This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
fix cx_dup for CXt_LOOP_PLAIN
[perl5.git] / pod / perlguts.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
954c1994 3perlguts - Introduction to the Perl API
a0d0e21e
LW
4
5=head1 DESCRIPTION
6
b3b6085d 7This document attempts to describe how to use the Perl API, as well as
10e2eb10
FC
8to provide some info on the basic workings of the Perl core. It is far
9from complete and probably contains many errors. Please refer any
b3b6085d 10questions or comments to the author below.
a0d0e21e 11
0a753a76
PP
12=head1 Variables
13
5f05dabc 14=head2 Datatypes
a0d0e21e
LW
15
16Perl has three typedefs that handle Perl's three main data types:
17
18 SV Scalar Value
19 AV Array Value
20 HV Hash Value
21
d1b91892 22Each typedef has specific routines that manipulate the various data types.
a0d0e21e
LW
23
24=head2 What is an "IV"?
25
954c1994 26Perl uses a special typedef IV which is a simple signed integer type that is
5f05dabc 27guaranteed to be large enough to hold a pointer (as well as an integer).
954c1994 28Additionally, there is the UV, which is simply an unsigned IV.
a0d0e21e 29
d1b91892 30Perl also uses two special typedefs, I32 and I16, which will always be at
10e2eb10 31least 32-bits and 16-bits long, respectively. (Again, there are U32 and U16,
20dbd849
NC
32as well.) They will usually be exactly 32 and 16 bits long, but on Crays
33they will both be 64 bits.
a0d0e21e 34
54310121 35=head2 Working with SVs
a0d0e21e 36
20dbd849
NC
37An SV can be created and loaded with one command. There are five types of
38values that can be loaded: an integer value (IV), an unsigned integer
39value (UV), a double (NV), a string (PV), and another scalar (SV).
61984ee1
KW
40("PV" stands for "Pointer Value". You might think that it is misnamed
41because it is described as pointing only to strings. However, it is
3ee1a09c 42possible to have it point to other things. For example, it could point
d6605d24 43to an array of UVs. But,
61984ee1
KW
44using it for non-strings requires care, as the underlying assumption of
45much of the internals is that PVs are just for strings. Often, for
6602b933 46example, a trailing C<NUL> is tacked on automatically. The non-string use
61984ee1 47is documented only in this paragraph.)
a0d0e21e 48
20dbd849 49The seven routines are:
a0d0e21e
LW
50
51 SV* newSViv(IV);
20dbd849 52 SV* newSVuv(UV);
a0d0e21e 53 SV* newSVnv(double);
06f6df17
RGS
54 SV* newSVpv(const char*, STRLEN);
55 SV* newSVpvn(const char*, STRLEN);
46fc3d4c 56 SV* newSVpvf(const char*, ...);
a0d0e21e
LW
57 SV* newSVsv(SV*);
58
06f6df17
RGS
59C<STRLEN> is an integer type (Size_t, usually defined as size_t in
60F<config.h>) guaranteed to be large enough to represent the size of
61any string that perl can handle.
62
3bf17896 63In the unlikely case of a SV requiring more complex initialization, you
06f6df17
RGS
64can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
65type NULL is returned, else an SV of type PV is returned with len + 1 (for
6602b933 66the C<NUL>) bytes of storage allocated, accessible via SvPVX. In both cases
da8c5729 67the SV has the undef value.
20dbd849 68
06f6df17 69 SV *sv = newSV(0); /* no storage allocated */
a9b0660e
KW
70 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage
71 * allocated */
20dbd849 72
06f6df17 73To change the value of an I<already-existing> SV, there are eight routines:
a0d0e21e
LW
74
75 void sv_setiv(SV*, IV);
deb3007b 76 void sv_setuv(SV*, UV);
a0d0e21e 77 void sv_setnv(SV*, double);
08105a92 78 void sv_setpv(SV*, const char*);
06f6df17 79 void sv_setpvn(SV*, const char*, STRLEN)
46fc3d4c 80 void sv_setpvf(SV*, const char*, ...);
a9b0660e
KW
81 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
82 SV **, I32, bool *);
a0d0e21e
LW
83 void sv_setsv(SV*, SV*);
84
85Notice that you can choose to specify the length of the string to be
9da1e3b5
MUN
86assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
87allow Perl to calculate the length by using C<sv_setpv> or by specifying
880 as the second argument to C<newSVpv>. Be warned, though, that Perl will
89determine the string's length by using C<strlen>, which depends on the
6602b933 90string terminating with a C<NUL> character, and not otherwise containing
a9b0660e 91NULs.
