbit) ) {
                        if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
                            const I32 wflagbit = *RExC_parse == 'o' ? WASTED_O : WASTED_G;
                            if (! (wastedflags & wflagbit) ) {
@@ -5236,8 +6071,9 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                                );
                            }
                        }
                                );
                            }
                        }
-                   }
-                   else if (*RExC_parse == 'c') {
+                       break;
+                       
+                   case CONTINUE_PAT_MOD: /* 'c' */
                        if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
                            if (! (wastedflags & WASTED_C) ) {
                                wastedflags |= WASTED_GC;
                        if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
                            if (! (wastedflags & WASTED_C) ) {
                                wastedflags |= WASTED_GC;
@@ -5249,33 +6085,51 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
                                );
                            }
                        }
                                );
                            }
                        }
-                   }
-                   else { pmflag(flagsp, *RExC_parse); }
-
-                   ++RExC_parse;
-               }
-               if (*RExC_parse == '-') {
-                   flagsp = &negflags;
-                   wastedflags = 0;  /* reset so (?g-c) warns twice */
+                       break;
+                   case KEEPCOPY_PAT_MOD: /* 'p' */
+                        if (flagsp == &negflags) {
+                            if (SIZE_ONLY)
+                                ckWARNreg(RExC_parse + 1,"Useless use of (?-p)");
+                        } else {
+                            *flagsp |= RXf_PMf_KEEPCOPY;
+                        }
+                       break;
+                    case '-':
+                        if (flagsp == &negflags) {
+                            RExC_parse++;
+                           vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
+                           /*NOTREACHED*/
+                       }
+                       flagsp = &negflags;
+                       wastedflags = 0;  /* reset so (?g-c) warns twice */
+                       break;
+                    case ':':
+                       paren = ':';
+                       /*FALLTHROUGH*/
+                    case ')':
+                        RExC_flags |= posflags;
+                        RExC_flags &= ~negflags;
+                        if (paren != ':') {
+                            oregflags |= posflags;
+                            oregflags &= ~negflags;
+                        }
+                        nextchar(pRExC_state);
+                       if (paren != ':') {
+                           *flagp = TRYAGAIN;
+                           return NULL;
+                       } else {
+                            ret = NULL;
+                           goto parse_rest;
+                       }
+                       /*NOTREACHED*/
+                    default:
+                       RExC_parse++;
+                       vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
+                       /*NOTREACHED*/
+                    }                           
                    ++RExC_parse;
                    ++RExC_parse;
-                   goto parse_flags;
-               }
-               RExC_flags |= posflags;
-               RExC_flags &= ~negflags;
-               if (*RExC_parse == ':') {
-                   RExC_parse++;
-                   paren = ':';
-                   break;
-               }               
-             unknown:
-               if (*RExC_parse != ')') {
-                   RExC_parse++;
-                   vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
                }
                }
-               nextchar(pRExC_state);
-               *flagp = TRYAGAIN;
-               return NULL;
-           }
+           }} /* one for the default block, one for the switch */
        }
        else {                  /* (...) */
          capturing_parens:
        }
        else {                  /* (...) */
          capturing_parens:
@@ -5286,7 +6140,9 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            if (!SIZE_ONLY ){
                if (!RExC_nestroot) 
                    RExC_nestroot = parno;
            if (!SIZE_ONLY ){
                if (!RExC_nestroot) 
                    RExC_nestroot = parno;
-               if (RExC_seen & REG_SEEN_RECURSE) {
+               if (RExC_seen & REG_SEEN_RECURSE
+                   && !RExC_open_parens[parno-1])
+               {
                    DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
                        "Setting open paren #%"IVdf" to %d\n", 
                        (IV)parno, REG_NODE_NUM(ret)));
                    DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
                        "Setting open paren #%"IVdf" to %d\n", 
                        (IV)parno, REG_NODE_NUM(ret)));
@@ -5300,10 +6156,18 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     }
     else                        /* ! paren */
        ret = NULL;
     }
     else                        /* ! paren */
        ret = NULL;
-
+   
+   parse_rest:
     /* Pick up the branches, linking them together. */
     parse_start = RExC_parse;   /* MJD */
     br = regbranch(pRExC_state, &flags, 1,depth+1);
     /* Pick up the branches, linking them together. */
     parse_start = RExC_parse;   /* MJD */
     br = regbranch(pRExC_state, &flags, 1,depth+1);
+
+    if (freeze_paren) {
+        if (RExC_npar > after_freeze)
+            after_freeze = RExC_npar;
+        RExC_npar = freeze_paren;
+    }
+
     /*     branch_len = (paren != 0); */
 
     if (br == NULL)
     /*     branch_len = (paren != 0); */
 
     if (br == NULL)
@@ -5329,7 +6193,7 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
     }
     else if (paren != '?')             /* Not Conditional */
        ret = br;
     }
     else if (paren != '?')             /* Not Conditional */
        ret = br;
-    *flagp |= flags & (SPSTART | HASWIDTH);
+    *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
     lastbr = br;
     while (*RExC_parse == '|') {
        if (!SIZE_ONLY && RExC_extralen) {
     lastbr = br;
     while (*RExC_parse == '|') {
        if (!SIZE_ONLY && RExC_extralen) {
@@ -5339,15 +6203,18 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
        if (SIZE_ONLY)
            RExC_extralen += 2;         /* Account for LONGJMP. */
        nextchar(pRExC_state);
        if (SIZE_ONLY)
            RExC_extralen += 2;         /* Account for LONGJMP. */
        nextchar(pRExC_state);
+       if (freeze_paren) {
+           if (RExC_npar > after_freeze)
+               after_freeze = RExC_npar;
+            RExC_npar = freeze_paren;      
+        }
         br = regbranch(pRExC_state, &flags, 0, depth+1);
 
        if (br == NULL)
            return(NULL);
         REGTAIL(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
        lastbr = br;
         br = regbranch(pRExC_state, &flags, 0, depth+1);
 
        if (br == NULL)
            return(NULL);
         REGTAIL(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
        lastbr = br;
-       if (flags&HASWIDTH)
-           *flagp |= HASWIDTH;
-       *flagp |= flags&SPSTART;
+       *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
     }
 
     if (have_branch || paren != ':') {
     }
 
     if (have_branch || paren != ':') {
@@ -5357,7 +6224,6 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            ender = reg_node(pRExC_state, TAIL);
            break;
        case 1:
            ender = reg_node(pRExC_state, TAIL);
            break;
        case 1:
-           RExC_cpar++;
            ender = reganode(pRExC_state, CLOSE, parno);
            if (!SIZE_ONLY && RExC_seen & REG_SEEN_RECURSE) {
                DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
            ender = reganode(pRExC_state, CLOSE, parno);
            if (!SIZE_ONLY && RExC_seen & REG_SEEN_RECURSE) {
                DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
@@ -5441,7 +6307,8 @@ S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
            FAIL("Junk on end of regexp");      /* "Can't happen". */
        /* NOTREACHED */
     }
            FAIL("Junk on end of regexp");      /* "Can't happen". */
        /* NOTREACHED */
     }
-
+    if (after_freeze)
+        RExC_npar = after_freeze;
     return(ret);
 }
 
     return(ret);
 }
 
@@ -5459,7 +6326,11 @@ S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
     register regnode *latest;
     I32 flags = 0, c = 0;
     GET_RE_DEBUG_FLAGS_DECL;
     register regnode *latest;
     I32 flags = 0, c = 0;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REGBRANCH;
+
     DEBUG_PARSE("brnc");
     DEBUG_PARSE("brnc");
+
     if (first)
        ret = NULL;
     else {
     if (first)
        ret = NULL;
     else {
@@ -5488,7 +6359,7 @@ S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
        }
        else if (ret == NULL)
            ret = latest;
        }
        else if (ret == NULL)
            ret = latest;
-       *flagp |= flags&HASWIDTH;
+       *flagp |= flags&(HASWIDTH|POSTPONED);
        if (chain == NULL)      /* First piece. */
            *flagp |= flags&SPSTART;
        else {
        if (chain == NULL)      /* First piece. */
            *flagp |= flags&SPSTART;
        else {
@@ -5533,6 +6404,9 @@ S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
     char *parse_start;
     const char *maxpos = NULL;
     GET_RE_DEBUG_FLAGS_DECL;
     char *parse_start;
     const char *maxpos = NULL;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REGPIECE;
+
     DEBUG_PARSE("piec");
 
     ret = regatom(pRExC_state, &flags,depth+1);
     DEBUG_PARSE("piec");
 
     ret = regatom(pRExC_state, &flags,depth+1);
@@ -5610,7 +6484,7 @@ S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
                *flagp = WORST;
            if (max > 0)
                *flagp |= HASWIDTH;
                *flagp = WORST;
            if (max > 0)
                *flagp |= HASWIDTH;
-           if (max && max < min)
+           if (max < min)
                vFAIL("Can't do {n,m} with n > m");
            if (!SIZE_ONLY) {
                ARG1_SET(ret, (U16)min);
                vFAIL("Can't do {n,m} with n > m");
            if (!SIZE_ONLY) {
                ARG1_SET(ret, (U16)min);
@@ -5670,11 +6544,11 @@ S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
        goto do_curly;
     }
   nest_check:
        goto do_curly;
     }
   nest_check:
-    if (!SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3 && ckWARN(WARN_REGEXP)) {
-       vWARN3(RExC_parse,
-              "%.*s matches null string many times",
-              (int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
-              origparse);
+    if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3) {
+       ckWARN3reg(RExC_parse,
+                  "%.*s matches null string many times",
+                  (int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
+                  origparse);
     }
 
     if (RExC_parse < RExC_end && *RExC_parse == '?') {
     }
 
     if (RExC_parse < RExC_end && *RExC_parse == '?') {
@@ -5709,267 +6583,242 @@ S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 /* reg_namedseq(pRExC_state,UVp)
    
    This is expected to be called by a parser routine that has 
 /* reg_namedseq(pRExC_state,UVp)
    
    This is expected to be called by a parser routine that has 
-   recognized'\N' and needs to handle the rest. RExC_parse is 
+   recognized '\N' and needs to handle the rest. RExC_parse is
    expected to point at the first char following the N at the time
    of the call.
    expected to point at the first char following the N at the time
    of the call.
+
+   The \N may be inside (indicated by valuep not being NULL) or outside a
+   character class.
+
+   \N may begin either a named sequence, or if outside a character class, mean
+   to match a non-newline.  For non single-quoted regexes, the tokenizer has
+   attempted to decide which, and in the case of a named sequence converted it
+   into one of the forms: \N{} (if the sequence is null), or \N{U+c1.c2...},
+   where c1... are the characters in the sequence.  For single-quoted regexes,
+   the tokenizer passes the \N sequence through unchanged; this code will not
+   attempt to determine this nor expand those.  The net effect is that if the
+   beginning of the passed-in pattern isn't '{U+' or there is no '}', it
+   signals that this \N occurrence means to match a non-newline.
+   
+   Only the \N{U+...} form should occur in a character class, for the same
+   reason that '.' inside a character class means to just match a period: it
+   just doesn't make sense.
    
    If valuep is non-null then it is assumed that we are parsing inside 
    of a charclass definition and the first codepoint in the resolved
    string is returned via *valuep and the routine will return NULL. 
    In this mode if a multichar string is returned from the charnames 
    
    If valuep is non-null then it is assumed that we are parsing inside 
    of a charclass definition and the first codepoint in the resolved
    string is returned via *valuep and the routine will return NULL. 
    In this mode if a multichar string is returned from the charnames 
-   handler a warning will be issued, and only the first char in the 
+   handler, a warning will be issued, and only the first char in the 
    sequence will be examined. If the string returned is zero length
    then the value of *valuep is undefined and NON-NULL will 
    be returned to indicate failure. (This will NOT be a valid pointer 
    to a regnode.)
    
    sequence will be examined. If the string returned is zero length
    then the value of *valuep is undefined and NON-NULL will 
    be returned to indicate failure. (This will NOT be a valid pointer 
    to a regnode.)
    
-   If value is null then it is assumed that we are parsing normal text
-   and inserts a new EXACT node into the program containing the resolved
-   string and returns a pointer to the new node. If the string is 
-   zerolength a NOTHING node is emitted.
-   
+   If valuep is null then it is assumed that we are parsing normal text and a
+   new EXACT node is inserted into the program containing the resolved string,
+   and a pointer to the new node is returned.  But if the string is zero length
+   a NOTHING node is emitted instead.
+
    On success RExC_parse is set to the char following the endbrace.
    On success RExC_parse is set to the char following the endbrace.
-   Parsing failures will generate a fatal errorvia vFAIL(...)
-   
-   NOTE: We cache all results from the charnames handler locally in 
-   the RExC_charnames hash (created on first use) to prevent a charnames 
-   handler from playing silly-buggers and returning a short string and 
-   then a long string for a given pattern. Since the regexp program 
-   size is calculated during an initial parse this would result
-   in a buffer overrun so we cache to prevent the charname result from
-   changing during the course of the parse.
-   
+   Parsing failures will generate a fatal error via vFAIL(...)
  */
 STATIC regnode *
  */
 STATIC regnode *
-S_reg_namedseq(pTHX_ RExC_state_t *pRExC_state, UV *valuep
+S_reg_namedseq(pTHX_ RExC_state_t *pRExC_state, UV *valuep, I32 *flagp)
 {
 {
-    char * name;        /* start of the content of the name */
     char * endbrace;    /* endbrace following the name */
     char * endbrace;    /* endbrace following the name */
-    SV *sv_str = NULL;  
-    SV *sv_name = NULL;
-    STRLEN len; /* this has various purposes throughout the code */
-    bool cached = 0; /* if this is true then we shouldn't refcount dev sv_str */
     regnode *ret = NULL;
     regnode *ret = NULL;
+#ifdef DEBUGGING
+    char* parse_start = RExC_parse - 2;            /* points to the '\N' */
+#endif
+
+    GET_RE_DEBUG_FLAGS_DECL;
+    PERL_ARGS_ASSERT_REG_NAMEDSEQ;
+
+    GET_RE_DEBUG_FLAGS;
+   
+    /* Disambiguate between \N meaning a named character versus \N meaning
+     * don't match a newline. */
+    if (*RExC_parse != '{'
+       || (! (endbrace = strchr(RExC_parse, '}'))) /* no trailing brace */
+       || ! (endbrace == RExC_parse + 1        /* nothing between the {} */
+             || (endbrace - RExC_parse > 3     /* U+ and at least one hex */
+                 && strnEQ(RExC_parse + 1, "U+", 2))))
+    {
+       if (valuep) {
+           /* no bare \N in a charclass */
+           vFAIL("\\N in a character class must be a named character: \\N{...}");
+       }
+       nextchar(pRExC_state);
+       ret = reg_node(pRExC_state, REG_ANY);
+       *flagp |= HASWIDTH|SIMPLE;
+       RExC_naughty++;
+       RExC_parse--;
+        Set_Node_Length(ret, 1); /* MJD */
+       return ret;
+    }
+
+    /* Here, we have decided it is a named sequence */
+    RExC_parse++;      /* Skip past the '{' */
+    if (endbrace == RExC_parse) {   /* empty: \N{} */
+       if (! valuep) {
+           RExC_parse = endbrace + 1;  
+           return reg_node(pRExC_state,NOTHING);
+       }
+
+       if (SIZE_ONLY) {
+           ckWARNreg(RExC_parse,
+                   "Ignoring zero length \\N{} in character class"
+           );
+           RExC_parse = endbrace + 1;  
+       }
+       *valuep = 0;
+       return (regnode *) &RExC_parse; /* Invalid regnode pointer */
+    }
+
+    RExC_utf8 = 1;     /* named sequences imply Unicode semantics */
+    RExC_parse += 2;   /* Skip past the 'U+' */
+
+    if (valuep) {   /* In a bracketed char class */
+       /* We only pay attention to the first char of 
+       multichar strings being returned. I kinda wonder
+       if this makes sense as it does change the behaviour
+       from earlier versions, OTOH that behaviour was broken
+       as well. XXX Solution is to recharacterize as
+       [rest-of-class]|multi1|multi2... */
+
+       STRLEN length_of_hex;
+       I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
+           | PERL_SCAN_DISALLOW_PREFIX
+           | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
     
     
-    if (*RExC_parse != '{') {
-        vFAIL("Missing braces on \\N{}");
-    }
-    name = RExC_parse+1;
-    endbrace = strchr(RExC_parse, '}');
-    if ( ! endbrace ) {
-        RExC_parse++;
-        vFAIL("Missing right brace on \\N{}");
-    } 
-    RExC_parse = endbrace + 1;  
-    
-    
-    /* RExC_parse points at the beginning brace, 
-       endbrace points at the last */
-    if ( name[0]=='U' && name[1]=='+' ) {
-        /* its a "unicode hex" notation {U+89AB} */
-        I32 fl = PERL_SCAN_ALLOW_UNDERSCORES
-            | PERL_SCAN_DISALLOW_PREFIX
-            | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
-        UV cp;
-        len = (STRLEN)(endbrace - name - 2);
-        cp = grok_hex(name + 2, &len, &fl, NULL);
-        if ( len != (STRLEN)(endbrace - name - 2) ) {
-            cp = 0xFFFD;
-        }    
-        if (cp > 0xff)
-            RExC_utf8 = 1;
-        if ( valuep ) {
-            *valuep = cp;
-            return NULL;
-        }
-        sv_str= Perl_newSVpvf_nocontext("%c",(int)cp);
-    } else {
-        /* fetch the charnames handler for this scope */
-        HV * const table = GvHV(PL_hintgv);
-        SV **cvp= table ? 
-            hv_fetchs(table, "charnames", FALSE) :
-            NULL;
-        SV *cv= cvp ? *cvp : NULL;
-        HE *he_str;
-        int count;
-        /* create an SV with the name as argument */
-        sv_name = newSVpvn(name, endbrace - name);
-        
-        if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
-            vFAIL2("Constant(\\N{%s}) unknown: "
-                  "(possibly a missing \"use charnames ...\")",
-                  SvPVX(sv_name));
-        }
-        if (!cvp || !SvOK(*cvp)) { /* when $^H{charnames} = undef; */
-            vFAIL2("Constant(\\N{%s}): "
-                  "$^H{charnames} is not defined",SvPVX(sv_name));
-        }
-        
-        
-        
-        if (!RExC_charnames) {
-            /* make sure our cache is allocated */
-            RExC_charnames = newHV();
-            sv_2mortal((SV*)RExC_charnames);
-        } 
-            /* see if we have looked this one up before */
-        he_str = hv_fetch_ent( RExC_charnames, sv_name, 0, 0 );
-        if ( he_str ) {
-            sv_str = HeVAL(he_str);
-            cached = 1;
-        } else {
-            dSP ;
+       char * endchar = strchr(RExC_parse, '.');
+       if (endchar) {
+           ckWARNreg(endchar, "Using just the first character returned by \\N{} in character class");
+       }
+       else endchar = endbrace;
 