9abd00ed
GS
92
93The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
94formatted output becomes the value.
95
328bf373 96C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
9abd00ed
GS
97either a pointer to a variable argument list or the address and length of
98an array of SVs. The last argument points to a boolean; on return, if that
99boolean is true, then locale-specific information has been used to format
c2611fb3 100the string, and the string's contents are therefore untrustworthy (see
9abd00ed
GS
101L<perlsec>). This pointer may be NULL if that information is not
102important. Note that this function requires you to specify the length of
103the format.
104
9da1e3b5
MUN
105The C<sv_set*()> functions are not generic enough to operate on values
106that have "magic". See L<Magic Virtual Tables> later in this document.
a0d0e21e 107
6602b933
KW
108All SVs that contain strings should be terminated with a C<NUL> character.
109If it is not C<NUL>-terminated there is a risk of
5f05dabc 110core dumps and corruptions from code which passes the string to C
6602b933
KW
111functions or system calls which expect a C<NUL>-terminated string.
112Perl's own functions typically add a trailing C<NUL> for this reason.
5f05dabc
PP
113Nevertheless, you should be very careful when you pass a string stored
114in an SV to a C function or system call.
115
a0d0e21e
LW
116To access the actual value that an SV points to, you can use the macros:
117
118 SvIV(SV*)
954c1994 119 SvUV(SV*)
a0d0e21e
LW
120 SvNV(SV*)
121 SvPV(SV*, STRLEN len)
1fa8b10d 122 SvPV_nolen(SV*)
a0d0e21e 123
954c1994 124which will automatically coerce the actual scalar type into an IV, UV, double,
a0d0e21e
LW
125or string.
126
127In the C<SvPV> macro, the length of the string returned is placed into the
1fa8b10d
JD
128variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
129not care what the length of the data is, use the C<SvPV_nolen> macro.
130Historically the C<SvPV> macro with the global variable C<PL_na> has been
131used in this case. But that can be quite inefficient because C<PL_na> must
132be accessed in thread-local storage in threaded Perl. In any case, remember
133that Perl allows arbitrary strings of data that may both contain NULs and
6602b933 134might not be terminated by a C<NUL>.
a0d0e21e 135
ce2f5d8f 136Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
10e2eb10
FC
137len);>. It might work with your
138compiler, but it won't work for everyone.
ce2f5d8f
KA
139Break this sort of statement up into separate assignments:
140
1aa6ea50
JC
141 SV *s;
142 STRLEN len;
61955433 143 char *ptr;
1aa6ea50
JC
144 ptr = SvPV(s, len);
145 foo(ptr, len);
ce2f5d8f 146
07fa94a1 147If you want to know if the scalar value is TRUE, you can use:
a0d0e21e
LW
148
149 SvTRUE(SV*)
150
151Although Perl will automatically grow strings for you, if you need to force
152Perl to allocate more memory for your SV, you can use the macro
153
154 SvGROW(SV*, STRLEN newlen)
155
156which will determine if more memory needs to be allocated. If so, it will
157call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
5f05dabc 158decrease, the allocated memory of an SV and that it does not automatically
6602b933 159add space for the trailing C<NUL> byte (perl's own string functions typically do
8ebc5c01 160C<SvGROW(sv, len + 1)>).
a0d0e21e 161
21134f66
TC
162If you want to write to an existing SV's buffer and set its value to a
163string, use SvPV_force() or one of its variants to force the SV to be
164a PV. This will remove any of various types of non-stringness from
165the SV while preserving the content of the SV in the PV. This can be
166used, for example, to append data from an API function to a buffer
167without extra copying:
168
169 (void)SvPVbyte_force(sv, len);
170 s = SvGROW(sv, len + needlen + 1);
171 /* something that modifies up to needlen bytes at s+len, but
172 modifies newlen bytes
173 eg. newlen = read(fd, s + len, needlen);
174 ignoring errors for these examples
175 */
176 s[len + newlen] = '\0';
177 SvCUR_set(sv, len + newlen);
178 SvUTF8_off(sv);
179 SvSETMAGIC(sv);
180
181If you already have the data in memory or if you want to keep your
182code simple, you can use one of the sv_cat*() variants, such as
183sv_catpvn(). If you want to insert anywhere in the string you can use
184sv_insert() or sv_insert_flags().
185
186If you don't need the existing content of the SV, you can avoid some
187copying with:
188
189 sv_setpvn(sv, "", 0);
190 s = SvGROW(sv, needlen + 1);
191 /* something that modifies up to needlen bytes at s, but modifies
192 newlen bytes
193 eg. newlen = read(fd, s. needlen);
194 */
195 s[newlen] = '\0';
196 SvCUR_set(sv, newlen);
197 SvPOK_only(sv); /* also clears SVf_UTF8 */
198 SvSETMAGIC(sv);
199
200Again, if you already have the data in memory or want to avoid the
201complexity of the above, you can use sv_setpvn().
202
203If you have a buffer allocated with Newx() and want to set that as the
204SV's value, you can use sv_usepvn_flags(). That has some requirements
205if you want to avoid perl re-allocating the buffer to fit the trailing
206NUL:
207
208 Newx(buf, somesize+1, char);
209 /* ... fill in buf ... */
210 buf[somesize] = '\0';
211 sv_usepvn_flags(sv, buf, somesize, SV_SMAGIC | SV_HAS_TRAILING_NUL);
212 /* buf now belongs to perl, don't release it */
213
a0d0e21e
LW
214If you have an SV and want to know what kind of data Perl thinks is stored
215in it, you can use the following macros to check the type of SV you have.
216
217 SvIOK(SV*)
218 SvNOK(SV*)
219 SvPOK(SV*)
220
221You can get and set the current length of the string stored in an SV with
222the following macros:
223
224 SvCUR(SV*)
225 SvCUR_set(SV*, I32 val)
226
cb1a09d0
AD
227You can also get a pointer to the end of the string stored in the SV
228with the macro:
229
230 SvEND(SV*)
231
232But note that these last three macros are valid only if C<SvPOK()> is true.
a0d0e21e 233
d1b91892
AD
234If you want to append something to the end of string stored in an C<SV*>,
235you can use the following functions:
236
08105a92 237 void sv_catpv(SV*, const char*);
e65f3abd 238 void sv_catpvn(SV*, const char*, STRLEN);
46fc3d4c 239 void sv_catpvf(SV*, const char*, ...);
a9b0660e
KW
240 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
241 I32, bool);
d1b91892
AD
242 void sv_catsv(SV*, SV*);
243
244The first function calculates the length of the string to be appended by
245using C<strlen>. In the second, you specify the length of the string
46fc3d4c 246yourself. The third function processes its arguments like C<sprintf> and
9abd00ed
GS
247appends the formatted output. The fourth function works like C<vsprintf>.
248You can specify the address and length of an array of SVs instead of the
10e2eb10
FC
249va_list argument. The fifth function
250extends the string stored in the first
9abd00ed
GS
251SV with the string stored in the second SV. It also forces the second SV
252to be interpreted as a string.
253
254The C<sv_cat*()> functions are not generic enough to operate on values that
255have "magic". See L<Magic Virtual Tables> later in this document.
d1b91892 256
a0d0e21e
LW
257If you know the name of a scalar variable, you can get a pointer to its SV
258by using the following:
259
64ace3f8 260 SV* get_sv("package::varname", 0);
a0d0e21e
LW
261
262This returns NULL if the variable does not exist.