 
-            ENTER ;
-            SAVETMPS ;
-            PUSHMARK(SP) ;
-            
-            XPUSHs(sv_name);
-            
-            PUTBACK ;
-            
-            count= call_sv(cv, G_SCALAR);
-            
-            if (count == 1) { /* XXXX is this right? dmq */
-                sv_str = POPs;
-                SvREFCNT_inc_simple_void(sv_str);
-            } 
-            
-            SPAGAIN ;
-            PUTBACK ;
-            FREETMPS ;
-            LEAVE ;
-            
-            if ( !sv_str || !SvOK(sv_str) ) {
-                vFAIL2("Constant(\\N{%s}): Call to &{$^H{charnames}} "
-                      "did not return a defined value",SvPVX(sv_name));
-            }
-            if (hv_store_ent( RExC_charnames, sv_name, sv_str, 0))
-                cached = 1;
-        }
+       length_of_hex = (STRLEN)(endchar - RExC_parse);
+       *valuep = grok_hex(RExC_parse, &length_of_hex, &flags, NULL);
+
+       /* The tokenizer should have guaranteed validity, but it's possible to
+        * bypass it by using single quoting, so check */
+       if ( length_of_hex != (STRLEN)(endchar - RExC_parse) ) {
+           *valuep = UNICODE_REPLACEMENT;
+       }    
+
+       RExC_parse = endbrace + 1;
+       if (endchar == endbrace) return NULL;
+
+        ret = (regnode *) &RExC_parse; /* Invalid regnode pointer */
     }
     }
-    if (valuep) {
-        char *p = SvPV(sv_str, len);
-        if (len) {
-            STRLEN numlen = 1;
-            if ( SvUTF8(sv_str) ) {
-                *valuep = utf8_to_uvchr((U8*)p, &numlen);
-                if (*valuep > 0x7F)
-                    RExC_utf8 = 1; 
-                /* XXXX
-                  We have to turn on utf8 for high bit chars otherwise
-                  we get failures with
-                  
-                   "ss" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
-                   "SS" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
-                
-                  This is different from what \x{} would do with the same
-                  codepoint, where the condition is > 0xFF.
-                  - dmq
-                */
-                
-                
-            } else {
-                *valuep = (UV)*p;
-                /* warn if we havent used the whole string? */
-            }
-            if (numlen<len && SIZE_ONLY && ckWARN(WARN_REGEXP)) {
-                vWARN2(RExC_parse,
-                    "Ignoring excess chars from \\N{%s} in character class",
-                    SvPVX(sv_name)
-                );
-            }        
-        } else if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
-            vWARN2(RExC_parse,
-                    "Ignoring zero length \\N{%s} in character class",
-                    SvPVX(sv_name)
-                );
-        }
-        if (sv_name)    
-            SvREFCNT_dec(sv_name);    
-        if (!cached)
-            SvREFCNT_dec(sv_str);    
-        return len ? NULL : (regnode *)&len;
-    } else if(SvCUR(sv_str)) {     
-        
-        char *s; 
-        char *p, *pend;        
-        STRLEN charlen = 1;
-        char * parse_start = name-3; /* needed for the offsets */
-        GET_RE_DEBUG_FLAGS_DECL;     /* needed for the offsets */
-        
-        ret = reg_node(pRExC_state,
-            (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
-        s= STRING(ret);
-        
-        if ( RExC_utf8 && !SvUTF8(sv_str) ) {
-            sv_utf8_upgrade(sv_str);
-        } else if ( !RExC_utf8 && SvUTF8(sv_str) ) {
-            RExC_utf8= 1;
-        }
-        
-        p = SvPV(sv_str, len);
-        pend = p + len;
-        /* len is the length written, charlen is the size the char read */
-        for ( len = 0; p < pend; p += charlen ) {
-            if (UTF) {
-                UV uvc = utf8_to_uvchr((U8*)p, &charlen);
-                if (FOLD) {
-                    STRLEN foldlen,numlen;
-                    U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
-                    uvc = toFOLD_uni(uvc, tmpbuf, &foldlen);
-                    /* Emit all the Unicode characters. */
-                    
-                    for (foldbuf = tmpbuf;
-                        foldlen;
-                        foldlen -= numlen) 
-                    {
-                        uvc = utf8_to_uvchr(foldbuf, &numlen);
-                        if (numlen > 0) {
-                            const STRLEN unilen = reguni(pRExC_state, uvc, s);
-                            s       += unilen;
-                            len     += unilen;
-                            /* In EBCDIC the numlen
-                            * and unilen can differ. */
-                            foldbuf += numlen;
-                            if (numlen >= foldlen)
-                                break;
-                        }
-                        else
-                            break; /* "Can't happen." */
-                    }                          
-                } else {
-                    const STRLEN unilen = reguni(pRExC_state, uvc, s);
-                   if (unilen > 0) {
-                      s   += unilen;
-                      len += unilen;
-                   }
-               }
-           } else {
-                len++;
-                REGC(*p, s++);
-            }
-        }
-        if (SIZE_ONLY) {
-            RExC_size += STR_SZ(len);
-        } else {
-            STR_LEN(ret) = len;
-            RExC_emit += STR_SZ(len);
-        }
-        Set_Node_Cur_Length(ret); /* MJD */
-        RExC_parse--; 
-        nextchar(pRExC_state);
-    } else {
-        ret = reg_node(pRExC_state,NOTHING);
-    }
-    if (!cached) {
-        SvREFCNT_dec(sv_str);
-    }
-    if (sv_name) {
-        SvREFCNT_dec(sv_name); 
+    else {     /* Not a char class */
+       char *s;            /* String to put in generated EXACT node */
+       STRLEN len = 0;     /* Its current length */
+       char *endchar;      /* Points to '.' or '}' ending cur char in the input
+                              stream */
+
+       ret = reg_node(pRExC_state,
+                       (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
+       s= STRING(ret);
+
+       /* Exact nodes can hold only a U8 length's of text = 255.  Loop through
+        * the input which is of the form now 'c1.c2.c3...}' until find the
+        * ending brace or exeed length 255.  The characters that exceed this
+        * limit are dropped.  The limit could be relaxed should it become
+        * desirable by reparsing this as (?:\N{NAME}), so could generate
+        * multiple EXACT nodes, as is done for just regular input.  But this
+        * is primarily a named character, and not intended to be a huge long
+        * string, so 255 bytes should be good enough */
+       while (1) {
+           STRLEN this_char_length;
+           I32 grok_flags = PERL_SCAN_ALLOW_UNDERSCORES
+                           | PERL_SCAN_DISALLOW_PREFIX
+                           | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
+           UV cp;  /* Ord of current character */
+
+           /* Code points are separated by dots.  If none, there is only one
+            * code point, and is terminated by the brace */
+           endchar = strchr(RExC_parse, '.');
+           if (! endchar) endchar = endbrace;
+
+           /* The values are Unicode even on EBCDIC machines */
+           this_char_length = (STRLEN)(endchar - RExC_parse);
+           cp = grok_hex(RExC_parse, &this_char_length, &grok_flags, NULL);
+           if ( this_char_length == 0 
+               || this_char_length != (STRLEN)(endchar - RExC_parse) )
+           {
+               cp = UNICODE_REPLACEMENT;   /* Substitute a valid character */
+           }    
+
+           if (! FOLD) {       /* Not folding, just append to the string */
+               STRLEN unilen;
+
+               /* Quit before adding this character if would exceed limit */
+               if (len + UNISKIP(cp) > U8_MAX) break;
+
+               unilen = reguni(pRExC_state, cp, s);
+               if (unilen > 0) {
+                   s   += unilen;
+                   len += unilen;
+               }
+           } else {    /* Folding, output the folded equivalent */
+               STRLEN foldlen,numlen;
+               U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
+               cp = toFOLD_uni(cp, tmpbuf, &foldlen);
+
+               /* Quit before exceeding size limit */
+               if (len + foldlen > U8_MAX) break;
+               
+               for (foldbuf = tmpbuf;
+                   foldlen;
+                   foldlen -= numlen) 
+               {
+                   cp = utf8_to_uvchr(foldbuf, &numlen);
+                   if (numlen > 0) {
+                       const STRLEN unilen = reguni(pRExC_state, cp, s);
+                       s       += unilen;
+                       len     += unilen;
+                       /* In EBCDIC the numlen and unilen can differ. */
+                       foldbuf += numlen;
+                       if (numlen >= foldlen)
+                           break;
+                   }
+                   else
+                       break; /* "Can't happen." */
+               }                          
+           }
+
+           /* Point to the beginning of the next character in the sequence. */
+           RExC_parse = endchar + 1;
+
+           /* Quit if no more characters */
+           if (RExC_parse >= endbrace) break;
+       }
+
+
+       if (SIZE_ONLY) {
+           if (RExC_parse < endbrace) {
+               ckWARNreg(RExC_parse - 1,
+                         "Using just the first characters returned by \\N{}");
+           }
+
+           RExC_size += STR_SZ(len);
+       } else {
+           STR_LEN(ret) = len;
+           RExC_emit += STR_SZ(len);
+       }
+
+       RExC_parse = endbrace + 1;
+
+       *flagp |= HASWIDTH; /* Not SIMPLE, as that causes the engine to fail
+                              with malformed in t/re/pat_advanced.t */
+       RExC_parse --;
+       Set_Node_Cur_Length(ret); /* MJD */
+       nextchar(pRExC_state);
     }
     }
-    return ret;
 
 
+    return ret;
 }
 
 
 }
 
 
@@ -5987,12 +6836,13 @@ STATIC UV
 S_reg_recode(pTHX_ const char value, SV **encp)
 {
     STRLEN numlen = 1;
 S_reg_recode(pTHX_ const char value, SV **encp)
 {
     STRLEN numlen = 1;
-    SV * const sv = sv_2mortal(newSVpvn(&value, numlen));
-    const char * const s = encp && *encp ? sv_recode_to_utf8(sv, *encp)
-                                        : SvPVX(sv);
+    SV * const sv = newSVpvn_flags(&value, numlen, SVs_TEMP);
+    const char * const s = *encp ? sv_recode_to_utf8(sv, *encp) : SvPVX(sv);
     const STRLEN newlen = SvCUR(sv);
     UV uv = UNICODE_REPLACEMENT;
 
     const STRLEN newlen = SvCUR(sv);
     UV uv = UNICODE_REPLACEMENT;
 
+    PERL_ARGS_ASSERT_REG_RECODE;
+
     if (newlen)
        uv = SvUTF8(sv)
             ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
     if (newlen)
        uv = SvUTF8(sv)
             ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
@@ -6000,8 +6850,7 @@ S_reg_recode(pTHX_ const char value, SV **encp)
 
     if (!newlen || numlen != newlen) {
        uv = UNICODE_REPLACEMENT;
 
     if (!newlen || numlen != newlen) {
        uv = UNICODE_REPLACEMENT;
-       if (encp)
-           *encp = NULL;
+       *encp = NULL;
     }
     return uv;
 }
     }
     return uv;
 }
@@ -6009,15 +6858,26 @@ S_reg_recode(pTHX_ const char value, SV **encp)
 
 /*
  - regatom - the lowest level
 
 /*
  - regatom - the lowest level
- *
- * Optimization:  gobbles an entire sequence of ordinary characters so that
- * it can turn them into a single node, which is smaller to store and
- * faster to run.  Backslashed characters are exceptions, each becoming a
- * separate node; the code is simpler that way and it's not worth fixing.
- *
- * [Yes, it is worth fixing, some scripts can run twice the speed.]
- * [It looks like its ok, as in S_study_chunk we merge adjacent EXACT nodes]
- */
+
+   Try to identify anything special at the start of the pattern. If there
+   is, then handle it as required. This may involve generating a single regop,
+   such as for an assertion; or it may involve recursing, such as to
+   handle a () structure.
+
+   If the string doesn't start with something special then we gobble up
+   as much literal text as we can.
+
+   Once we have been able to handle whatever type of thing started the
+   sequence, we return.
+
+   Note: we have to be careful with escapes, as they can be both literal
+   and special, and in the case of \10 and friends can either, depending
+   on context. Specifically there are two seperate switches for handling
+   escape sequences, with the one for handling literal escapes requiring
+   a dummy entry for all of the special escapes that are actually handled
+   by the other.
+*/
+
 STATIC regnode *
 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 {
 STATIC regnode *
 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
 {
@@ -6029,14 +6889,16 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
     DEBUG_PARSE("atom");
     *flagp = WORST;            /* Tentatively. */
 
     DEBUG_PARSE("atom");
     *flagp = WORST;            /* Tentatively. */
 
+    PERL_ARGS_ASSERT_REGATOM;
+
 tryagain:
 tryagain:
-    switch (*RExC_parse) {
+    switch ((U8)*RExC_parse) {
     case '^':
        RExC_seen_zerolen++;
        nextchar(pRExC_state);
     case '^':
        RExC_seen_zerolen++;
        nextchar(pRExC_state);
-       if (RExC_flags & PMf_MULTILINE)
+       if (RExC_flags & RXf_PMf_MULTILINE)
            ret = reg_node(pRExC_state, MBOL);
            ret = reg_node(pRExC_state, MBOL);
-       else if (RExC_flags & PMf_SINGLELINE)
+       else if (RExC_flags & RXf_PMf_SINGLELINE)
            ret = reg_node(pRExC_state, SBOL);
        else
            ret = reg_node(pRExC_state, BOL);
            ret = reg_node(pRExC_state, SBOL);
        else
            ret = reg_node(pRExC_state, BOL);
@@ -6046,9 +6908,9 @@ tryagain:
        nextchar(pRExC_state);
        if (*RExC_parse)
            RExC_seen_zerolen++;
        nextchar(pRExC_state);
        if (*RExC_parse)
            RExC_seen_zerolen++;
-       if (RExC_flags & PMf_MULTILINE)
+       if (RExC_flags & RXf_PMf_MULTILINE)
            ret = reg_node(pRExC_state, MEOL);
            ret = reg_node(pRExC_state, MEOL);
-       else if (RExC_flags & PMf_SINGLELINE)
+       else if (RExC_flags & RXf_PMf_SINGLELINE)
            ret = reg_node(pRExC_state, SEOL);
        else
            ret = reg_node(pRExC_state, EOL);
            ret = reg_node(pRExC_state, SEOL);
        else
            ret = reg_node(pRExC_state, EOL);
@@ -6056,7 +6918,7 @@ tryagain:
        break;
     case '.':
        nextchar(pRExC_state);
        break;
     case '.':
        nextchar(pRExC_state);
-       if (RExC_flags & PMf_SINGLELINE)
+       if (RExC_flags & RXf_PMf_SINGLELINE)
            ret = reg_node(pRExC_state, SANY);
        else
            ret = reg_node(pRExC_state, REG_ANY);
            ret = reg_node(pRExC_state, SANY);
        else
            ret = reg_node(pRExC_state, REG_ANY);
@@ -6091,7 +6953,7 @@ tryagain:
                }
                return(NULL);
        }
                }
                return(NULL);
        }
-       *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
+       *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
        break;
     case '|':
     case ')':
        break;
     case '|':
     case ')':
@@ -6114,105 +6976,147 @@ tryagain:
        RExC_parse++;
        vFAIL("Quantifier follows nothing");
        break;
        RExC_parse++;
        vFAIL("Quantifier follows nothing");
        break;
+    case 0xDF:
+    case 0xC3:
+    case 0xCE:
+        do_foldchar:
+        if (!LOC && FOLD) {
+            U32 len,cp;
+           len=0; /* silence a spurious compiler warning */
+            if ((cp = what_len_TRICKYFOLD_safe(RExC_parse,RExC_end,UTF,len))) {
+                *flagp |= HASWIDTH; /* could be SIMPLE too, but needs a handler in regexec.regrepeat */
+                RExC_parse+=len-1; /* we get one from nextchar() as well. :-( */
+                ret = reganode(pRExC_state, FOLDCHAR, cp);
+                Set_Node_Length(ret, 1); /* MJD */
+                nextchar(pRExC_state); /* kill whitespace under /x */
+                return ret;
+            }
+        }
+        goto outer_default;
     case '\\':
     case '\\':
-       switch (*++RExC_parse) {
+       /* Special Escapes
+
+          This switch handles escape sequences that resolve to some kind
+          of special regop and not to literal text. Escape sequnces that
+          resolve to literal text are handled below in the switch marked
+          "Literal Escapes".
+
+          Every entry in this switch *must* have a corresponding entry
+          in the literal escape switch. However, the opposite is not
+          required, as the default for this switch is to jump to the
+          literal text handling code.
+       */
+       switch ((U8)*++RExC_parse) {
+       case 0xDF:
+       case 0xC3:
+       case 0xCE:
+                  goto do_foldchar;        
+       /* Special Escapes */
        case 'A':
            RExC_seen_zerolen++;
            ret = reg_node(pRExC_state, SBOL);
            *flagp |= SIMPLE;
        case 'A':
            RExC_seen_zerolen++;
            ret = reg_node(pRExC_state, SBOL);
            *flagp |= SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'G':
            ret = reg_node(pRExC_state, GPOS);
            RExC_seen |= REG_SEEN_GPOS;
            *flagp |= SIMPLE;
        case 'G':
            ret = reg_node(pRExC_state, GPOS);
            RExC_seen |= REG_SEEN_GPOS;
            *flagp |= SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
+       case 'K':
+           RExC_seen_zerolen++;
+           ret = reg_node(pRExC_state, KEEPS);
+           *flagp |= SIMPLE;
+           /* XXX:dmq : disabling in-place substitution seems to
+            * be necessary here to avoid cases of memory corruption, as
+            * with: C<$_="x" x 80; s/x\K/y/> -- rgs
+            */
+           RExC_seen |= REG_SEEN_LOOKBEHIND;
+           goto finish_meta_pat;
        case 'Z':
            ret = reg_node(pRExC_state, SEOL);
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
        case 'Z':
            ret = reg_node(pRExC_state, SEOL);
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
-           nextchar(pRExC_state);
-           break;
+           goto finish_meta_pat;
        case 'z':
            ret = reg_node(pRExC_state, EOS);
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
        case 'z':
            ret = reg_node(pRExC_state, EOS);
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'C':
            ret = reg_node(pRExC_state, CANY);
            RExC_seen |= REG_SEEN_CANY;
            *flagp |= HASWIDTH|SIMPLE;
        case 'C':
            ret = reg_node(pRExC_state, CANY);
            RExC_seen |= REG_SEEN_CANY;
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'X':
            ret = reg_node(pRExC_state, CLUMP);
            *flagp |= HASWIDTH;
        case 'X':
            ret = reg_node(pRExC_state, CLUMP);
            *flagp |= HASWIDTH;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'w':
            ret = reg_node(pRExC_state, (U8)(LOC ? ALNUML     : ALNUM));
            *flagp |= HASWIDTH|SIMPLE;
        case 'w':
            ret = reg_node(pRExC_state, (U8)(LOC ? ALNUML     : ALNUM));
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'W':
            ret = reg_node(pRExC_state, (U8)(LOC ? NALNUML    : NALNUM));
            *flagp |= HASWIDTH|SIMPLE;
        case 'W':
            ret = reg_node(pRExC_state, (U8)(LOC ? NALNUML    : NALNUM));
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'b':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
            ret = reg_node(pRExC_state, (U8)(LOC ? BOUNDL     : BOUND));
            *flagp |= SIMPLE;
        case 'b':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
            ret = reg_node(pRExC_state, (U8)(LOC ? BOUNDL     : BOUND));
            *flagp |= SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'B':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
            ret = reg_node(pRExC_state, (U8)(LOC ? NBOUNDL    : NBOUND));
            *flagp |= SIMPLE;
        case 'B':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
            ret = reg_node(pRExC_state, (U8)(LOC ? NBOUNDL    : NBOUND));
            *flagp |= SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 's':
            ret = reg_node(pRExC_state, (U8)(LOC ? SPACEL     : SPACE));
            *flagp |= HASWIDTH|SIMPLE;
        case 's':
            ret = reg_node(pRExC_state, (U8)(LOC ? SPACEL     : SPACE));
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'S':
            ret = reg_node(pRExC_state, (U8)(LOC ? NSPACEL    : NSPACE));
            *flagp |= HASWIDTH|SIMPLE;
        case 'S':
            ret = reg_node(pRExC_state, (U8)(LOC ? NSPACEL    : NSPACE));
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'd':
            ret = reg_node(pRExC_state, DIGIT);
            *flagp |= HASWIDTH|SIMPLE;
        case 'd':
            ret = reg_node(pRExC_state, DIGIT);
            *flagp |= HASWIDTH|SIMPLE;
-           nextchar(pRExC_state);
-            Set_Node_Length(ret, 2); /* MJD */
-           break;
+           goto finish_meta_pat;
        case 'D':
            ret = reg_node(pRExC_state, NDIGIT);
            *flagp |= HASWIDTH|SIMPLE;
        case 'D':
            ret = reg_node(pRExC_state, NDIGIT);
            *flagp |= HASWIDTH|SIMPLE;
+           goto finish_meta_pat;
+       case 'R':
+           ret = reg_node(pRExC_state, LNBREAK);
+           *flagp |= HASWIDTH|SIMPLE;
+           goto finish_meta_pat;
+       case 'h':
+           ret = reg_node(pRExC_state, HORIZWS);
+           *flagp |= HASWIDTH|SIMPLE;
+           goto finish_meta_pat;
+       case 'H':
+           ret = reg_node(pRExC_state, NHORIZWS);
+           *flagp |= HASWIDTH|SIMPLE;
+           goto finish_meta_pat;
+       case 'v':
+           ret = reg_node(pRExC_state, VERTWS);
+           *flagp |= HASWIDTH|SIMPLE;
+           goto finish_meta_pat;
+       case 'V':
+           ret = reg_node(pRExC_state, NVERTWS);
+           *flagp |= HASWIDTH|SIMPLE;
+         finish_meta_pat:          
            nextchar(pRExC_state);
             Set_Node_Length(ret, 2); /* MJD */
            nextchar(pRExC_state);
             Set_Node_Length(ret, 2); /* MJD */
-           break;
+           break;          
        case 'p':
        case 'P':
            {   
                char* const oldregxend = RExC_end;
        case 'p':
        case 'P':
            {   
                char* const oldregxend = RExC_end;
+#ifdef DEBUGGING
                char* parse_start = RExC_parse - 2;
                char* parse_start = RExC_parse - 2;
+#endif
 