263
d1b91892 264If you want to know if this variable (or any other SV) is actually C<defined>,
a0d0e21e
LW
265you can call:
266
267 SvOK(SV*)
268
06f6df17 269The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
9adebda4 270
10e2eb10
FC
271Its address can be used whenever an C<SV*> is needed. Make sure that
272you don't try to compare a random sv with C<&PL_sv_undef>. For example
9adebda4
SB
273when interfacing Perl code, it'll work correctly for:
274
275 foo(undef);
276
277But won't work when called as:
278
279 $x = undef;
280 foo($x);
281
282So to repeat always use SvOK() to check whether an sv is defined.
283
284Also you have to be careful when using C<&PL_sv_undef> as a value in
285AVs or HVs (see L<AVs, HVs and undefined values>).
a0d0e21e 286
06f6df17
RGS
287There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
288boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
289addresses can be used whenever an C<SV*> is needed.
a0d0e21e 290
9cde0e7f 291Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
a0d0e21e
LW
292Take this code:
293
294 SV* sv = (SV*) 0;
295 if (I-am-to-return-a-real-value) {
296 sv = sv_2mortal(newSViv(42));
297 }
298 sv_setsv(ST(0), sv);
299
300This code tries to return a new SV (which contains the value 42) if it should
04343c6d 301return a real value, or undef otherwise. Instead it has returned a NULL
a0d0e21e 302pointer which, somewhere down the line, will cause a segmentation violation,
06f6df17
RGS
303bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
304first line and all will be well.
a0d0e21e
LW
305
306To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
3fe9a6f1 307call is not necessary (see L<Reference Counts and Mortality>).
a0d0e21e 308
94dde4fb
SC
309=head2 Offsets
310
311Perl provides the function C<sv_chop> to efficiently remove characters
312from the beginning of a string; you give it an SV and a pointer to
da75cd15 313somewhere inside the PV, and it discards everything before the
10e2eb10 314pointer. The efficiency comes by means of a little hack: instead of
94dde4fb
SC
315actually removing the characters, C<sv_chop> sets the flag C<OOK>
316(offset OK) to signal to other functions that the offset hack is in
883bb8c0
KW
317effect, and it moves the PV pointer (called C<SvPVX>) forward
318by the number of bytes chopped off, and adjusts C<SvCUR> and C<SvLEN>
319accordingly. (A portion of the space between the old and new PV
320pointers is used to store the count of chopped bytes.)
94dde4fb
SC
321
322Hence, at this point, the start of the buffer that we allocated lives
323at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
324into the middle of this allocated storage.
325
f942a0df
FC
326This is best demonstrated by example. Normally copy-on-write will prevent
327the substitution from operator from using this hack, but if you can craft a
328string for which copy-on-write is not possible, you can see it in play. In
329the current implementation, the final byte of a string buffer is used as a
330copy-on-write reference count. If the buffer is not big enough, then
331copy-on-write is skipped. First have a look at an empty string:
332
333 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a .= ""; Dump $a'
334 SV = PV(0x7ffb7c008a70) at 0x7ffb7c030390
335 REFCNT = 1
336 FLAGS = (POK,pPOK)
337 PV = 0x7ffb7bc05b50 ""\0
338 CUR = 0
339 LEN = 10
340
341Notice here the LEN is 10. (It may differ on your platform.) Extend the
342length of the string to one less than 10, and do a substitution:
94dde4fb 343
f942a0df
FC
344 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a.="123456789"; $a=~s/.//; Dump($a)'
345 SV = PV(0x7ffa04008a70) at 0x7ffa04030390
94dde4fb
SC
346 REFCNT = 1
347 FLAGS = (POK,OOK,pPOK)
f942a0df
FC
348 OFFSET = 1
349 PV = 0x7ffa03c05b61 ( "\1" . ) "23456789"\0
350 CUR = 8
351 LEN = 9
94dde4fb 352
f942a0df 353Here the number of bytes chopped off (1) is shown next as the OFFSET. The
94dde4fb
SC
354portion of the string between the "real" and the "fake" beginnings is
355shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
f942a0df
FC
356the fake beginning, not the real one. (The first character of the string
357buffer happens to have changed to "\1" here, not "1", because the current
358implementation stores the offset count in the string buffer. This is
359subject to change.)
94dde4fb 360
fe854a6f 361Something similar to the offset hack is performed on AVs to enable
319cef53
SC
362efficient shifting and splicing off the beginning of the array; while
363C<AvARRAY> points to the first element in the array that is visible from
10e2eb10 364Perl, C<AvALLOC> points to the real start of the C array. These are
319cef53 365usually the same, but a C<shift> operation can be carried out by
6de131f0 366increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
319cef53 367Again, the location of the real start of the C array only comes into
10e2eb10 368play when freeing the array. See C<av_shift> in F<av.c>.
319cef53 369
d1b91892 370=head2 What's Really Stored in an SV?
a0d0e21e
LW
371
372Recall that the usual method of determining the type of scalar you have is
5f05dabc 373to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 374usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
375macros will do the appropriate conversion of string to integer/double or
376integer/double to string.
377
378If you I<really> need to know if you have an integer, double, or string
379pointer in an SV, you can use the following three macros instead:
380
381 SvIOKp(SV*)
382 SvNOKp(SV*)
383 SvPOKp(SV*)
384
385These will tell you if you truly have an integer, double, or string pointer
d1b91892 386stored in your SV. The "p" stands for private.
a0d0e21e 387
da8c5729 388There are various ways in which the private and public flags may differ.
9090718a
FC
389For example, in perl 5.16 and earlier a tied SV may have a valid
390underlying value in the IV slot (so SvIOKp is true), but the data
391should be accessed via the FETCH routine rather than directly,
392so SvIOK is false. (In perl 5.18 onwards, tied scalars use
393the flags the same way as untied scalars.) Another is when
d7f8936a 394numeric conversion has occurred and precision has been lost: only the
10e2eb10 395private flag is set on 'lossy' values. So when an NV is converted to an
9e9796d6
JH
396IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
397
07fa94a1 398In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 399
54310121 400=head2 Working with AVs
a0d0e21e 401
07fa94a1
JO
402There are two ways to create and load an AV. The first method creates an
403empty AV:
a0d0e21e
LW
404
405 AV* newAV();
406
54310121 407The second method both creates the AV and initially populates it with SVs:
a0d0e21e 408
c70927a6 409 AV* av_make(SSize_t num, SV **ptr);
a0d0e21e 410
5f05dabc 411The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 412AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 413
da8c5729 414Once the AV has been created, the following operations are possible on it:
a0d0e21e
LW
415
416 void av_push(AV*, SV*);
417 SV* av_pop(AV*);
418 SV* av_shift(AV*);
c70927a6 419 void av_unshift(AV*, SSize_t num);
a0d0e21e
LW
420
421These should be familiar operations, with the exception of C<av_unshift>.
422This routine adds C<num> elements at the front of the array with the C<undef>
423value. You must then use C<av_store> (described below) to assign values
424to these new elements.
425
426Here are some other functions:
427
c70927a6
FC
428 SSize_t av_top_index(AV*);
429 SV** av_fetch(AV*, SSize_t key, I32 lval);
430 SV** av_store(AV*, SSize_t key, SV* val);
a0d0e21e 431
dab460cd 432The C<av_top_index> function returns the highest index value in an array (just
5f05dabc
PP
433like $#array in Perl). If the array is empty, -1 is returned. The
434C<av_fetch> function returns the value at index C<key>, but if C<lval>
435is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
436The C<av_store> function stores the value C<val> at index C<key>, and does
437not increment the reference count of C<val>. Thus the caller is responsible
438for taking care of that, and if C<av_store> returns NULL, the caller will
439have to decrement the reference count to avoid a memory leak. Note that
440C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
441return value.