                if (RExC_parse[1] == '{') {
                  /* a lovely hack--pretend we saw [\pX] instead */
 
                if (RExC_parse[1] == '{') {
                  /* a lovely hack--pretend we saw [\pX] instead */
@@ -6244,94 +7148,98 @@ tryagain:
            }
            break;
         case 'N': 
            }
            break;
         case 'N': 
-            /* Handle \N{NAME} here and not below because it can be 
+            /* Handle \N and \N{NAME} here and not below because it can be
             multicharacter. join_exact() will join them up later on. 
             Also this makes sure that things like /\N{BLAH}+/ and 
             \N{BLAH} being multi char Just Happen. dmq*/
             ++RExC_parse;
             multicharacter. join_exact() will join them up later on. 
             Also this makes sure that things like /\N{BLAH}+/ and 
             \N{BLAH} being multi char Just Happen. dmq*/
             ++RExC_parse;
-            ret= reg_namedseq(pRExC_state, NULL); 
+            ret= reg_namedseq(pRExC_state, NULL, flagp); 
             break;
        case 'k':    /* Handle \k<NAME> and \k'NAME' */
             break;
        case 'k':    /* Handle \k<NAME> and \k'NAME' */
+       parse_named_seq:
         {   
             char ch= RExC_parse[1];        
         {   
             char ch= RExC_parse[1];        
-           if (ch != '<' && ch != '\'') {
-               if (SIZE_ONLY)
-                   vWARN( RExC_parse + 1, 
-                       "Possible broken named back reference treated as literal k");
-               parse_start--;
-               goto defchar;
+           if (ch != '<' && ch != '\'' && ch != '{') {
+               RExC_parse++;
+               vFAIL2("Sequence %.2s... not terminated",parse_start);
            } else {
            } else {
+               /* this pretty much dupes the code for (?P=...) in reg(), if
+                   you change this make sure you change that */
                char* name_start = (RExC_parse += 2);
                char* name_start = (RExC_parse += 2);
-               I32 num = 0;
+               U32 num = 0;
                 SV *sv_dat = reg_scan_name(pRExC_state,
                     SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
                 SV *sv_dat = reg_scan_name(pRExC_state,
                     SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
-                ch= (ch == '<') ? '>' : '\'';
-                    
+                ch= (ch == '<') ? '>' : (ch == '{') ? '}' : '\'';
                 if (RExC_parse == name_start || *RExC_parse != ch)
                 if (RExC_parse == name_start || *RExC_parse != ch)
-                    vFAIL2("Sequence \\k%c... not terminated",
-                        (ch == '>' ? '<' : ch));
-                
+                    vFAIL2("Sequence %.3s... not terminated",parse_start);
+
+                if (!SIZE_ONLY) {
+                    num = add_data( pRExC_state, 1, "S" );
+                    RExC_rxi->data->data[num]=(void*)sv_dat;
+                    SvREFCNT_inc_simple_void(sv_dat);
+                }
+
                 RExC_sawback = 1;
                 ret = reganode(pRExC_state,
                           (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
                           num);
                 *flagp |= HASWIDTH;
                 RExC_sawback = 1;
                 ret = reganode(pRExC_state,
                           (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
                           num);
                 *flagp |= HASWIDTH;
-                
-               
-                if (!SIZE_ONLY) {
-                    num = add_data( pRExC_state, 1, "S" );
-                    ARG_SET(ret,num);
-                    RExC_rx->data->data[num]=(void*)sv_dat;
-                    SvREFCNT_inc(sv_dat);
-                }    
+
                 /* override incorrect value set in reganode MJD */
                 Set_Node_Offset(ret, parse_start+1);
                 Set_Node_Cur_Length(ret); /* MJD */
                 nextchar(pRExC_state);
                 /* override incorrect value set in reganode MJD */
                 Set_Node_Offset(ret, parse_start+1);
                 Set_Node_Cur_Length(ret); /* MJD */
                 nextchar(pRExC_state);
-                              
+
             }
             break;
             }
             break;
-        }            
-       case 'n':
-       case 'r':
-       case 't':
-       case 'f':
-       case 'e':
-       case 'a':
-       case 'x':
-       case 'c':
-       case '0':
-           goto defchar;
-       case 'R': 
+       }
+       case 'g': 
        case '1': case '2': case '3': case '4':
        case '5': case '6': case '7': case '8': case '9':
            {
                I32 num;
        case '1': case '2': case '3': case '4':
        case '5': case '6': case '7': case '8': case '9':
            {
                I32 num;
-               bool isrel=(*RExC_parse=='R');
-               if (isrel)
+               bool isg = *RExC_parse == 'g';
+               bool isrel = 0; 
+               bool hasbrace = 0;
+               if (isg) {
                    RExC_parse++;
                    RExC_parse++;
+                   if (*RExC_parse == '{') {
+                       RExC_parse++;
+                       hasbrace = 1;
+                   }
+                   if (*RExC_parse == '-') {
+                       RExC_parse++;
+                       isrel = 1;
+                   }
+                   if (hasbrace && !isDIGIT(*RExC_parse)) {
+                       if (isrel) RExC_parse--;
+                        RExC_parse -= 2;                           
+                       goto parse_named_seq;
+               }   }
                num = atoi(RExC_parse);
                num = atoi(RExC_parse);
+               if (isg && num == 0)
+                   vFAIL("Reference to invalid group 0");
                 if (isrel) {
                 if (isrel) {
-                    num = RExC_cpar - num;
+                    num = RExC_npar - num;
                     if (num < 1)
                         vFAIL("Reference to nonexistent or unclosed group");
                 }
                     if (num < 1)
                         vFAIL("Reference to nonexistent or unclosed group");
                 }
-               if (num > 9 && num >= RExC_npar)
+               if (!isg && num > 9 && num >= RExC_npar)
                    goto defchar;
                else {
                    char * const parse_start = RExC_parse - 1; /* MJD */
                    while (isDIGIT(*RExC_parse))
                        RExC_parse++;
                    goto defchar;
                else {
                    char * const parse_start = RExC_parse - 1; /* MJD */
                    while (isDIGIT(*RExC_parse))
                        RExC_parse++;
-
+                   if (parse_start == RExC_parse - 1) 
+                       vFAIL("Unterminated \\g... pattern");
+                    if (hasbrace) {
+                        if (*RExC_parse != '}') 
+                            vFAIL("Unterminated \\g{...} pattern");
+                        RExC_parse++;
+                    }    
                    if (!SIZE_ONLY) {
                        if (num > (I32)RExC_rx->nparens)
                            vFAIL("Reference to nonexistent group");
                    if (!SIZE_ONLY) {
                        if (num > (I32)RExC_rx->nparens)
                            vFAIL("Reference to nonexistent group");
-                       /* People make this error all the time apparently.
-                          So we cant fail on it, even though we should 
-                       
-                       else if (num >= RExC_cpar)
-                           vFAIL("Reference to unclosed group will always match");
-                       */
                    }
                    RExC_sawback = 1;
                    ret = reganode(pRExC_state,
                    }
                    RExC_sawback = 1;
                    ret = reganode(pRExC_state,
@@ -6360,15 +7268,14 @@ tryagain:
        break;
 
     case '#':
        break;
 
     case '#':
-       if (RExC_flags & PMf_EXTENDED) {
-           while (RExC_parse < RExC_end && *RExC_parse != '\n')
-               RExC_parse++;
-           if (RExC_parse < RExC_end)
+       if (RExC_flags & RXf_PMf_EXTENDED) {
+           if ( reg_skipcomment( pRExC_state ) )
                goto tryagain;
        }
        /* FALL THROUGH */
 
                goto tryagain;
        }
        /* FALL THROUGH */
 
-    default: {
+    default:
+        outer_default:{
            register STRLEN len;
            register UV ender;
            register char *p;
            register STRLEN len;
            register UV ender;
            register char *p;
@@ -6391,9 +7298,14 @@ tryagain:
            {
                char * const oldp = p;
 
            {
                char * const oldp = p;
 
-               if (RExC_flags & PMf_EXTENDED)
-                   p = regwhite(p, RExC_end);
-               switch (*p) {
+               if (RExC_flags & RXf_PMf_EXTENDED)
+                   p = regwhite( pRExC_state, p );
+               switch ((U8)*p) {
+               case 0xDF:
+               case 0xC3:
+               case 0xCE:
+                          if (LOC || !FOLD || !is_TRICKYFOLD_safe(p,RExC_end,UTF))
+                               goto normal_default;
                case '^':
                case '$':
                case '.':
                case '^':
                case '$':
                case '.':
@@ -6403,27 +7315,47 @@ tryagain:
                case '|':
                    goto loopdone;
                case '\\':
                case '|':
                    goto loopdone;
                case '\\':
-                   switch (*++p) {
-                   case 'A':
-                   case 'C':
-                   case 'X':
-                   case 'G':
-                   case 'Z':
-                   case 'z':
-                   case 'w':
-                   case 'W':
-                   case 'b':
-                   case 'B':
-                   case 's':
-                   case 'S':
-                   case 'd':
-                   case 'D':
-                   case 'p':
-                   case 'P':
-                    case 'N':
-                    case 'R':
+                   /* Literal Escapes Switch
+
+                      This switch is meant to handle escape sequences that
+                      resolve to a literal character.
+
+                      Every escape sequence that represents something
+                      else, like an assertion or a char class, is handled
+                      in the switch marked 'Special Escapes' above in this
+                      routine, but also has an entry here as anything that
+                      isn't explicitly mentioned here will be treated as
+                      an unescaped equivalent literal.
+                   */
+
+                   switch ((U8)*++p) {
+                   /* These are all the special escapes. */
+                   case 0xDF:
+                   case 0xC3:
+                   case 0xCE:
+                          if (LOC || !FOLD || !is_TRICKYFOLD_safe(p,RExC_end,UTF))
+                               goto normal_default;                
+                   case 'A':             /* Start assertion */
+                   case 'b': case 'B':   /* Word-boundary assertion*/
+                   case 'C':             /* Single char !DANGEROUS! */
+                   case 'd': case 'D':   /* digit class */
+                   case 'g': case 'G':   /* generic-backref, pos assertion */
+                   case 'h': case 'H':   /* HORIZWS */
+                   case 'k': case 'K':   /* named backref, keep marker */
+                   case 'N':             /* named char sequence */
+                   case 'p': case 'P':   /* Unicode property */
+                             case 'R':   /* LNBREAK */
+                   case 's': case 'S':   /* space class */
+                   case 'v': case 'V':   /* VERTWS */
+                   case 'w': case 'W':   /* word class */
+                   case 'X':             /* eXtended Unicode "combining character sequence" */
+                   case 'z': case 'Z':   /* End of line/string assertion */
                        --p;
                        goto loopdone;
                        --p;
                        goto loopdone;
+
+                   /* Anything after here is an escape that resolves to a
+                      literal. (Except digits, which may or may not)
+                    */
                    case 'n':
                        ender = '\n';
                        p++;
                    case 'n':
                        ender = '\n';
                        p++;
@@ -6487,6 +7419,18 @@ tryagain:
                             I32 flags = 0;
                            STRLEN numlen = 3;
                            ender = grok_oct(p, &numlen, &flags, NULL);
                             I32 flags = 0;
                            STRLEN numlen = 3;
                            ender = grok_oct(p, &numlen, &flags, NULL);
+
+                           /* An octal above 0xff is interpreted differently
+                            * depending on if the re is in utf8 or not.  If it
+                            * is in utf8, the value will be itself, otherwise
+                            * it is interpreted as modulo 0x100.  It has been
+                            * decided to discourage the use of octal above the
+                            * single-byte range.  For now, warn only when
+                            * it ends up modulo */
+                           if (SIZE_ONLY && ender >= 0x100
+                                   && ! UTF && ! PL_encoding) {
+                               ckWARNregdep(p, "Use of octal value above 377 is deprecated");
+                           }
                            p += numlen;
                        }
                        else {
                            p += numlen;
                        }
                        else {
@@ -6500,8 +7444,8 @@ tryagain:
                        {
                            SV* enc = PL_encoding;
                            ender = reg_recode((const char)(U8)ender, &enc);
                        {
                            SV* enc = PL_encoding;
                            ender = reg_recode((const char)(U8)ender, &enc);
-                           if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
-                               vWARN(p, "Invalid escape in the specified encoding");
+                           if (!enc && SIZE_ONLY)
+                               ckWARNreg(p, "Invalid escape in the specified encoding");
                            RExC_utf8 = 1;
                        }
                        break;
                            RExC_utf8 = 1;
                        }
                        break;
@@ -6510,8 +7454,8 @@ tryagain:
                            FAIL("Trailing \\");
                        /* FALL THROUGH */
                    default:
                            FAIL("Trailing \\");
                        /* FALL THROUGH */
                    default:
-                       if (!SIZE_ONLY&& isALPHA(*p) && ckWARN(WARN_REGEXP))
-                           vWARN2(p + 1, "Unrecognized escape \\%c passed through", UCHARAT(p));
+                       if (!SIZE_ONLY&& isALPHA(*p))
+                           ckWARN2reg(p + 1, "Unrecognized escape \\%c passed through", UCHARAT(p));
                        goto normal_default;
                    }
                    break;
                        goto normal_default;
                    }
                    break;
@@ -6527,13 +7471,13 @@ tryagain:
                        ender = *p++;
                    break;
                }
                        ender = *p++;
                    break;
                }
-               if (RExC_flags & PMf_EXTENDED)
-                   p = regwhite(p, RExC_end);
+               if ( RExC_flags & RXf_PMf_EXTENDED)
+                   p = regwhite( pRExC_state, p );
                if (UTF && FOLD) {
                    /* Prime the casefolded buffer. */
                    ender = toFOLD_uni(ender, tmpbuf, &foldlen);
                }
                if (UTF && FOLD) {
                    /* Prime the casefolded buffer. */
                    ender = toFOLD_uni(ender, tmpbuf, &foldlen);
                }
-               if (ISMULT2(p)) { /* Back off on ?+*. */
+               if (p < RExC_end && ISMULT2(p)) { /* Back off on ?+*. */
                    if (len)
                        p = oldp;
                    else if (UTF) {
                    if (len)
                        p = oldp;
                    else if (UTF) {
@@ -6635,15 +7579,25 @@ tryagain:
 }
 
 STATIC char *
 }
 
 STATIC char *
-S_regwhite(char *p, const char *e)
+S_regwhite( RExC_state_t *pRExC_state, char *p )
 {
 {
+    const char *e = RExC_end;
+
+    PERL_ARGS_ASSERT_REGWHITE;
+
     while (p < e) {
        if (isSPACE(*p))
            ++p;
        else if (*p == '#') {
     while (p < e) {
        if (isSPACE(*p))
            ++p;
        else if (*p == '#') {
+            bool ended = 0;
            do {
            do {
-               p++;
-           } while (p < e && *p != '\n');
+               if (*p++ == '\n') {
+                   ended = 1;
+                   break;
+               }
+           } while (p < e);
+           if (!ended)
+               RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
        }
        else
            break;
        }
        else
            break;
@@ -6667,6 +7621,8 @@ S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
     dVAR;
     I32 namedclass = OOB_NAMEDCLASS;
 
     dVAR;
     I32 namedclass = OOB_NAMEDCLASS;
 