d1b91892 442
da8c5729
MH
443A few more:
444
a0d0e21e 445 void av_clear(AV*);
a0d0e21e 446 void av_undef(AV*);
c70927a6 447 void av_extend(AV*, SSize_t key);
5f05dabc
PP
448
449The C<av_clear> function deletes all the elements in the AV* array, but
450does not actually delete the array itself. The C<av_undef> function will
451delete all the elements in the array plus the array itself. The
adc882cf
GS
452C<av_extend> function extends the array so that it contains at least C<key+1>
453elements. If C<key+1> is less than the currently allocated length of the array,
454then nothing is done.
a0d0e21e
LW
455
456If you know the name of an array variable, you can get a pointer to its AV
457by using the following:
458
cbfd0a87 459 AV* get_av("package::varname", 0);
a0d0e21e
LW
460
461This returns NULL if the variable does not exist.
462
04343c6d
GS
463See L<Understanding the Magic of Tied Hashes and Arrays> for more
464information on how to use the array access functions on tied arrays.
465
54310121 466=head2 Working with HVs
a0d0e21e
LW
467
468To create an HV, you use the following routine:
469
470 HV* newHV();
471
da8c5729 472Once the HV has been created, the following operations are possible on it:
a0d0e21e 473
08105a92
GS
474 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
475 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 476
5f05dabc
PP
477The C<klen> parameter is the length of the key being passed in (Note that
478you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
479length of the key). The C<val> argument contains the SV pointer to the
54310121 480scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
481you want C<hv_store> to calculate it for you). The C<lval> parameter
482indicates whether this fetch is actually a part of a store operation, in
483which case a new undefined value will be added to the HV with the supplied
484key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 485
5f05dabc
PP
486Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
487C<SV*>. To access the scalar value, you must first dereference the return
488value. However, you should check to make sure that the return value is
489not NULL before dereferencing it.
a0d0e21e 490
da8c5729
MH
491The first of these two functions checks if a hash table entry exists, and the
492second deletes it.
a0d0e21e 493
08105a92
GS
494 bool hv_exists(HV*, const char* key, U32 klen);
495 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 496
5f05dabc
PP
497If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
498create and return a mortal copy of the deleted value.
499
a0d0e21e
LW
500And more miscellaneous functions:
501
502 void hv_clear(HV*);
a0d0e21e 503 void hv_undef(HV*);
5f05dabc
PP
504
505Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
506table but does not actually delete the hash table. The C<hv_undef> deletes
507both the entries and the hash table itself.
a0d0e21e 508
a9b0660e 509Perl keeps the actual data in a linked list of structures with a typedef of HE.
d1b91892
AD
510These contain the actual key and value pointers (plus extra administrative
511overhead). The key is a string pointer; the value is an C<SV*>. However,
512once you have an C<HE*>, to get the actual key and value, use the routines
513specified below.
514
a0d0e21e
LW
515 I32 hv_iterinit(HV*);
516 /* Prepares starting point to traverse hash table */
517 HE* hv_iternext(HV*);
518 /* Get the next entry, and return a pointer to a
519 structure that has both the key and value */
520 char* hv_iterkey(HE* entry, I32* retlen);
521 /* Get the key from an HE structure and also return
522 the length of the key string */
cb1a09d0 523 SV* hv_iterval(HV*, HE* entry);
d1be9408 524 /* Return an SV pointer to the value of the HE
a0d0e21e 525 structure */
cb1a09d0 526 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
527 /* This convenience routine combines hv_iternext,
528 hv_iterkey, and hv_iterval. The key and retlen
529 arguments are return values for the key and its
530 length. The value is returned in the SV* argument */
a0d0e21e
LW
531
532If you know the name of a hash variable, you can get a pointer to its HV
533by using the following:
534
6673a63c 535 HV* get_hv("package::varname", 0);
a0d0e21e
LW
536
537This returns NULL if the variable does not exist.
538
a43e7901 539The hash algorithm is defined in the C<PERL_HASH> macro:
a0d0e21e 540
a43e7901 541 PERL_HASH(hash, key, klen)
ab192400 542
a43e7901
YO
543The exact implementation of this macro varies by architecture and version
544of perl, and the return value may change per invocation, so the value
545is only valid for the duration of a single perl process.
a0d0e21e 546
04343c6d
GS
547See L<Understanding the Magic of Tied Hashes and Arrays> for more
548information on how to use the hash access functions on tied hashes.
549
1e422769
PP
550=head2 Hash API Extensions
551
552Beginning with version 5.004, the following functions are also supported:
553
554 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
555 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 556
1e422769
PP
557 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
558 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 559
1e422769
PP
560 SV* hv_iterkeysv (HE* entry);
561
562Note that these functions take C<SV*> keys, which simplifies writing
563of extension code that deals with hash structures. These functions
564also allow passing of C<SV*> keys to C<tie> functions without forcing
565you to stringify the keys (unlike the previous set of functions).
566
567They also return and accept whole hash entries (C<HE*>), making their
568use more efficient (since the hash number for a particular string
4a4eefd0
GS
569doesn't have to be recomputed every time). See L<perlapi> for detailed
570descriptions.
1e422769
PP
571
572The following macros must always be used to access the contents of hash
573entries. Note that the arguments to these macros must be simple
574variables, since they may get evaluated more than once. See
4a4eefd0 575L<perlapi> for detailed descriptions of these macros.
1e422769
PP
576
577 HePV(HE* he, STRLEN len)
578 HeVAL(HE* he)
579 HeHASH(HE* he)
580 HeSVKEY(HE* he)
581 HeSVKEY_force(HE* he)
582 HeSVKEY_set(HE* he, SV* sv)
583
584These two lower level macros are defined, but must only be used when
585dealing with keys that are not C<SV*>s:
586
587 HeKEY(HE* he)
588 HeKLEN(HE* he)
589
04343c6d
GS
590Note that both C<hv_store> and C<hv_store_ent> do not increment the
591reference count of the stored C<val>, which is the caller's responsibility.
592If these functions return a NULL value, the caller will usually have to
593decrement the reference count of C<val> to avoid a memory leak.
1e422769 594
a9381218
MHM
595=head2 AVs, HVs and undefined values
596
10e2eb10
FC
597Sometimes you have to store undefined values in AVs or HVs. Although
598this may be a rare case, it can be tricky. That's because you're
a9381218
MHM
599used to using C<&PL_sv_undef> if you need an undefined SV.
600
601For example, intuition tells you that this XS code:
602
603 AV *av = newAV();
604 av_store( av, 0, &PL_sv_undef );
605
606is equivalent to this Perl code:
607
608 my @av;
609 $av[0] = undef;
610
f3c4ec28 611Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
a9381218
MHM
612for indicating that an array element has not yet been initialized.
613Thus, C<exists $av[0]> would be true for the above Perl code, but
f3c4ec28
FC
614false for the array generated by the XS code. In perl 5.20, storing
615&PL_sv_undef will create a read-only element, because the scalar
616&PL_sv_undef itself is stored, not a copy.
a9381218 617
f3c4ec28 618Similar problems can occur when storing C<&PL_sv_undef> in HVs:
a9381218
MHM
619
620 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
621
622This will indeed make the value C<undef>, but if you try to modify
623the value of C<key>, you'll get the following error:
624
625 Modification of non-creatable hash value attempted
626
627In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
10e2eb10 628in restricted hashes. This caused such hash entries not to appear
a9381218
MHM
629when iterating over the hash or when checking for the keys
630with the C<hv_exists> function.
631
8abccac8 632You can run into similar problems when you store C<&PL_sv_yes> or
10e2eb10 633C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
634will give you the following error:
635
636 Modification of a read-only value attempted
637
638To make a long story short, you can use the special variables
8abccac8 639C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
640HVs, but you have to make sure you know what you're doing.