+    PERL_ARGS_ASSERT_REGPPOSIXCC;
+
     if (value == '[' && RExC_parse + 1 < RExC_end &&
        /* I smell either [: or [= or [. -- POSIX has been here, right? */
        POSIXCC(UCHARAT(RExC_parse))) {
     if (value == '[' && RExC_parse + 1 < RExC_end &&
        /* I smell either [: or [= or [. -- POSIX has been here, right? */
        POSIXCC(UCHARAT(RExC_parse))) {
@@ -6781,6 +7737,9 @@ STATIC void
 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
 {
     dVAR;
 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
 {
     dVAR;
+
+    PERL_ARGS_ASSERT_CHECKPOSIXCC;
+
     if (POSIXCC(UCHARAT(RExC_parse))) {
        const char *s = RExC_parse;
        const char  c = *s++;
     if (POSIXCC(UCHARAT(RExC_parse))) {
        const char *s = RExC_parse;
        const char  c = *s++;
@@ -6788,10 +7747,9 @@ S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
        while (isALNUM(*s))
            s++;
        if (*s && c == *s && s[1] == ']') {
        while (isALNUM(*s))
            s++;
        if (*s && c == *s && s[1] == ']') {
-           if (ckWARN(WARN_REGEXP))
-               vWARN3(s+2,
-                       "POSIX syntax [%c %c] belongs inside character classes",
-                       c, c);
+           ckWARN3reg(s+2,
+                      "POSIX syntax [%c %c] belongs inside character classes",
+                      c, c);
 
            /* [[=foo=]] and [[.foo.]] are still future. */
            if (POSIXCC_NOTYET(c)) {
 
            /* [[=foo=]] and [[.foo.]] are still future. */
            if (POSIXCC_NOTYET(c)) {
@@ -6806,19 +7764,77 @@ S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
 }
 
 
 }
 
 
+#define _C_C_T_(NAME,TEST,WORD)                         \
+ANYOF_##NAME:                                           \
+    if (LOC)                                            \
+       ANYOF_CLASS_SET(ret, ANYOF_##NAME);             \
+    else {                                              \
+       for (value = 0; value < 256; value++)           \
+           if (TEST)                                   \
+               ANYOF_BITMAP_SET(ret, value);           \
+    }                                                   \
+    yesno = '+';                                        \
+    what = WORD;                                        \
+    break;                                              \
+case ANYOF_N##NAME:                                     \
+    if (LOC)                                            \
+       ANYOF_CLASS_SET(ret, ANYOF_N##NAME);            \
+    else {                                              \
+       for (value = 0; value < 256; value++)           \
+           if (!TEST)                                  \
+               ANYOF_BITMAP_SET(ret, value);           \
+    }                                                   \
+    yesno = '!';                                        \
+    what = WORD;                                        \
+    break
+
+#define _C_C_T_NOLOC_(NAME,TEST,WORD)                   \
+ANYOF_##NAME:                                           \
+       for (value = 0; value < 256; value++)           \
+           if (TEST)                                   \
+               ANYOF_BITMAP_SET(ret, value);           \
+    yesno = '+';                                        \
+    what = WORD;                                        \
+    break;                                              \
+case ANYOF_N##NAME:                                     \
+       for (value = 0; value < 256; value++)           \
+           if (!TEST)                                  \
+               ANYOF_BITMAP_SET(ret, value);           \
+    yesno = '!';                                        \
+    what = WORD;                                        \
+    break
+
+/* 
+   We dont use PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS as the direct test
+   so that it is possible to override the option here without having to 
+   rebuild the entire core. as we are required to do if we change regcomp.h
+   which is where PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS is defined.
+*/
+#if PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS
+#define BROKEN_UNICODE_CHARCLASS_MAPPINGS
+#endif
+
+#ifdef BROKEN_UNICODE_CHARCLASS_MAPPINGS
+#define POSIX_CC_UNI_NAME(CCNAME) CCNAME
+#else
+#define POSIX_CC_UNI_NAME(CCNAME) "Posix" CCNAME
+#endif
+
 /*
    parse a class specification and produce either an ANYOF node that
 /*
    parse a class specification and produce either an ANYOF node that
-   matches the pattern. If the pattern matches a single char only and
-   that char is < 256 then we produce an EXACT node instead.
+   matches the pattern or if the pattern matches a single char only and
+   that char is < 256 and we are case insensitive then we produce an 
+   EXACT node instead.
 */
 */
+
 STATIC regnode *
 S_regclass(pTHX_ RExC_state_t *pRExC_state, U32 depth)
 {
     dVAR;
 STATIC regnode *
 S_regclass(pTHX_ RExC_state_t *pRExC_state, U32 depth)
 {
     dVAR;
-    register UV value = 0;
     register UV nextvalue;
     register IV prevvalue = OOB_UNICODE;
     register IV range = 0;
     register UV nextvalue;
     register IV prevvalue = OOB_UNICODE;
     register IV range = 0;
+    UV value = 0; /* XXX:dmq: needs to be referenceable (unfortunately) */
     register regnode *ret;
     STRLEN numlen;
     IV namedclass;
     register regnode *ret;
     STRLEN numlen;
     IV namedclass;
@@ -6837,6 +7853,8 @@ S_regclass(pTHX_ RExC_state_t *pRExC_state, U32 depth)
         case we need to change the emitted regop to an EXACT. */
     const char * orig_parse = RExC_parse;
     GET_RE_DEBUG_FLAGS_DECL;
         case we need to change the emitted regop to an EXACT. */
     const char * orig_parse = RExC_parse;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REGCLASS;
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
@@ -6921,6 +7939,10 @@ parseit:
            case 'S':   namedclass = ANYOF_NSPACE;      break;
            case 'd':   namedclass = ANYOF_DIGIT;       break;
            case 'D':   namedclass = ANYOF_NDIGIT;      break;
            case 'S':   namedclass = ANYOF_NSPACE;      break;
            case 'd':   namedclass = ANYOF_DIGIT;       break;
            case 'D':   namedclass = ANYOF_NDIGIT;      break;
+           case 'v':   namedclass = ANYOF_VERTWS;      break;
+           case 'V':   namedclass = ANYOF_NVERTWS;     break;
+           case 'h':   namedclass = ANYOF_HORIZWS;     break;
+           case 'H':   namedclass = ANYOF_NHORIZWS;    break;
             case 'N':  /* Handle \N{NAME} in class */
                 {
                     /* We only pay attention to the first char of 
             case 'N':  /* Handle \N{NAME} in class */
                 {
                     /* We only pay attention to the first char of 
@@ -6929,7 +7951,7 @@ parseit:
                     from earlier versions, OTOH that behaviour was broken
                     as well. */
                     UV v; /* value is register so we cant & it /grrr */
                     from earlier versions, OTOH that behaviour was broken
                     as well. */
                     UV v; /* value is register so we cant & it /grrr */
-                    if (reg_namedseq(pRExC_state, &v)) {
+                    if (reg_namedseq(pRExC_state, &v, NULL)) {
                         goto parseit;
                     }
                     value= v; 
                         goto parseit;
                     }
                     value= v; 
@@ -7023,16 +8045,16 @@ parseit:
                {
                    SV* enc = PL_encoding;
                    value = reg_recode((const char)(U8)value, &enc);
                {
                    SV* enc = PL_encoding;
                    value = reg_recode((const char)(U8)value, &enc);
-                   if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
-                       vWARN(RExC_parse,
-                             "Invalid escape in the specified encoding");
+                   if (!enc && SIZE_ONLY)
+                       ckWARNreg(RExC_parse,
+                                 "Invalid escape in the specified encoding");
                    break;
                }
            default:
                    break;
                }
            default:
-               if (!SIZE_ONLY && isALPHA(value) && ckWARN(WARN_REGEXP))
-                   vWARN2(RExC_parse,
-                          "Unrecognized escape \\%c in character class passed through",
-                          (int)value);
+               if (!SIZE_ONLY && isALPHA(value))
+                   ckWARN2reg(RExC_parse,
+                              "Unrecognized escape \\%c in character class passed through",
+                              (int)value);
                break;
            }
        } /* end of \blah */
                break;
            }
        } /* end of \blah */
@@ -7051,14 +8073,13 @@ parseit:
            /* a bad range like a-\d, a-[:digit:] ? */
            if (range) {
                if (!SIZE_ONLY) {
            /* a bad range like a-\d, a-[:digit:] ? */
            if (range) {
                if (!SIZE_ONLY) {
-                   if (ckWARN(WARN_REGEXP)) {
-                       const int w =
-                           RExC_parse >= rangebegin ?
-                           RExC_parse - rangebegin : 0;
-                       vWARN4(RExC_parse,
+                   const int w =
+                       RExC_parse >= rangebegin ?
+                       RExC_parse - rangebegin : 0;
+                   ckWARN4reg(RExC_parse,
                               "False [] range \"%*.*s\"",
                               w, w, rangebegin);
                               "False [] range \"%*.*s\"",
                               w, w, rangebegin);
-                   }
+
                    if (prevvalue < 256) {
                        ANYOF_BITMAP_SET(ret, prevvalue);
                        ANYOF_BITMAP_SET(ret, '-');
                    if (prevvalue < 256) {
                        ANYOF_BITMAP_SET(ret, prevvalue);
                        ANYOF_BITMAP_SET(ret, '-');
@@ -7073,6 +8094,8 @@ parseit:
                range = 0; /* this was not a true range */
            }
 
                range = 0; /* this was not a true range */
            }
 
+
+    
            if (!SIZE_ONLY) {
                const char *what = NULL;
                char yesno = 0;
            if (!SIZE_ONLY) {
                const char *what = NULL;
                char yesno = 0;
@@ -7084,72 +8107,27 @@ parseit:
                 * A similar issue a little earlier when switching on value.
                 * --jhi */
                switch ((I32)namedclass) {
                 * A similar issue a little earlier when switching on value.
                 * --jhi */
                switch ((I32)namedclass) {
-               case ANYOF_ALNUM:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isALNUM(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Word";      
-                   break;
-               case ANYOF_NALNUM:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isALNUM(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Word";
-                   break;
-               case ANYOF_ALNUMC:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isALNUMC(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Alnum";
-                   break;
-               case ANYOF_NALNUMC:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isALNUMC(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Alnum";
-                   break;
-               case ANYOF_ALPHA:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isALPHA(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Alpha";
-                   break;
-               case ANYOF_NALPHA:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isALPHA(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Alpha";
-                   break;
+               
+               case _C_C_T_(ALNUMC, isALNUMC(value), POSIX_CC_UNI_NAME("Alnum"));
+               case _C_C_T_(ALPHA, isALPHA(value), POSIX_CC_UNI_NAME("Alpha"));
+               case _C_C_T_(BLANK, isBLANK(value), POSIX_CC_UNI_NAME("Blank"));
+               case _C_C_T_(CNTRL, isCNTRL(value), POSIX_CC_UNI_NAME("Cntrl"));
+               case _C_C_T_(GRAPH, isGRAPH(value), POSIX_CC_UNI_NAME("Graph"));
+               case _C_C_T_(LOWER, isLOWER(value), POSIX_CC_UNI_NAME("Lower"));
+               case _C_C_T_(PRINT, isPRINT(value), POSIX_CC_UNI_NAME("Print"));
+               case _C_C_T_(PSXSPC, isPSXSPC(value), POSIX_CC_UNI_NAME("Space"));
+               case _C_C_T_(PUNCT, isPUNCT(value), POSIX_CC_UNI_NAME("Punct"));
+               case _C_C_T_(UPPER, isUPPER(value), POSIX_CC_UNI_NAME("Upper"));
+#ifdef BROKEN_UNICODE_CHARCLASS_MAPPINGS
+               case _C_C_T_(ALNUM, isALNUM(value), "Word");
+               case _C_C_T_(SPACE, isSPACE(value), "SpacePerl");
+#else
+               case _C_C_T_(SPACE, isSPACE(value), "PerlSpace");
+               case _C_C_T_(ALNUM, isALNUM(value), "PerlWord");
+#endif         
+               case _C_C_T_(XDIGIT, isXDIGIT(value), "XDigit");
+               case _C_C_T_NOLOC_(VERTWS, is_VERTWS_latin1(&value), "VertSpace");
+               case _C_C_T_NOLOC_(HORIZWS, is_HORIZWS_latin1(&value), "HorizSpace");
                case ANYOF_ASCII:
                    if (LOC)
                        ANYOF_CLASS_SET(ret, ANYOF_ASCII);
                case ANYOF_ASCII:
                    if (LOC)
                        ANYOF_CLASS_SET(ret, ANYOF_ASCII);
@@ -7183,251 +8161,31 @@ parseit:
                    }
                    yesno = '!';
                    what = "ASCII";
                    }
                    yesno = '!';
                    what = "ASCII";
-                   break;
-               case ANYOF_BLANK:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_BLANK);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isBLANK(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Blank";
-                   break;
-               case ANYOF_NBLANK:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isBLANK(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Blank";
-                   break;
-               case ANYOF_CNTRL:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isCNTRL(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Cntrl";
-                   break;
-               case ANYOF_NCNTRL:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isCNTRL(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Cntrl";
-                   break;
+                   break;              
                case ANYOF_DIGIT:
                    if (LOC)
                        ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
                    else {
                case ANYOF_DIGIT:
                    if (LOC)
                        ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
                    else {
-                       /* consecutive digits assumed */
-                       for (value = '0'; value <= '9'; value++)
-                           ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Digit";
-                   break;
-               case ANYOF_NDIGIT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
-                   else {
-                       /* consecutive digits assumed */
-                       for (value = 0; value < '0'; value++)
-                           ANYOF_BITMAP_SET(ret, value);
-                       for (value = '9' + 1; value < 256; value++)
-                           ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Digit";
-                   break;
-               case ANYOF_GRAPH:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isGRAPH(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Graph";
-                   break;
-               case ANYOF_NGRAPH:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isGRAPH(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Graph";
-                   break;
-               case ANYOF_LOWER:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_LOWER);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isLOWER(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Lower";
-                   break;
-               case ANYOF_NLOWER:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isLOWER(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Lower";
-                   break;
-               case ANYOF_PRINT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_PRINT);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isPRINT(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Print";
-                   break;
-               case ANYOF_NPRINT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isPRINT(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Print";
-                   break;
-               case ANYOF_PSXSPC:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isPSXSPC(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Space";
-                   break;
-               case ANYOF_NPSXSPC:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isPSXSPC(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Space";
-                   break;
-               case ANYOF_PUNCT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isPUNCT(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Punct";
-                   break;
-               case ANYOF_NPUNCT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isPUNCT(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Punct";
-                   break;
-               case ANYOF_SPACE:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_SPACE);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isSPACE(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "SpacePerl";
-                   break;
-               case ANYOF_NSPACE:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isSPACE(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "SpacePerl";
-                   break;
-               case ANYOF_UPPER:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_UPPER);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isUPPER(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '+';
-                   what = "Upper";
-                   break;
-               case ANYOF_NUPPER:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (!isUPPER(value))
-                               ANYOF_BITMAP_SET(ret, value);
-                   }
-                   yesno = '!';
-                   what = "Upper";
-                   break;
-               case ANYOF_XDIGIT:
-                   if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
-                   else {
-                       for (value = 0; value < 256; value++)
-                           if (isXDIGIT(value))
-                               ANYOF_BITMAP_SET(ret, value);
+                       /* consecutive digits assumed */
+                       for (value = '0'; value <= '9'; value++)
+                           ANYOF_BITMAP_SET(ret, value);
                    }
                    yesno = '+';
                    }
                    yesno = '+';
-                   what = "XDigit";
+                   what = POSIX_CC_UNI_NAME("Digit");
                    break;
                    break;
-               case ANYOF_NXDIGIT:
+               case ANYOF_NDIGIT:
                    if (LOC)
                    if (LOC)
-                       ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
+                       ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
                    else {
                    else {
-                       for (value = 0; value < 256; value++)
-                           if (!isXDIGIT(value))
-                               ANYOF_BITMAP_SET(ret, value);
+                       /* consecutive digits assumed */
+                       for (value = 0; value < '0'; value++)
+                           ANYOF_BITMAP_SET(ret, value);
+                       for (value = '9' + 1; value < 256; value++)
+                           ANYOF_BITMAP_SET(ret, value);
                    }
                    yesno = '!';
                    }
                    yesno = '!';
-                   what = "XDigit";
-                   break;
+                   what = POSIX_CC_UNI_NAME("Digit");
+                   break;              
                case ANYOF_MAX:
                    /* this is to handle \p and \P */
                    break;
                case ANYOF_MAX:
                    /* this is to handle \p and \P */
                    break;
@@ -7491,12 +8249,16 @@ parseit:
                {
                    if (isLOWER(prevvalue)) {
                        for (i = prevvalue; i <= ceilvalue; i++)
                {
                    if (isLOWER(prevvalue)) {
                        for (i = prevvalue; i <= ceilvalue; i++)
-                           if (isLOWER(i))
+                           if (isLOWER(i) && !ANYOF_BITMAP_TEST(ret,i)) {
+                               stored++;
                                ANYOF_BITMAP_SET(ret, i);
                                ANYOF_BITMAP_SET(ret, i);
+                           }
                    } else {
                        for (i = prevvalue; i <= ceilvalue; i++)
                    } else {
                        for (i = prevvalue; i <= ceilvalue; i++)
-                           if (isUPPER(i))
+                           if (isUPPER(i) && !ANYOF_BITMAP_TEST(ret,i)) {
+                               stored++;
                                ANYOF_BITMAP_SET(ret, i);
                                ANYOF_BITMAP_SET(ret, i);
+                           }
                    }
                }
                else
                    }
                }
                else
@@ -7561,8 +8323,8 @@ parseit:
 
                                  if (!unicode_alternate)
                                      unicode_alternate = newAV();
 
                                  if (!unicode_alternate)
                                      unicode_alternate = newAV();
-                                 sv = newSVpvn((char*)foldbuf, foldlen);
-                                 SvUTF8_on(sv);
+                                 sv = newSVpvn_utf8((char*)foldbuf, foldlen,
+                                                    TRUE);
                                  av_push(unicode_alternate, sv);
                              }
                         }
                                  av_push(unicode_alternate, sv);
                              }
                         }
@@ -7608,7 +8370,7 @@ parseit:
         return ret;
     /****** !SIZE_ONLY AFTER HERE *********/
 
         return ret;
     /****** !SIZE_ONLY AFTER HERE *********/
 
-    if( stored == 1 && value < 256
+    if( stored == 1 && (value < 128 || (value < 256 && !UTF))
         && !( ANYOF_FLAGS(ret) & ( ANYOF_FLAGS_ALL ^ ANYOF_FOLD ) )
     ) {
         /* optimize single char class to an EXACT node
         && !( ANYOF_FLAGS(ret) & ( ANYOF_FLAGS_ALL ^ ANYOF_FOLD ) )
     ) {
         /* optimize single char class to an EXACT node
@@ -7622,6 +8384,7 @@ parseit:
         *STRING(ret)= (char)value;
         STR_LEN(ret)= 1;
         RExC_emit += STR_SZ(1);
         *STRING(ret)= (char)value;
         STR_LEN(ret)= 1;
         RExC_emit += STR_SZ(1);
+       SvREFCNT_dec(listsv);
         return ret;
     }
     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
         return ret;
     }
     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
@@ -7658,20 +8421,70 @@ parseit:
         * used later (regexec.c:S_reginclass()). */
        av_store(av, 0, listsv);
        av_store(av, 1, NULL);
         * used later (regexec.c:S_reginclass()). */
        av_store(av, 0, listsv);
        av_store(av, 1, NULL);
-       av_store(av, 2, (SV*)unicode_alternate);
-       rv = newRV_noinc((SV*)av);
+       av_store(av, 2, MUTABLE_SV(unicode_alternate));
+       rv = newRV_noinc(MUTABLE_SV(av));
        n = add_data(pRExC_state, 1, "s");
        n = add_data(pRExC_state, 1, "s");
-       RExC_rx->data->data[n] = (void*)rv;
+       RExC_rxi->data->data[n] = (void*)rv;
        ARG_SET(ret, n);
     }
     return ret;
 }
        ARG_SET(ret, n);
     }
     return ret;
 }
+#undef _C_C_T_
+
+
+/* reg_skipcomment()
+
+   Absorbs an /x style # comments from the input stream.
+   Returns true if there is more text remaining in the stream.
+   Will set the REG_SEEN_RUN_ON_COMMENT flag if the comment
+   terminates the pattern without including a newline.
+
+   Note its the callers responsibility to ensure that we are
+   actually in /x mode
+
+*/
+
+STATIC bool
+S_reg_skipcomment(pTHX_ RExC_state_t *pRExC_state)
+{
+    bool ended = 0;
+
+    PERL_ARGS_ASSERT_REG_SKIPCOMMENT;
+
+    while (RExC_parse < RExC_end)
+        if (*RExC_parse++ == '\n') {
+            ended = 1;
+            break;
+        }
+    if (!ended) {
+        /* we ran off the end of the pattern without ending
+           the comment, so we have to add an \n when wrapping */
+        RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
+        return 0;
+    } else
+        return 1;
+}
+
+/* nextchar()
+
+   Advance that parse position, and optionall absorbs
+   "whitespace" from the inputstream.
+
+   Without /x "whitespace" means (?#...) style comments only,
+   with /x this means (?#...) and # comments and whitespace proper.
+
+   Returns the RExC_parse point from BEFORE the scan occurs.
+
+   This is the /x friendly way of saying RExC_parse++.
+*/
 
 STATIC char*
 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
 {
     char* const retval = RExC_parse++;
 
 
 STATIC char*
 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
 {
     char* const retval = RExC_parse++;
 
+    PERL_ARGS_ASSERT_NEXTCHAR;
+
     for (;;) {
        if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
                RExC_parse[2] == '#') {
     for (;;) {
        if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
                RExC_parse[2] == '#') {
@@ -7683,15 +8496,14 @@ S_nextchar(pTHX_ RExC_state_t *pRExC_state)
            RExC_parse++;
            continue;
        }
            RExC_parse++;
            continue;
        }
-       if (RExC_flags & PMf_EXTENDED) {
+       if (RExC_flags & RXf_PMf_EXTENDED) {
            if (isSPACE(*RExC_parse)) {
                RExC_parse++;
                continue;
            }
            else if (*RExC_parse == '#') {
            if (isSPACE(*RExC_parse)) {
                RExC_parse++;
                continue;
            }
            else if (*RExC_parse == '#') {
-               while (RExC_parse < RExC_end)
-                   if (*RExC_parse++ == '\n') break;
-               continue;
+               if ( reg_skipcomment( pRExC_state ) )
+                   continue;
            }
        }
        return retval;
            }
        }
        return retval;
@@ -7709,23 +8521,24 @@ S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
     regnode * const ret = RExC_emit;
     GET_RE_DEBUG_FLAGS_DECL;
 
     regnode * const ret = RExC_emit;
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_REG_NODE;
+
     if (SIZE_ONLY) {
        SIZE_ALIGN(RExC_size);
        RExC_size += 1;
        return(ret);
     }
     if (SIZE_ONLY) {
        SIZE_ALIGN(RExC_size);
        RExC_size += 1;
        return(ret);
     }
-#ifdef DEBUGGING
-    if (OP(RExC_emit) == 255)
-        Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %s: %d ",
-            reg_name[op], OP(RExC_emit));
-#endif  
+    if (RExC_emit >= RExC_emit_bound)
+        Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
+
     NODE_ALIGN_FILL(ret);
     ptr = ret;
     FILL_ADVANCE_NODE(ptr, op);
     NODE_ALIGN_FILL(ret);
     ptr = ret;
     FILL_ADVANCE_NODE(ptr, op);
+#ifdef RE_TRACK_PATTERN_OFFSETS
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n", 
               "reg_node", __LINE__, 
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n", 
               "reg_node", __LINE__, 
-              reg_name[op],
+              PL_reg_name[op],
               (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] 
                ? "Overwriting end of array!\n" : "OK",
               (UV)(RExC_emit - RExC_emit_start),
               (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] 
                ? "Overwriting end of array!\n" : "OK",
               (UV)(RExC_emit - RExC_emit_start),
@@ -7733,7 +8546,7 @@ S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
               (UV)RExC_offsets[0])); 
        Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
     }
               (UV)RExC_offsets[0])); 
        Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
     }
-
+#endif
     RExC_emit = ptr;
     return(ret);
 }
     RExC_emit = ptr;
     return(ret);
 }
@@ -7749,6 +8562,8 @@ S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
     regnode * const ret = RExC_emit;
     GET_RE_DEBUG_FLAGS_DECL;
 
     regnode * const ret = RExC_emit;
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_REGANODE;
+
     if (SIZE_ONLY) {
        SIZE_ALIGN(RExC_size);
        RExC_size += 2;
     if (SIZE_ONLY) {
        SIZE_ALIGN(RExC_size);
        RExC_size += 2;
@@ -7768,18 +8583,18 @@ S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
        */
        return(ret);
     }
        */
        return(ret);
     }
-#ifdef DEBUGGING
-    if (OP(RExC_emit) == 255)
-        Perl_croak(aTHX_ "panic: reganode overwriting end of allocated program space");
-#endif 
+    if (RExC_emit >= RExC_emit_bound)
+        Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
+
     NODE_ALIGN_FILL(ret);
     ptr = ret;
     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
     NODE_ALIGN_FILL(ret);
     ptr = ret;
     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
+#ifdef RE_TRACK_PATTERN_OFFSETS
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n", 
               "reganode",
              __LINE__,
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n", 
               "reganode",
              __LINE__,
-             reg_name[op],
+             PL_reg_name[op],
               (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ? 
               "Overwriting end of array!\n" : "OK",
               (UV)(RExC_emit - RExC_emit_start),
               (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ? 
               "Overwriting end of array!\n" : "OK",
               (UV)(RExC_emit - RExC_emit_start),
@@ -7787,7 +8602,7 @@ S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
               (UV)RExC_offsets[0])); 
        Set_Cur_Node_Offset;
     }
               (UV)RExC_offsets[0])); 
        Set_Cur_Node_Offset;
     }
-            
+#endif            
     RExC_emit = ptr;
     return(ret);
 }
     RExC_emit = ptr;
     return(ret);
 }
@@ -7799,6 +8614,9 @@ STATIC STRLEN
 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
 {
     dVAR;
 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
 {
     dVAR;
+
+    PERL_ARGS_ASSERT_REGUNI;
+
     return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
 }
 
     return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
 }
 
@@ -7817,8 +8635,11 @@ S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
     const int offset = regarglen[(U8)op];
     const int size = NODE_STEP_REGNODE + offset;
     GET_RE_DEBUG_FLAGS_DECL;
     const int offset = regarglen[(U8)op];
     const int size = NODE_STEP_REGNODE + offset;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REGINSERT;
+    PERL_UNUSED_ARG(depth);
 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
-    DEBUG_PARSE_FMT("inst"," - %s",reg_name[op]);
+    DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
     if (SIZE_ONLY) {
        RExC_size += size;
        return;
     if (SIZE_ONLY) {
        RExC_size += size;
        return;
@@ -7829,30 +8650,31 @@ S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
     dst = RExC_emit;
     if (RExC_open_parens) {
         int paren;
     dst = RExC_emit;
     if (RExC_open_parens) {
         int paren;
-        DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);
+        /*DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);*/
         for ( paren=0 ; paren < RExC_npar ; paren++ ) {
             if ( RExC_open_parens[paren] >= opnd ) {
         for ( paren=0 ; paren < RExC_npar ; paren++ ) {
             if ( RExC_open_parens[paren] >= opnd ) {
-                DEBUG_PARSE_FMT("open"," - %d",size);
+                /*DEBUG_PARSE_FMT("open"," - %d",size);*/
                 RExC_open_parens[paren] += size;
             } else {
                 RExC_open_parens[paren] += size;
             } else {
-                DEBUG_PARSE_FMT("open"," - %s","ok");
+                /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
             }
             if ( RExC_close_parens[paren] >= opnd ) {
             }
             if ( RExC_close_parens[paren] >= opnd ) {
-                DEBUG_PARSE_FMT("close"," - %d",size);
+                /*DEBUG_PARSE_FMT("close"," - %d",size);*/
                 RExC_close_parens[paren] += size;
             } else {
                 RExC_close_parens[paren] += size;
             } else {
-                DEBUG_PARSE_FMT("close"," - %s","ok");
+                /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
             }
         }
     }
 
     while (src > opnd) {
        StructCopy(--src, --dst, regnode);
             }
         }
     }
 
     while (src > opnd) {
        StructCopy(--src, --dst, regnode);
+#ifdef RE_TRACK_PATTERN_OFFSETS
         if (RExC_offsets) {     /* MJD 20010112 */
            MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
                   "reg_insert",
                  __LINE__,
         if (RExC_offsets) {     /* MJD 20010112 */
            MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
                   "reg_insert",
                  __LINE__,
-                 reg_name[op],
+                 PL_reg_name[op],
                   (UV)(dst - RExC_emit_start) > RExC_offsets[0] 
                    ? "Overwriting end of array!\n" : "OK",
                   (UV)(src - RExC_emit_start),
                   (UV)(dst - RExC_emit_start) > RExC_offsets[0] 
                    ? "Overwriting end of array!\n" : "OK",
                   (UV)(src - RExC_emit_start),
@@ -7861,15 +8683,17 @@ S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
            Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
            Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
         }
            Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
            Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
         }
+#endif
     }
     
 
     place = opnd;              /* Op node, where operand used to be. */
     }
     
 
     place = opnd;              /* Op node, where operand used to be. */
+#ifdef RE_TRACK_PATTERN_OFFSETS
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n", 
               "reginsert",
              __LINE__,
     if (RExC_offsets) {         /* MJD */
        MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n", 
               "reginsert",
              __LINE__,
-             reg_name[op],
+             PL_reg_name[op],
               (UV)(place - RExC_emit_start) > RExC_offsets[0] 
               ? "Overwriting end of array!\n" : "OK",
               (UV)(place - RExC_emit_start),
               (UV)(place - RExC_emit_start) > RExC_offsets[0] 
               ? "Overwriting end of array!\n" : "OK",
               (UV)(place - RExC_emit_start),
@@ -7878,6 +8702,7 @@ S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
        Set_Node_Offset(place, RExC_parse);
        Set_Node_Length(place, 1);
     }
        Set_Node_Offset(place, RExC_parse);
        Set_Node_Length(place, 1);
     }
+#endif    
     src = NEXTOPER(place);
     FILL_ADVANCE_NODE(place, op);
     Zero(src, offset, regnode);
     src = NEXTOPER(place);
     FILL_ADVANCE_NODE(place, op);
     Zero(src, offset, regnode);
@@ -7894,6 +8719,8 @@ S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 de
     dVAR;
     register regnode *scan;
     GET_RE_DEBUG_FLAGS_DECL;
     dVAR;
     register regnode *scan;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_REGTAIL;
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
 #ifndef DEBUGGING
     PERL_UNUSED_ARG(depth);
 #endif
@@ -7912,7 +8739,7 @@ S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 de
             PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
                 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
                     (temp == NULL ? "->" : ""),
             PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
                 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
                     (temp == NULL ? "->" : ""),
-                    (temp == NULL ? reg_name[OP(val)] : "")
+                    (temp == NULL ? PL_reg_name[OP(val)] : "")
             );
         });
         if (temp == NULL)
             );
         });
         if (temp == NULL)
@@ -7954,9 +8781,10 @@ S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,
 #ifdef EXPERIMENTAL_INPLACESCAN
     I32 min = 0;
 #endif
 #ifdef EXPERIMENTAL_INPLACESCAN
     I32 min = 0;
 #endif
-
     GET_RE_DEBUG_FLAGS_DECL;
 
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_REGTAIL_STUDY;
+
 
     if (SIZE_ONLY)
         return exact;
 
     if (SIZE_ONLY)
         return exact;
@@ -7993,7 +8821,7 @@ S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,
             PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
                 SvPV_nolen_const(mysv),
                 REG_NODE_NUM(scan),
             PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
                 SvPV_nolen_const(mysv),
                 REG_NODE_NUM(scan),
-                reg_name[exact]);
+                PL_reg_name[exact]);
         });
        if (temp == NULL)
            break;
         });
        if (temp == NULL)
            break;
@@ -8003,10 +8831,10 @@ S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,
         SV * const mysv_val=sv_newmortal();
         DEBUG_PARSE_MSG("");
         regprop(RExC_rx, mysv_val, val);
         SV * const mysv_val=sv_newmortal();
         DEBUG_PARSE_MSG("");
         regprop(RExC_rx, mysv_val, val);
-        PerlIO_printf(Perl_debug_log, "~ attach to %s (%d) offset to %d\n",
-            SvPV_nolen_const(mysv_val),
-            REG_NODE_NUM(val),
-            val - scan
+        PerlIO_printf(Perl_debug_log, "~ attach to %s (%"IVdf") offset to %"IVdf"\n",
+                     SvPV_nolen_const(mysv_val),
+                     (IV)REG_NODE_NUM(val),
+                     (IV)(val - scan)
         );
     });
     if (reg_off_by_arg[OP(scan)]) {
         );
     });
     if (reg_off_by_arg[OP(scan)]) {
@@ -8023,9 +8851,11 @@ S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,
 /*
  - regcurly - a little FSA that accepts {\d+,?\d*}
  */
 /*
  - regcurly - a little FSA that accepts {\d+,?\d*}
  */
-STATIC I32
-S_regcurly(register const char *s)
+I32
+Perl_regcurly(register const char *s)
 {
 {
+    PERL_ARGS_ASSERT_REGCURLY;
+
     if (*s++ != '{')
        return FALSE;
     if (!isDIGIT(*s))
     if (*s++ != '{')
        return FALSE;
     if (!isDIGIT(*s))
@@ -8045,6 +8875,29 @@ S_regcurly(register const char *s)
 /*
  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
  */
 /*
  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
  */
+#ifdef DEBUGGING
+static void 
+S_regdump_extflags(pTHX_ const char *lead, const U32 flags)
+{
+    int bit;
+    int set=0;
+
+    for (bit=0; bit<32; bit++) {
+        if (flags & (1<<bit)) {
+            if (!set++ && lead) 
+                PerlIO_printf(Perl_debug_log, "%s",lead);
+            PerlIO_printf(Perl_debug_log, "%s ",PL_reg_extflags_name[bit]);
+        }              
+    }     
+    if (lead)  {
+        if (set) 
+            PerlIO_printf(Perl_debug_log, "\n");
+        else 
+            PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
+    }            
+}   
+#endif
+
 void
 Perl_regdump(pTHX_ const regexp *r)
 {
 void
 Perl_regdump(pTHX_ const regexp *r)
 {
@@ -8052,8 +8905,12 @@ Perl_regdump(pTHX_ const regexp *r)
     dVAR;
     SV * const sv = sv_newmortal();
     SV *dsv= sv_newmortal();
     dVAR;
     SV * const sv = sv_newmortal();
     SV *dsv= sv_newmortal();
+    RXi_GET_DECL(r,ri);
+    GET_RE_DEBUG_FLAGS_DECL;
 
 
-    (void)dumpuntil(r, r->program, r->program + 1, NULL, NULL, sv, 0, 0);
+    PERL_ARGS_ASSERT_REGDUMP;
+
+    (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
 
     /* Header fields of interest. */
     if (r->anchored_substr) {
 
     /* Header fields of interest. */
     if (r->anchored_substr) {
@@ -8092,40 +8949,42 @@ Perl_regdump(pTHX_ const regexp *r)
                      (r->check_substr == r->float_substr
                       && r->check_utf8 == r->float_utf8
                       ? "(checking floating" : "(checking anchored"));
                      (r->check_substr == r->float_substr
                       && r->check_utf8 == r->float_utf8
                       ? "(checking floating" : "(checking anchored"));
-    if (r->reganch & ROPT_NOSCAN)
+    if (r->extflags & RXf_NOSCAN)
        PerlIO_printf(Perl_debug_log, " noscan");
        PerlIO_printf(Perl_debug_log, " noscan");
-    if (r->reganch & ROPT_CHECK_ALL)
+    if (r->extflags & RXf_CHECK_ALL)
        PerlIO_printf(Perl_debug_log, " isall");
     if (r->check_substr || r->check_utf8)
        PerlIO_printf(Perl_debug_log, ") ");
 
        PerlIO_printf(Perl_debug_log, " isall");
     if (r->check_substr || r->check_utf8)
        PerlIO_printf(Perl_debug_log, ") ");
 
-    if (r->regstclass) {
-       regprop(r, sv, r->regstclass);
+    if (ri->regstclass) {
+       regprop(r, sv, ri->regstclass);
        PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
     }
        PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
     }
-    if (r->reganch & ROPT_ANCH) {
+    if (r->extflags & RXf_ANCH) {
        PerlIO_printf(Perl_debug_log, "anchored");
        PerlIO_printf(Perl_debug_log, "anchored");
-       if (r->reganch & ROPT_ANCH_BOL)
+       if (r->extflags & RXf_ANCH_BOL)
            PerlIO_printf(Perl_debug_log, "(BOL)");
            PerlIO_printf(Perl_debug_log, "(BOL)");
-       if (r->reganch & ROPT_ANCH_MBOL)
+       if (r->extflags & RXf_ANCH_MBOL)
            PerlIO_printf(Perl_debug_log, "(MBOL)");
            PerlIO_printf(Perl_debug_log, "(MBOL)");
-       if (r->reganch & ROPT_ANCH_SBOL)
+       if (r->extflags & RXf_ANCH_SBOL)
            PerlIO_printf(Perl_debug_log, "(SBOL)");
            PerlIO_printf(Perl_debug_log, "(SBOL)");
-       if (r->reganch & ROPT_ANCH_GPOS)
+       if (r->extflags & RXf_ANCH_GPOS)
            PerlIO_printf(Perl_debug_log, "(GPOS)");
        PerlIO_putc(Perl_debug_log, ' ');
     }
            PerlIO_printf(Perl_debug_log, "(GPOS)");
        PerlIO_putc(Perl_debug_log, ' ');
     }
-    if (r->reganch & ROPT_GPOS_SEEN)
-       PerlIO_printf(Perl_debug_log, "GPOS ");
-    if (r->reganch & ROPT_SKIP)
+    if (r->extflags & RXf_GPOS_SEEN)
+       PerlIO_printf(Perl_debug_log, "GPOS:%"UVuf" ", (UV)r->gofs);
+    if (r->intflags & PREGf_SKIP)
        PerlIO_printf(Perl_debug_log, "plus ");
        PerlIO_printf(Perl_debug_log, "plus ");
-    if (r->reganch & ROPT_IMPLICIT)
+    if (r->intflags & PREGf_IMPLICIT)
        PerlIO_printf(Perl_debug_log, "implicit ");
        PerlIO_printf(Perl_debug_log, "implicit ");
-    PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
-    if (r->reganch & ROPT_EVAL_SEEN)
+    PerlIO_printf(Perl_debug_log, "minlen %"IVdf" ", (IV)r->minlen);
+    if (r->extflags & RXf_EVAL_SEEN)
        PerlIO_printf(Perl_debug_log, "with eval ");
     PerlIO_printf(Perl_debug_log, "\n");
        PerlIO_printf(Perl_debug_log, "with eval ");
     PerlIO_printf(Perl_debug_log, "\n");
+    DEBUG_FLAGS_r(regdump_extflags("r->extflags: ",r->extflags));            
 #else
 #else
+    PERL_ARGS_ASSERT_REGDUMP;
     PERL_UNUSED_CONTEXT;
     PERL_UNUSED_ARG(r);
 #endif /* DEBUGGING */
     PERL_UNUSED_CONTEXT;
     PERL_UNUSED_ARG(r);
 #endif /* DEBUGGING */
@@ -8134,51 +8993,62 @@ Perl_regdump(pTHX_ const regexp *r)
 /*
 - regprop - printable representation of opcode
 */
 /*
 - regprop - printable representation of opcode
 */
+#define EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags) \
+STMT_START { \
+        if (do_sep) {                           \
+            Perl_sv_catpvf(aTHX_ sv,"%s][%s",PL_colors[1],PL_colors[0]); \
+            if (flags & ANYOF_INVERT)           \
+                /*make sure the invert info is in each */ \
+                sv_catpvs(sv, "^");             \
+            do_sep = 0;                         \
+        }                                       \
+} STMT_END
+
 void
 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
 {
 #ifdef DEBUGGING
     dVAR;
     register int k;
 void
 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
 {
 #ifdef DEBUGGING
     dVAR;
     register int k;
+    RXi_GET_DECL(prog,progi);
     GET_RE_DEBUG_FLAGS_DECL;
     GET_RE_DEBUG_FLAGS_DECL;
+    
+    PERL_ARGS_ASSERT_REGPROP;
 
 
-    sv_setpvn(sv, "", 0);
+    sv_setpvs(sv, "");
 
     if (OP(o) > REGNODE_MAX)           /* regnode.type is unsigned */
        /* It would be nice to FAIL() here, but this may be called from
           regexec.c, and it would be hard to supply pRExC_state. */
        Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
 
     if (OP(o) > REGNODE_MAX)           /* regnode.type is unsigned */
        /* It would be nice to FAIL() here, but this may be called from
           regexec.c, and it would be hard to supply pRExC_state. */
        Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
-    sv_catpv(sv, reg_name[OP(o)]); /* Take off const! */
+    sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
 
     k = PL_regkind[OP(o)];
 
     if (k == EXACT) {
 
     k = PL_regkind[OP(o)];
 
     if (k == EXACT) {
-       SV * const dsv = sv_2mortal(newSVpvs(""));
+       sv_catpvs(sv, " ");
        /* Using is_utf8_string() (via PERL_PV_UNI_DETECT) 
         * is a crude hack but it may be the best for now since 
         * we have no flag "this EXACTish node was UTF-8" 
         * --jhi */
        /* Using is_utf8_string() (via PERL_PV_UNI_DETECT) 
         * is a crude hack but it may be the best for now since 
         * we have no flag "this EXACTish node was UTF-8" 
         * --jhi */
-       const char * const s = 
-           pv_pretty(dsv, STRING(o), STR_LEN(o), 60, 
-               PL_colors[0], PL_colors[1],
-               PERL_PV_ESCAPE_UNI_DETECT |
-               PERL_PV_PRETTY_ELIPSES    |
-               PERL_PV_PRETTY_LTGT    
-            ); 
-       Perl_sv_catpvf(aTHX_ sv, " %s", s );
+       pv_pretty(sv, STRING(o), STR_LEN(o), 60, PL_colors[0], PL_colors[1],
+                 PERL_PV_ESCAPE_UNI_DETECT |
+                 PERL_PV_PRETTY_ELLIPSES   |
+                 PERL_PV_PRETTY_LTGT       |
+                 PERL_PV_PRETTY_NOCLEAR
+                 );
     } else if (k == TRIE) {
        /* print the details of the trie in dumpuntil instead, as
     } else if (k == TRIE) {
        /* print the details of the trie in dumpuntil instead, as
-        * prog->data isn't available here */
+        * progi->data isn't available here */
         const char op = OP(o);
         const char op = OP(o);
-        const I32 n = ARG(o);
+        const U32 n = ARG(o);
         const reg_ac_data * const ac = IS_TRIE_AC(op) ?
         const reg_ac_data * const ac = IS_TRIE_AC(op) ?
-               (reg_ac_data *)prog->data->data[n] :
+               (reg_ac_data *)progi->data->data[n] :
                NULL;
                NULL;
-        const reg_trie_data * const trie = !IS_TRIE_AC(op) ?
-            (reg_trie_data*)prog->data->data[n] :
-            ac->trie;
+        const reg_trie_data * const trie
+           = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
         
         
-        Perl_sv_catpvf(aTHX_ sv, "-%s",reg_name[o->flags]);
+        Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
         DEBUG_TRIE_COMPILE_r(
             Perl_sv_catpvf(aTHX_ sv,
                 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
         DEBUG_TRIE_COMPILE_r(
             Perl_sv_catpvf(aTHX_ sv,
                 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
@@ -8195,7 +9065,7 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
             int i;
             int rangestart = -1;
             U8* bitmap = IS_ANYOF_TRIE(op) ? (U8*)ANYOF_BITMAP(o) : (U8*)TRIE_BITMAP(trie);
             int i;
             int rangestart = -1;
             U8* bitmap = IS_ANYOF_TRIE(op) ? (U8*)ANYOF_BITMAP(o) : (U8*)TRIE_BITMAP(trie);
-            Perl_sv_catpvf(aTHX_ sv, "[");
+            sv_catpvs(sv, "[");
             for (i = 0; i <= 256; i++) {
                 if (i < 256 && BITMAP_TEST(bitmap,i)) {
                     if (rangestart == -1)
             for (i = 0; i <= 256; i++) {
                 if (i < 256 && BITMAP_TEST(bitmap,i)) {
                     if (rangestart == -1)
@@ -8212,7 +9082,7 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
                     rangestart = -1;
                 }
             }
                     rangestart = -1;
                 }
             }
-            Perl_sv_catpvf(aTHX_ sv, "]");
+            sv_catpvs(sv, "]");
         } 
         
     } else if (k == CURLY) {
         } 
         
     } else if (k == CURLY) {
@@ -8222,19 +9092,44 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
     }
     else if (k == WHILEM && o->flags)                  /* Ordinal/of */
        Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
     }
     else if (k == WHILEM && o->flags)                  /* Ordinal/of */
        Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
-    else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP || OP(o)==ACCEPT) 
+    else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP || OP(o)==ACCEPT) {
        Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
        Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
-    else if (k == GOSUB) 
+       if ( RXp_PAREN_NAMES(prog) ) {
+            if ( k != REF || OP(o) < NREF) {       
+               AV *list= MUTABLE_AV(progi->data->data[progi->name_list_idx]);
+               SV **name= av_fetch(list, ARG(o), 0 );
+               if (name)
+                   Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
+            }      
+            else {
+                AV *list= MUTABLE_AV(progi->data->data[ progi->name_list_idx ]);
+                SV *sv_dat= MUTABLE_SV(progi->data->data[ ARG( o ) ]);
+                I32 *nums=(I32*)SvPVX(sv_dat);
+                SV **name= av_fetch(list, nums[0], 0 );
+                I32 n;
+                if (name) {
+                    for ( n=0; n<SvIVX(sv_dat); n++ ) {
+                        Perl_sv_catpvf(aTHX_ sv, "%s%"IVdf,
+                                   (n ? "," : ""), (IV)nums[n]);
+                    }
+                    Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
+                }
+            }
+        }            
+    } else if (k == GOSUB) 
        Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o)); /* Paren and offset */
     else if (k == VERB) {
         if (!o->flags) 
             Perl_sv_catpvf(aTHX_ sv, ":%"SVf, 
        Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o)); /* Paren and offset */
     else if (k == VERB) {
         if (!o->flags) 
             Perl_sv_catpvf(aTHX_ sv, ":%"SVf, 
-                (SV*)prog->data->data[ ARG( o ) ]);
+                          SVfARG((MUTABLE_SV(progi->data->data[ ARG( o ) ]))));
     } else if (k == LOGICAL)
        Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
     } else if (k == LOGICAL)
        Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
+    else if (k == FOLDCHAR)
+       Perl_sv_catpvf(aTHX_ sv, "[0x%"UVXf"]", PTR2UV(ARG(o)) );
     else if (k == ANYOF) {
        int i, rangestart = -1;
        const U8 flags = ANYOF_FLAGS(o);
     else if (k == ANYOF) {
        int i, rangestart = -1;
        const U8 flags = ANYOF_FLAGS(o);
+       int do_sep = 0;
 
        /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
        static const char * const anyofs[] = {
 
        /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
        static const char * const anyofs[] = {
@@ -8250,8 +9145,8 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
            "[:^alpha:]",
            "[:ascii:]",
            "[:^ascii:]",
            "[:^alpha:]",
            "[:ascii:]",
            "[:^ascii:]",
-           "[:ctrl:]",
-           "[:^ctrl:]",
+           "[:cntrl:]",
+           "[:^cntrl:]",
            "[:graph:]",
            "[:^graph:]",
            "[:lower:]",
            "[:graph:]",
            "[:^graph:]",
            "[:lower:]",
@@ -8277,6 +9172,8 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
        Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
        if (flags & ANYOF_INVERT)
            sv_catpvs(sv, "^");
        Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
        if (flags & ANYOF_INVERT)
            sv_catpvs(sv, "^");
+       
+       /* output what the standard cp 0-255 bitmap matches */
        for (i = 0; i <= 256; i++) {
            if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
                if (rangestart == -1)
        for (i = 0; i <= 256; i++) {
            if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
                if (rangestart == -1)
@@ -8290,15 +9187,23 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
                    sv_catpvs(sv, "-");
                    put_byte(sv, i - 1);
                }
                    sv_catpvs(sv, "-");
                    put_byte(sv, i - 1);
                }
+               do_sep = 1;
                rangestart = -1;
            }
        }
                rangestart = -1;
            }
        }
-
+        
+        EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags);
+        /* output any special charclass tests (used mostly under use locale) */
        if (o->flags & ANYOF_CLASS)
            for (i = 0; i < (int)(sizeof(anyofs)/sizeof(char*)); i++)
        if (o->flags & ANYOF_CLASS)
            for (i = 0; i < (int)(sizeof(anyofs)/sizeof(char*)); i++)
-               if (ANYOF_CLASS_TEST(o,i))
+               if (ANYOF_CLASS_TEST(o,i)) {
                    sv_catpv(sv, anyofs[i]);
                    sv_catpv(sv, anyofs[i]);
-
+                   do_sep = 1;
+               }
+        
+        EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags);
+        
+        /* output information about the unicode matching */
        if (flags & ANYOF_UNICODE)
            sv_catpvs(sv, "{unicode}");
        else if (flags & ANYOF_UNICODE_ALL)
        if (flags & ANYOF_UNICODE)
            sv_catpvs(sv, "{unicode}");
        else if (flags & ANYOF_UNICODE_ALL)
@@ -8311,7 +9216,7 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
            if (lv) {
                if (sw) {
                    U8 s[UTF8_MAXBYTES_CASE+1];
            if (lv) {
                if (sw) {
                    U8 s[UTF8_MAXBYTES_CASE+1];
-               
+
                    for (i = 0; i <= 256; i++) { /* just the first 256 */
                        uvchr_to_utf8(s, i);
                        
                    for (i = 0; i <= 256; i++) { /* just the first 256 */
                        uvchr_to_utf8(s, i);
                        
@@ -8382,10 +9287,13 @@ Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
 }
 
 SV *
 }
 
 SV *
-Perl_re_intuit_string(pTHX_ regexp *prog)
+Perl_re_intuit_string(pTHX_ REGEXP * const r)
 {                              /* Assume that RE_INTUIT is set */
     dVAR;
 {                              /* Assume that RE_INTUIT is set */
     dVAR;
+    struct regexp *const prog = (struct regexp *)SvANY(r);
     GET_RE_DEBUG_FLAGS_DECL;
     GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_RE_INTUIT_STRING;
     PERL_UNUSED_CONTEXT;
 
     DEBUG_COMPILE_r(
     PERL_UNUSED_CONTEXT;
 
     DEBUG_COMPILE_r(
@@ -8408,72 +9316,175 @@ Perl_re_intuit_string(pTHX_ regexp *prog)
 }
 
 /* 
 }
 
 /* 
-   pregfree - free a regexp
+   pregfree() 
    
    
-   See regdupe below if you change anything here. 
+   handles refcounting and freeing the perl core regexp structure. When 
+   it is necessary to actually free the structure the first thing it 
+   does is call the 'free' method of the regexp_engine associated to to 
+   the regexp, allowing the handling of the void *pprivate; member 
+   first. (This routine is not overridable by extensions, which is why 
+   the extensions free is called first.)
+   
+   See regdupe and regdupe_internal if you change anything here. 
 */
 */
+#ifndef PERL_IN_XSUB_RE
+void
+Perl_pregfree(pTHX_ REGEXP *r)
+{
+    SvREFCNT_dec(r);
+}
 
 void
 
 void
-Perl_pregfree(pTHX_ struct regexp *r)
+Perl_pregfree2(pTHX_ REGEXP *rx)
 {
     dVAR;
 {
     dVAR;
+    struct regexp *const r = (struct regexp *)SvANY(rx);
+    GET_RE_DEBUG_FLAGS_DECL;
+
+    PERL_ARGS_ASSERT_PREGFREE2;
+
+    if (r->mother_re) {
+        ReREFCNT_dec(r->mother_re);
+    } else {
+        CALLREGFREE_PVT(rx); /* free the private data */
+        SvREFCNT_dec(RXp_PAREN_NAMES(r));
+    }        
+    if (r->substrs) {
+        SvREFCNT_dec(r->anchored_substr);
+        SvREFCNT_dec(r->anchored_utf8);
+        SvREFCNT_dec(r->float_substr);
+        SvREFCNT_dec(r->float_utf8);
+       Safefree(r->substrs);
+    }
+    RX_MATCH_COPY_FREE(rx);
+#ifdef PERL_OLD_COPY_ON_WRITE
+    SvREFCNT_dec(r->saved_copy);
+#endif
+    Safefree(r->offs);
+}
+
+/*  reg_temp_copy()
+    
+    This is a hacky workaround to the structural issue of match results
+    being stored in the regexp structure which is in turn stored in
+    PL_curpm/PL_reg_curpm. The problem is that due to qr// the pattern
+    could be PL_curpm in multiple contexts, and could require multiple
+    result sets being associated with the pattern simultaneously, such
+    as when doing a recursive match with (??{$qr})
+    
+    The solution is to make a lightweight copy of the regexp structure 
+    when a qr// is returned from the code executed by (??{$qr}) this
+    lightweight copy doesnt actually own any of its data except for
+    the starp/end and the actual regexp structure itself. 
+    
+*/    
+    
+    
+REGEXP *
+Perl_reg_temp_copy (pTHX_ REGEXP *ret_x, REGEXP *rx)
+{
+    struct regexp *ret;
+    struct regexp *const r = (struct regexp *)SvANY(rx);
+    register const I32 npar = r->nparens+1;
+
+    PERL_ARGS_ASSERT_REG_TEMP_COPY;
+
+    if (!ret_x)
+       ret_x = (REGEXP*) newSV_type(SVt_REGEXP);
+    ret = (struct regexp *)SvANY(ret_x);
+    
+    (void)ReREFCNT_inc(rx);
+    /* We can take advantage of the existing "copied buffer" mechanism in SVs
+       by pointing directly at the buffer, but flagging that the allocated
+       space in the copy is zero. As we've just done a struct copy, it's now
+       a case of zero-ing that, rather than copying the current length.  */
+    SvPV_set(ret_x, RX_WRAPPED(rx));
+    SvFLAGS(ret_x) |= SvFLAGS(rx) & (SVf_POK|SVp_POK|SVf_UTF8);
+    memcpy(&(ret->xpv_cur), &(r->xpv_cur),
+          sizeof(regexp) - STRUCT_OFFSET(regexp, xpv_cur));
+    SvLEN_set(ret_x, 0);
+    SvSTASH_set(ret_x, NULL);
+    SvMAGIC_set(ret_x, NULL);
+    Newx(ret->offs, npar, regexp_paren_pair);
+    Copy(r->offs, ret->offs, npar, regexp_paren_pair);
+    if (r->substrs) {
+        Newx(ret->substrs, 1, struct reg_substr_data);
+       StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
+
+       SvREFCNT_inc_void(ret->anchored_substr);
+       SvREFCNT_inc_void(ret->anchored_utf8);
+       SvREFCNT_inc_void(ret->float_substr);
+       SvREFCNT_inc_void(ret->float_utf8);
+
+       /* check_substr and check_utf8, if non-NULL, point to either their
+          anchored or float namesakes, and don't hold a second reference.  */
+    }
+    RX_MATCH_COPIED_off(ret_x);
+#ifdef PERL_OLD_COPY_ON_WRITE
+    ret->saved_copy = NULL;
+#endif
+    ret->mother_re = rx;
+    
+    return ret_x;
+}
+#endif
+
+/* regfree_internal() 
 
 
+   Free the private data in a regexp. This is overloadable by 
+   extensions. Perl takes care of the regexp structure in pregfree(), 
+   this covers the *pprivate pointer which technically perldoesnt 
+   know about, however of course we have to handle the 
+   regexp_internal structure when no extension is in use. 
+   
+   Note this is called before freeing anything in the regexp 
+   structure. 
+ */
+void
+Perl_regfree_internal(pTHX_ REGEXP * const rx)
+{
+    dVAR;
+    struct regexp *const r = (struct regexp *)SvANY(rx);
+    RXi_GET_DECL(r,ri);
     GET_RE_DEBUG_FLAGS_DECL;
 
     GET_RE_DEBUG_FLAGS_DECL;
 
-    if (!r || (--r->refcnt > 0))
-       return;
+    PERL_ARGS_ASSERT_REGFREE_INTERNAL;
+
     DEBUG_COMPILE_r({
        if (!PL_colorset)
            reginitcolors();
        {
            SV *dsv= sv_newmortal();
     DEBUG_COMPILE_r({
        if (!PL_colorset)
            reginitcolors();
        {
            SV *dsv= sv_newmortal();
-            RE_PV_QUOTED_DECL(s, (r->reganch & ROPT_UTF8),
-                dsv, r->precomp, r->prelen, 60);
+            RE_PV_QUOTED_DECL(s, RX_UTF8(rx),
+                dsv, RX_PRECOMP(rx), RX_PRELEN(rx), 60);
             PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n", 
                 PL_colors[4],PL_colors[5],s);
         }
     });
             PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n", 
                 PL_colors[4],PL_colors[5],s);
         }
     });
-
-    /* gcov results gave these as non-null 100% of the time, so there's no
-       optimisation in checking them before calling Safefree  */
-    Safefree(r->precomp);
-    Safefree(r->offsets);             /* 20010421 MJD */
-    RX_MATCH_COPY_FREE(r);
-#ifdef PERL_OLD_COPY_ON_WRITE
-    if (r->saved_copy)
-       SvREFCNT_dec(r->saved_copy);
+#ifdef RE_TRACK_PATTERN_OFFSETS
+    if (ri->u.offsets)
+        Safefree(ri->u.offsets);             /* 20010421 MJD */
 #endif
 #endif
-    if (r->substrs) {
-       if (r->anchored_substr)
-           SvREFCNT_dec(r->anchored_substr);
-       if (r->anchored_utf8)
-           SvREFCNT_dec(r->anchored_utf8);
-       if (r->float_substr)
-           SvREFCNT_dec(r->float_substr);
-       if (r->float_utf8)
-           SvREFCNT_dec(r->float_utf8);
-       Safefree(r->substrs);
-    }
-    if (r->paren_names)
-            SvREFCNT_dec(r->paren_names);
-    if (r->data) {
-       int n = r->data->count;
+    if (ri->data) {
+       int n = ri->data->count;
        PAD* new_comppad = NULL;
        PAD* old_comppad;
        PADOFFSET refcnt;
 
        while (--n >= 0) {
           /* If you add a ->what type here, update the comment in regcomp.h */
        PAD* new_comppad = NULL;
        PAD* old_comppad;
        PADOFFSET refcnt;
 
        while (--n >= 0) {
           /* If you add a ->what type here, update the comment in regcomp.h */
-           switch (r->data->what[n]) {
+           switch (ri->data->what[n]) {
            case 's':
            case 'S':
            case 's':
            case 'S':
-               SvREFCNT_dec((SV*)r->data->data[n]);
+           case 'u':
+               SvREFCNT_dec(MUTABLE_SV(ri->data->data[n]));
                break;
            case 'f':
                break;
            case 'f':
-               Safefree(r->data->data[n]);
+               Safefree(ri->data->data[n]);
                break;
            case 'p':
                break;
            case 'p':
-               new_comppad = (AV*)r->data->data[n];
+               new_comppad = MUTABLE_AV(ri->data->data[n]);
                break;
            case 'o':
                if (new_comppad == NULL)
                break;
            case 'o':
                if (new_comppad == NULL)
@@ -8483,13 +9494,13 @@ Perl_pregfree(pTHX_ struct regexp *r)
                    (SvTYPE(new_comppad) == SVt_PVAV) ? new_comppad : NULL
                );
                OP_REFCNT_LOCK;
                    (SvTYPE(new_comppad) == SVt_PVAV) ? new_comppad : NULL
                );
                OP_REFCNT_LOCK;
-               refcnt = OpREFCNT_dec((OP_4tree*)r->data->data[n]);
+               refcnt = OpREFCNT_dec((OP_4tree*)ri->data->data[n]);
                OP_REFCNT_UNLOCK;
                if (!refcnt)
                OP_REFCNT_UNLOCK;
                if (!refcnt)
-                    op_free((OP_4tree*)r->data->data[n]);
+                    op_free((OP_4tree*)ri->data->data[n]);
 
                PAD_RESTORE_LOCAL(old_comppad);
 
                PAD_RESTORE_LOCAL(old_comppad);
-               SvREFCNT_dec((SV*)new_comppad);
+               SvREFCNT_dec(MUTABLE_SV(new_comppad));
                new_comppad = NULL;
                break;
            case 'n':
                new_comppad = NULL;
                break;
            case 'n':
@@ -8498,17 +9509,16 @@ Perl_pregfree(pTHX_ struct regexp *r)
                 { /* Aho Corasick add-on structure for a trie node.
                      Used in stclass optimization only */
                     U32 refcount;
                 { /* Aho Corasick add-on structure for a trie node.
                      Used in stclass optimization only */
                     U32 refcount;
-                    reg_ac_data *aho=(reg_ac_data*)r->data->data[n];
+                    reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
                     OP_REFCNT_LOCK;
                     refcount = --aho->refcount;
                     OP_REFCNT_UNLOCK;
                     if ( !refcount ) {
                     OP_REFCNT_LOCK;
                     refcount = --aho->refcount;
                     OP_REFCNT_UNLOCK;
                     if ( !refcount ) {
-                        Safefree(aho->states);
-                        Safefree(aho->fail);
-                        aho->trie=NULL; /* not necessary to free this as it is 
-                                           handled by the 't' case */
-                        Safefree(r->data->data[n]); /* do this last!!!! */
-                        Safefree(r->regstclass);
+                        PerlMemShared_free(aho->states);
+                        PerlMemShared_free(aho->fail);
+                        /* do this last!!!! */
+                        PerlMemShared_free(ri->data->data[n]);
+                        PerlMemShared_free(ri->regstclass);
                     }
                 }
                 break;
                     }
                 }
                 break;
@@ -8516,112 +9526,182 @@ Perl_pregfree(pTHX_ struct regexp *r)
                {
                    /* trie structure. */
                    U32 refcount;
                {
                    /* trie structure. */
                    U32 refcount;
-                   reg_trie_data *trie=(reg_trie_data*)r->data->data[n];
+                   reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
                     OP_REFCNT_LOCK;
                     refcount = --trie->refcount;
                     OP_REFCNT_UNLOCK;
                     if ( !refcount ) {
                     OP_REFCNT_LOCK;
                     refcount = --trie->refcount;
                     OP_REFCNT_UNLOCK;
                     if ( !refcount ) {
-                        Safefree(trie->charmap);
-                        if (trie->widecharmap)
-                            SvREFCNT_dec((SV*)trie->widecharmap);
-                        Safefree(trie->states);
-                        Safefree(trie->trans);
+                        PerlMemShared_free(trie->charmap);
+                        PerlMemShared_free(trie->states);
+                        PerlMemShared_free(trie->trans);
                         if (trie->bitmap)
                         if (trie->bitmap)
-                            Safefree(trie->bitmap);
+                            PerlMemShared_free(trie->bitmap);
                         if (trie->wordlen)
                         if (trie->wordlen)
-                            Safefree(trie->wordlen);
+                            PerlMemShared_free(trie->wordlen);
                         if (trie->jump)
                         if (trie->jump)
-                            Safefree(trie->jump);
+                            PerlMemShared_free(trie->jump);
                         if (trie->nextword)
                         if (trie->nextword)
-                            Safefree(trie->nextword);
-#ifdef DEBUGGING
-                        if (trie->words)
-                            SvREFCNT_dec((SV*)trie->words);
-                        if (trie->revcharmap)
-                            SvREFCNT_dec((SV*)trie->revcharmap);
-#endif
-                        Safefree(r->data->data[n]); /* do this last!!!! */
+                            PerlMemShared_free(trie->nextword);
+                        /* do this last!!!! */
+                        PerlMemShared_free(ri->data->data[n]);
                    }
                }
                break;
            default:
                    }
                }
                break;
            default:
-               Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
+               Perl_croak(aTHX_ "panic: regfree data code '%c'", ri->data->what[n]);
            }
        }
            }
        }
-       Safefree(r->data->what);
-       Safefree(r->data);
-    }
-    Safefree(r->startp);
-    Safefree(r->endp);
-    if (r->swap) {
-        Safefree(r->swap->startp);
-        Safefree(r->swap->endp);
-        Safefree(r->swap);
+       Safefree(ri->data->what);
+       Safefree(ri->data);
     }
     }
-    Safefree(r);
+
+    Safefree(ri);
 }
 
 #define sv_dup_inc(s,t)        SvREFCNT_inc(sv_dup(s,t))
 }
 
 #define sv_dup_inc(s,t)        SvREFCNT_inc(sv_dup(s,t))
-#define av_dup_inc(s,t)        (AV*)SvREFCNT_inc(sv_dup((SV*)s,t))
-#define hv_dup_inc(s,t)        (HV*)SvREFCNT_inc(sv_dup((SV*)s,t))
+#define av_dup_inc(s,t)        MUTABLE_AV(SvREFCNT_inc(sv_dup((const SV *)s,t)))
+#define hv_dup_inc(s,t)        MUTABLE_HV(SvREFCNT_inc(sv_dup((const SV *)s,t)))
 #define SAVEPVN(p,n)   ((p) ? savepvn(p,n) : NULL)
 
 /* 
 #define SAVEPVN(p,n)   ((p) ? savepvn(p,n) : NULL)
 
 /* 
-   regdupe - duplicate a regexp. 
-   
-   This routine is called by sv.c's re_dup and is expected to clone a 
-   given regexp structure. It is a no-op when not under USE_ITHREADS. 
-   (Originally this *was* re_dup() for change history see sv.c)
+   re_dup - duplicate a regexp. 
    
    
-   See pregfree() above if you change anything here. 
+   This routine is expected to clone a given regexp structure. It is only
+   compiled under USE_ITHREADS.
+
+   After all of the core data stored in struct regexp is duplicated
+   the regexp_engine.dupe method is used to copy any private data
+   stored in the *pprivate pointer. This allows extensions to handle
+   any duplication it needs to do.
+
+   See pregfree() and regfree_internal() if you change anything here. 
 */
 #if defined(USE_ITHREADS)
 */
 #if defined(USE_ITHREADS)
-regexp *
-Perl_regdupe(pTHX_ const regexp *r, CLONE_PARAMS *param)
+#ifndef PERL_IN_XSUB_RE
+void
+Perl_re_dup_guts(pTHX_ const REGEXP *sstr, REGEXP *dstr, CLONE_PARAMS *param)
 {
     dVAR;
 {
     dVAR;
-    REGEXP *ret;
-    int i, len, npar;
-    struct reg_substr_datum *s;
+    I32 npar;
+    const struct regexp *r = (const struct regexp *)SvANY(sstr);
+    struct regexp *ret = (struct regexp *)SvANY(dstr);
+    
+    PERL_ARGS_ASSERT_RE_DUP_GUTS;
 
 
-    if (!r)
-       return (REGEXP *)NULL;
+    npar = r->nparens+1;
+    Newx(ret->offs, npar, regexp_paren_pair);
+    Copy(r->offs, ret->offs, npar, regexp_paren_pair);
+    if(ret->swap) {
+        /* no need to copy these */
+        Newx(ret->swap, npar, regexp_paren_pair);
+    }
 
 
-    if ((ret = (REGEXP *)ptr_table_fetch(PL_ptr_table, r)))
-       return ret;
+    if (ret->substrs) {
+       /* Do it this way to avoid reading from *r after the StructCopy().
+          That way, if any of the sv_dup_inc()s dislodge *r from the L1
+          cache, it doesn't matter.  */
+       const bool anchored = r->check_substr
+           ? r->check_substr == r->anchored_substr
+           : r->check_utf8 == r->anchored_utf8;
+        Newx(ret->substrs, 1, struct reg_substr_data);
+       StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
+
+       ret->anchored_substr = sv_dup_inc(ret->anchored_substr, param);
+       ret->anchored_utf8 = sv_dup_inc(ret->anchored_utf8, param);
+       ret->float_substr = sv_dup_inc(ret->float_substr, param);
+       ret->float_utf8 = sv_dup_inc(ret->float_utf8, param);
+
+       /* check_substr and check_utf8, if non-NULL, point to either their
+          anchored or float namesakes, and don't hold a second reference.  */
+
+       if (ret->check_substr) {
+           if (anchored) {
+               assert(r->check_utf8 == r->anchored_utf8);
+               ret->check_substr = ret->anchored_substr;
+               ret->check_utf8 = ret->anchored_utf8;
+           } else {
+               assert(r->check_substr == r->float_substr);
+               assert(r->check_utf8 == r->float_utf8);
+               ret->check_substr = ret->float_substr;
+               ret->check_utf8 = ret->float_utf8;
+           }
+       } else if (ret->check_utf8) {
+           if (anchored) {
+               ret->check_utf8 = ret->anchored_utf8;
+           } else {
+               ret->check_utf8 = ret->float_utf8;
+           }
+       }
+    }
 
 
-    len = r->offsets[0];
-    npar = r->nparens+1;
+    RXp_PAREN_NAMES(ret) = hv_dup_inc(RXp_PAREN_NAMES(ret), param);
 
 
-    Newxc(ret, sizeof(regexp) + (len+1)*sizeof(regnode), char, regexp);
-    Copy(r->program, ret->program, len+1, regnode);
+    if (ret->pprivate)
+       RXi_SET(ret,CALLREGDUPE_PVT(dstr,param));
 
 
-    Newx(ret->startp, npar, I32);
-    Copy(r->startp, ret->startp, npar, I32);
-    Newx(ret->endp, npar, I32);
-    Copy(r->startp, ret->startp, npar, I32);
-    if(r->swap) {
-        Newx(ret->swap, 1, regexp_paren_ofs);
-        /* no need to copy these */
-        Newx(ret->swap->startp, npar, I32);
-        Newx(ret->swap->endp, npar, I32);
-    } else {
-        ret->swap = NULL;
-    }
+    if (RX_MATCH_COPIED(dstr))
+       ret->subbeg  = SAVEPVN(ret->subbeg, ret->sublen);
+    else
+       ret->subbeg = NULL;
+#ifdef PERL_OLD_COPY_ON_WRITE
+    ret->saved_copy = NULL;
+#endif
 
 
-    Newx(ret->substrs, 1, struct reg_substr_data);
-    for (s = ret->substrs->data, i = 0; i < 3; i++, s++) {
-       s->min_offset = r->substrs->data[i].min_offset;
-       s->max_offset = r->substrs->data[i].max_offset;
-       s->end_shift  = r->substrs->data[i].end_shift;
-       s->substr     = sv_dup_inc(r->substrs->data[i].substr, param);
-       s->utf8_substr = sv_dup_inc(r->substrs->data[i].utf8_substr, param);
+    if (ret->mother_re) {
+       if (SvPVX_const(dstr) == SvPVX_const(ret->mother_re)) {
+           /* Our storage points directly to our mother regexp, but that's
+              1: a buffer in a different thread
+              2: something we no longer hold a reference on
+              so we need to copy it locally.  */
+           /* Note we need to sue SvCUR() on our mother_re, because it, in
+              turn, may well be pointing to its own mother_re.  */
+           SvPV_set(dstr, SAVEPVN(SvPVX_const(ret->mother_re),
+                                  SvCUR(ret->mother_re)+1));
+           SvLEN_set(dstr, SvCUR(ret->mother_re)+1);
+       }
+       ret->mother_re      = NULL;
     }
     }
+    ret->gofs = 0;
+}
+#endif /* PERL_IN_XSUB_RE */
+
+/*
+   regdupe_internal()
+   
+   This is the internal complement to regdupe() which is used to copy
+   the structure pointed to by the *pprivate pointer in the regexp.
+   This is the core version of the extension overridable cloning hook.
+   The regexp structure being duplicated will be copied by perl prior
+   to this and will be provided as the regexp *r argument, however 
+   with the /old/ structures pprivate pointer value. Thus this routine
+   may override any copying normally done by perl.
+   
+   It returns a pointer to the new regexp_internal structure.
+*/
+
+void *
+Perl_regdupe_internal(pTHX_ REGEXP * const rx, CLONE_PARAMS *param)
+{
+    dVAR;
+    struct regexp *const r = (struct regexp *)SvANY(rx);
+    regexp_internal *reti;
+    int len, npar;
+    RXi_GET_DECL(r,ri);
+
+    PERL_ARGS_ASSERT_REGDUPE_INTERNAL;
+    
+    npar = r->nparens+1;
+    len = ProgLen(ri);
+    
+    Newxc(reti, sizeof(regexp_internal) + len*sizeof(regnode), char, regexp_internal);
+    Copy(ri->program, reti->program, len+1, regnode);
+    
+
+    reti->regstclass = NULL;
 
 
-    ret->regstclass = NULL;
-    if (r->data) {
+    if (ri->data) {
        struct reg_data *d;
        struct reg_data *d;
-        const int count = r->data->count;
+        const int count = ri->data->count;
        int i;
 
        Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
        int i;
 
        Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
@@ -8630,199 +9710,73 @@ Perl_regdupe(pTHX_ const regexp *r, CLONE_PARAMS *param)
 
        d->count = count;
        for (i = 0; i < count; i++) {
 
        d->count = count;
        for (i = 0; i < count; i++) {
-           d->what[i] = r->data->what[i];
+           d->what[i] = ri->data->what[i];
            switch (d->what[i]) {
            switch (d->what[i]) {
-               /* legal options are one of: sSfpont
+               /* legal options are one of: sSfpontTu
                   see also regcomp.h and pregfree() */
            case 's':
            case 'S':
                   see also regcomp.h and pregfree() */
            case 's':
            case 'S':
-               d->data[i] = sv_dup_inc((SV *)r->data->data[i], param);
-               break;
-           case 'p':
-               d->data[i] = av_dup_inc((AV *)r->data->data[i], param);
+           case 'p': /* actually an AV, but the dup function is identical.  */
+           case 'u': /* actually an HV, but the dup function is identical.  */
+               d->data[i] = sv_dup_inc((const SV *)ri->data->data[i], param);
                break;
            case 'f':
                /* This is cheating. */
                Newx(d->data[i], 1, struct regnode_charclass_class);
                break;
            case 'f':
                /* This is cheating. */
                Newx(d->data[i], 1, struct regnode_charclass_class);
-               StructCopy(r->data->data[i], d->data[i],
+               StructCopy(ri->data->data[i], d->data[i],
                            struct regnode_charclass_class);
                            struct regnode_charclass_class);
-               ret->regstclass = (regnode*)d->data[i];
+               reti->regstclass = (regnode*)d->data[i];
                break;
            case 'o':
                break;
            case 'o':
-               /* Compiled op trees are readonly, and can thus be
-                  shared without duplication. */
-               OP_REFCNT_LOCK;
-               d->data[i] = (void*)OpREFCNT_inc((OP*)r->data->data[i]);
-               OP_REFCNT_UNLOCK;
-               break;
-           case 'n':
-               d->data[i] = r->data->data[i];
-               break;
-           case 't':
-               d->data[i] = r->data->data[i];
+               /* Compiled op trees are readonly and in shared memory,
+                  and can thus be shared without duplication. */
                OP_REFCNT_LOCK;
                OP_REFCNT_LOCK;
-               ((reg_trie_data*)d->data[i])->refcount++;
+               d->data[i] = (void*)OpREFCNT_inc((OP*)ri->data->data[i]);
                OP_REFCNT_UNLOCK;
                break;
            case 'T':
                OP_REFCNT_UNLOCK;
                break;
            case 'T':
-               d->data[i] = r->data->data[i];
-               OP_REFCNT_LOCK;
-               ((reg_ac_data*)d->data[i])->refcount++;
-               OP_REFCNT_UNLOCK;
                /* Trie stclasses are readonly and can thus be shared
                 * without duplication. We free the stclass in pregfree
                 * when the corresponding reg_ac_data struct is freed.
                 */
                /* Trie stclasses are readonly and can thus be shared
                 * without duplication. We free the stclass in pregfree
                 * when the corresponding reg_ac_data struct is freed.
                 */
-               ret->regstclass= r->regstclass;
+               reti->regstclass= ri->regstclass;
+               /* Fall through */
+           case 't':
+               OP_REFCNT_LOCK;
+               ((reg_trie_data*)ri->data->data[i])->refcount++;
+               OP_REFCNT_UNLOCK;
+               /* Fall through */
+           case 'n':
+               d->data[i] = ri->data->data[i];
                break;
             default:
                break;
             default:
-               Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'", r->data->what[i]);
+               Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'", ri->data->what[i]);
            }
        }
 
            }
        }
 
-       ret->data = d;
+       reti->data = d;
     }
     else
     }
     else
-       ret->data = NULL;
-
-    Newx(ret->offsets, 2*len+1, U32);
-    Copy(r->offsets, ret->offsets, 2*len+1, U32);
-
-    ret->precomp        = SAVEPVN(r->precomp, r->prelen);
-    ret->refcnt         = r->refcnt;
-    ret->minlen         = r->minlen;
-    ret->minlenret      = r->minlenret;
-    ret->prelen         = r->prelen;
-    ret->nparens        = r->nparens;
-    ret->lastparen      = r->lastparen;
-    ret->lastcloseparen = r->lastcloseparen;
-    ret->reganch        = r->reganch;
+       reti->data = NULL;
 
 
-    ret->sublen         = r->sublen;
+    reti->name_list_idx = ri->name_list_idx;
 
 
-    ret->engine         = r->engine;
-    
-    ret->paren_names    = hv_dup_inc(r->paren_names, param);
-
-    if (RX_MATCH_COPIED(ret))
-       ret->subbeg  = SAVEPVN(r->subbeg, r->sublen);
-    else
-       ret->subbeg = NULL;
-#ifdef PERL_OLD_COPY_ON_WRITE
-    ret->saved_copy = NULL;
+#ifdef RE_TRACK_PATTERN_OFFSETS
+    if (ri->u.offsets) {
+        Newx(reti->u.offsets, 2*len+1, U32);
+        Copy(ri->u.offsets, reti->u.offsets, 2*len+1, U32);
+    }
+#else
+    SetProgLen(reti,len);
 #endif
 
 #endif
 
-    ptr_table_store(PL_ptr_table, r, ret);
-    return ret;
-}
-#endif    
-
-/* 
-   reg_stringify() 
-   
-   converts a regexp embedded in a MAGIC struct to its stringified form, 
-   caching the converted form in the struct and returns the cached 
-   string. 
-
-   If lp is nonnull then it is used to return the length of the 
-   resulting string
-   
-   If flags is nonnull and the returned string contains UTF8 then 
-   (flags & 1) will be true.
-   
-   If haseval is nonnull then it is used to return whether the pattern 
-   contains evals.
-   
-   Normally called via macro: 
-   
-        CALLREG_STRINGIFY(mg,0,0);
-        
-   And internally with
-   
-        CALLREG_AS_STR(mg,lp,flags,haseval)        
-    
-   See sv_2pv_flags() in sv.c for an example of internal usage.
-    
- */
-
-char *
-Perl_reg_stringify(pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval ) {
-    dVAR;
-    const regexp * const re = (regexp *)mg->mg_obj;
-
-    if (!mg->mg_ptr) {
-       const char *fptr = "msix";
-       char reflags[6];
-       char ch;
-       int left = 0;
-       int right = 4;
-       bool need_newline = 0;
-       U16 reganch = (U16)((re->reganch & PMf_COMPILETIME) >> 12);
-
-       while((ch = *fptr++)) {
-           if(reganch & 1) {
-               reflags[left++] = ch;
-           }
-           else {
-               reflags[right--] = ch;
-           }
-           reganch >>= 1;
-       }
-       if(left != 4) {
-           reflags[left] = '-';
-           left = 5;
-       }
-
-       mg->mg_len = re->prelen + 4 + left;
-       /*
-        * If /x was used, we have to worry about a regex ending with a
-        * comment later being embedded within another regex. If so, we don't
-        * want this regex's "commentization" to leak out to the right part of
-        * the enclosing regex, we must cap it with a newline.
-        *
-        * So, if /x was used, we scan backwards from the end of the regex. If
-        * we find a '#' before we find a newline, we need to add a newline
-        * ourself. If we find a '\n' first (or if we don't find '#' or '\n'),
-        * we don't need to add anything.  -jfriedl
-        */
-       if (PMf_EXTENDED & re->reganch) {
-           const char *endptr = re->precomp + re->prelen;
-           while (endptr >= re->precomp) {
-               const char c = *(endptr--);
-               if (c == '\n')
-                   break; /* don't need another */
-               if (c == '#') {
-                   /* we end while in a comment, so we need a newline */
-                   mg->mg_len++; /* save space for it */
-                   need_newline = 1; /* note to add it */
-                   break;
-               }
-           }
-       }
-
-       Newx(mg->mg_ptr, mg->mg_len + 1 + left, char);
-       mg->mg_ptr[0] = '(';
-       mg->mg_ptr[1] = '?';
-       Copy(reflags, mg->mg_ptr+2, left, char);
-       *(mg->mg_ptr+left+2) = ':';
-       Copy(re->precomp, mg->mg_ptr+3+left, re->prelen, char);
-       if (need_newline)
-           mg->mg_ptr[mg->mg_len - 2] = '\n';
-       mg->mg_ptr[mg->mg_len - 1] = ')';
-       mg->mg_ptr[mg->mg_len] = 0;
-    }
-    if (haseval) 
-        *haseval = re->program[0].next_off;
-    if (flags)    
-       *flags = ((re->reganch & ROPT_UTF8) ? 1 : 0);
-    
-    if (lp)
-       *lp = mg->mg_len;
-    return mg->mg_ptr;
+    return (void*)reti;
 }
 
 }
 
+#endif    /* USE_ITHREADS */
 
 #ifndef PERL_IN_XSUB_RE
 
 #ifndef PERL_IN_XSUB_RE
+
 /*
  - regnext - dig the "next" pointer out of a node
  */
 /*
  - regnext - dig the "next" pointer out of a node
  */
@@ -8832,7 +9786,7 @@ Perl_regnext(pTHX_ register regnode *p)
     dVAR;
     register I32 offset;
 
     dVAR;
     register I32 offset;
 
-    if (p == &PL_regdummy)
+    if (!p)
        return(NULL);
 
     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
        return(NULL);
 
     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
@@ -8853,6 +9807,8 @@ S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
     SV *msv;
     const char *message;
 
     SV *msv;
     const char *message;
 
+    PERL_ARGS_ASSERT_RE_CROAK2;
+
     if (l1 > 510)
        l1 = 510;
     if (l1 + l2 > 510)
     if (l1 > 510)
        l1 = 510;
     if (l1 + l2 > 510)
@@ -8913,7 +9869,7 @@ Perl_save_re_context(pTHX)
        const REGEXP * const rx = PM_GETRE(PL_curpm);
        if (rx) {
            U32 i;
        const REGEXP * const rx = PM_GETRE(PL_curpm);
        if (rx) {
            U32 i;
-           for (i = 1; i <= rx->nparens; i++) {
+           for (i = 1; i <= RX_NPARENS(rx); i++) {
                char digits[TYPE_CHARS(long)];
                const STRLEN len = my_snprintf(digits, sizeof(digits), "%lu", (long)i);
                GV *const *const gvp
                char digits[TYPE_CHARS(long)];
                const STRLEN len = my_snprintf(digits, sizeof(digits), "%lu", (long)i);
                GV *const *const gvp
@@ -8934,7 +9890,7 @@ static void
 clear_re(pTHX_ void *r)
 {
     dVAR;
 clear_re(pTHX_ void *r)
 {
     dVAR;
-    ReREFCNT_dec((regexp *)r);
+    ReREFCNT_dec((REGEXP *)r);
 }
 
 #ifdef DEBUGGING
 }
 
 #ifdef DEBUGGING
@@ -8942,18 +9898,32 @@ clear_re(pTHX_ void *r)
 STATIC void
 S_put_byte(pTHX_ SV *sv, int c)
 {
 STATIC void
 S_put_byte(pTHX_ SV *sv, int c)
 {
-    if (isCNTRL(c) || c == 255 || !isPRINT(c))
+    PERL_ARGS_ASSERT_PUT_BYTE;
+
+    /* Our definition of isPRINT() ignores locales, so only bytes that are
+       not part of UTF-8 are considered printable. I assume that the same
+       holds for UTF-EBCDIC.
+       Also, code point 255 is not printable in either (it's E0 in EBCDIC,
+       which Wikipedia says:
+
+       EO, or Eight Ones, is an 8-bit EBCDIC character code represented as all
+       ones (binary 1111 1111, hexadecimal FF). It is similar, but not
+       identical, to the ASCII delete (DEL) or rubout control character.
+       ) So the old condition can be simplified to !isPRINT(c)  */
+    if (!isPRINT(c))
        Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
        Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
-    else if (c == '-' || c == ']' || c == '\\' || c == '^')
-       Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
-    else
-       Perl_sv_catpvf(aTHX_ sv, "%c", c);
+    else {
+       const char string = c;
+       if (c == '-' || c == ']' || c == '\\' || c == '^')
+           sv_catpvs(sv, "\\");
+       sv_catpvn(sv, &string, 1);
+    }
 }
 
 
 #define CLEAR_OPTSTART \
     if (optstart) STMT_START { \
 }
 
 
 #define CLEAR_OPTSTART \
     if (optstart) STMT_START { \
-           DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, " (%d nodes)\n", node - optstart)); \
+           DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, " (%"IVdf" nodes)\n", (IV)(node - optstart))); \
            optstart=NULL; \
     } STMT_END
 
            optstart=NULL; \
     } STMT_END
 
@@ -8968,8 +9938,12 @@ S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
     register U8 op = PSEUDO;   /* Arbitrary non-END op. */
     register const regnode *next;
     const regnode *optstart= NULL;
     register U8 op = PSEUDO;   /* Arbitrary non-END op. */
     register const regnode *next;
     const regnode *optstart= NULL;
+    
+    RXi_GET_DECL(r,ri);
     GET_RE_DEBUG_FLAGS_DECL;
 
     GET_RE_DEBUG_FLAGS_DECL;
 
+    PERL_ARGS_ASSERT_DUMPUNTIL;
+
 #ifdef DEBUG_DUMPUNTIL
     PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
         last ? last-start : 0,plast ? plast-start : 0);
 #ifdef DEBUG_DUMPUNTIL
     PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
         last ? last-start : 0,plast ? plast-start : 0);
@@ -8980,13 +9954,12 @@ S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
 
     while (PL_regkind[op] != END && (!last || node < last)) {
        /* While that wasn't END last time... */
 
     while (PL_regkind[op] != END && (!last || node < last)) {
        /* While that wasn't END last time... */
-
        NODE_ALIGN(node);
        op = OP(node);
        NODE_ALIGN(node);
        op = OP(node);
-       if (op == CLOSE)
+       if (op == CLOSE || op == WHILEM)
            indent--;
        next = regnext((regnode *)node);
            indent--;
        next = regnext((regnode *)node);
-       
+
        /* Where, what. */
        if (OP(node) == OPTIMIZED) {
            if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
        /* Where, what. */
        if (OP(node) == OPTIMIZED) {
            if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
@@ -8995,23 +9968,21 @@ S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
                goto after_print;
        } else
            CLEAR_OPTSTART;
                goto after_print;
        } else
            CLEAR_OPTSTART;
-           
+       
        regprop(r, sv, node);
        PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
                      (int)(2*indent + 1), "", SvPVX_const(sv));
        regprop(r, sv, node);
        PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
                      (int)(2*indent + 1), "", SvPVX_const(sv));
-
-       if (OP(node) != OPTIMIZED) {
-           if (next == NULL)           /* Next ptr. */
-               PerlIO_printf(Perl_debug_log, "(0)");
-           else if (PL_regkind[(U8)op] == BRANCH && PL_regkind[OP(next)] != BRANCH )
-               PerlIO_printf(Perl_debug_log, "(FAIL)");
-           else
-               PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
-               
-           /*if (PL_regkind[(U8)op]  != TRIE)*/
-               (void)PerlIO_putc(Perl_debug_log, '\n');
-       }
-
+        
+        if (OP(node) != OPTIMIZED) {                 
+            if (next == NULL)          /* Next ptr. */
+                PerlIO_printf(Perl_debug_log, " (0)");
+            else if (PL_regkind[(U8)op] == BRANCH && PL_regkind[OP(next)] != BRANCH )
+                PerlIO_printf(Perl_debug_log, " (FAIL)");
+            else 
+                PerlIO_printf(Perl_debug_log, " (%"IVdf")", (IV)(next - start));
+            (void)PerlIO_putc(Perl_debug_log, '\n'); 
+        }
+        
       after_print:
        if (PL_regkind[(U8)op] == BRANCHJ) {
            assert(next);
       after_print:
        if (PL_regkind[(U8)op] == BRANCHJ) {
            assert(next);
@@ -9031,33 +10002,35 @@ S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
        else if ( PL_regkind[(U8)op]  == TRIE ) {
            const regnode *this_trie = node;
            const char op = OP(node);
        else if ( PL_regkind[(U8)op]  == TRIE ) {
            const regnode *this_trie = node;
            const char op = OP(node);
-            const I32 n = ARG(node);
+            const U32 n = ARG(node);
            const reg_ac_data * const ac = op>=AHOCORASICK ?
            const reg_ac_data * const ac = op>=AHOCORASICK ?
-               (reg_ac_data *)r->data->data[n] :
+               (reg_ac_data *)ri->data->data[n] :
                NULL;
                NULL;
-           const reg_trie_data * const trie = op<AHOCORASICK ?
-               (reg_trie_data*)r->data->data[n] :
-               ac->trie;
+           const reg_trie_data * const trie =
+               (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
+#ifdef DEBUGGING
+           AV *const trie_words = MUTABLE_AV(ri->data->data[n + TRIE_WORDS_OFFSET]);
+#endif
            const regnode *nextbranch= NULL;
            I32 word_idx;
            const regnode *nextbranch= NULL;
            I32 word_idx;
-            sv_setpvn(sv, "", 0);
+            sv_setpvs(sv, "");
            for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
            for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
-               SV ** const elem_ptr = av_fetch(trie->words,word_idx,0);
+               SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
                
                 PerlIO_printf(Perl_debug_log, "%*s%s ",
                    (int)(2*(indent+3)), "",
                     elem_ptr ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr), SvCUR(*elem_ptr), 60,
                            PL_colors[0], PL_colors[1],
                            (SvUTF8(*elem_ptr) ? PERL_PV_ESCAPE_UNI : 0) |
                
                 PerlIO_printf(Perl_debug_log, "%*s%s ",
                    (int)(2*(indent+3)), "",
                     elem_ptr ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr), SvCUR(*elem_ptr), 60,
                            PL_colors[0], PL_colors[1],
                            (SvUTF8(*elem_ptr) ? PERL_PV_ESCAPE_UNI : 0) |
-                           PERL_PV_PRETTY_ELIPSES    |
+                           PERL_PV_PRETTY_ELLIPSES    |
                            PERL_PV_PRETTY_LTGT
                             )
                             : "???"
                 );
                 if (trie->jump) {
                     U16 dist= trie->jump[word_idx+1];
                            PERL_PV_PRETTY_LTGT
                             )
                             : "???"
                 );
                 if (trie->jump) {
                     U16 dist= trie->jump[word_idx+1];
-                   PerlIO_printf(Perl_debug_log, "(%u)\n",
-                        (dist ? this_trie + dist : next) - start);
+                   PerlIO_printf(Perl_debug_log, "(%"UVuf")\n",
+                                 (UV)((dist ? this_trie + dist : next) - start));
                     if (dist) {
                         if (!nextbranch)
                             nextbranch= this_trie + trie->jump[0];    
                     if (dist) {
                         if (!nextbranch)
                             nextbranch= this_trie + trie->jump[0];    
@@ -9102,12 +10075,10 @@ S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
        }
        if (op == CURLYX || op == OPEN)
            indent++;
        }
        if (op == CURLYX || op == OPEN)
            indent++;
-       else if (op == WHILEM)
-           indent--;
     }
     CLEAR_OPTSTART;
 #ifdef DEBUG_DUMPUNTIL    
     }
     CLEAR_OPTSTART;
 #ifdef DEBUG_DUMPUNTIL    
-    PerlIO_printf(Perl_debug_log, "--- %d\n",indent);
+    PerlIO_printf(Perl_debug_log, "--- %d\n", (int)indent);
 #endif
     return node;
 }
 #endif
     return node;
 }