641
642Generally, if you want to store an undefined value in an AV
643or HV, you should not use C<&PL_sv_undef>, but rather create a
644new undefined value using the C<newSV> function, for example:
645
646 av_store( av, 42, newSV(0) );
647 hv_store( hv, "foo", 3, newSV(0), 0 );
648
a0d0e21e
LW
649=head2 References
650
d1b91892 651References are a special type of scalar that point to other data types
a9b0660e 652(including other references).
a0d0e21e 653
07fa94a1 654To create a reference, use either of the following functions:
a0d0e21e 655
5f05dabc
PP
656 SV* newRV_inc((SV*) thing);
657 SV* newRV_noinc((SV*) thing);
a0d0e21e 658
5f05dabc 659The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
660functions are identical except that C<newRV_inc> increments the reference
661count of the C<thing>, while C<newRV_noinc> does not. For historical
662reasons, C<newRV> is a synonym for C<newRV_inc>.
663
664Once you have a reference, you can use the following macro to dereference
665the reference:
a0d0e21e
LW
666
667 SvRV(SV*)
668
669then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 670C<AV*> or C<HV*>, if required.
a0d0e21e 671
d1b91892 672To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
673
674 SvROK(SV*)
675
07fa94a1
JO
676To discover what type of value the reference refers to, use the following
677macro and then check the return value.
d1b91892
AD
678
679 SvTYPE(SvRV(SV*))
680
681The most useful types that will be returned are:
682
a5e62da0
FC
683 < SVt_PVAV Scalar
684 SVt_PVAV Array
685 SVt_PVHV Hash
686 SVt_PVCV Code
687 SVt_PVGV Glob (possibly a file handle)
688
689See L<perlapi/svtype> for more details.
d1b91892 690
cb1a09d0
AD
691=head2 Blessed References and Class Objects
692
06f6df17 693References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
694OO lexicon, an object is simply a reference that has been blessed into a
695package (or class). Once blessed, the programmer may now use the reference
696to access the various methods in the class.
697
698A reference can be blessed into a package with the following function:
699
700 SV* sv_bless(SV* sv, HV* stash);
701
06f6df17
RGS
702The C<sv> argument must be a reference value. The C<stash> argument
703specifies which class the reference will belong to. See
2ae324a7 704L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
705
706/* Still under construction */
707
ddd2cc91
DM
708The following function upgrades rv to reference if not already one.
709Creates a new SV for rv to point to. If C<classname> is non-null, the SV
710is blessed into the specified class. SV is returned.
cb1a09d0 711
08105a92 712 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 713
ddd2cc91
DM
714The following three functions copy integer, unsigned integer or double
715into an SV whose reference is C<rv>. SV is blessed if C<classname> is
716non-null.
cb1a09d0 717
08105a92 718 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 719 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 720 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 721
ddd2cc91
DM
722The following function copies the pointer value (I<the address, not the
723string!>) into an SV whose reference is rv. SV is blessed if C<classname>
724is non-null.
cb1a09d0 725
ddd2cc91 726 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
cb1a09d0 727
a9b0660e 728The following function copies a string into an SV whose reference is C<rv>.
ddd2cc91
DM
729Set length to 0 to let Perl calculate the string length. SV is blessed if
730C<classname> is non-null.
cb1a09d0 731
a9b0660e
KW
732 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
733 STRLEN length);
cb1a09d0 734
ddd2cc91
DM
735The following function tests whether the SV is blessed into the specified
736class. It does not check inheritance relationships.
9abd00ed 737
08105a92 738 int sv_isa(SV* sv, const char* name);
9abd00ed 739
ddd2cc91 740The following function tests whether the SV is a reference to a blessed object.
9abd00ed
GS
741
742 int sv_isobject(SV* sv);
743
ddd2cc91 744The following function tests whether the SV is derived from the specified
10e2eb10
FC
745class. SV can be either a reference to a blessed object or a string
746containing a class name. This is the function implementing the
ddd2cc91 747C<UNIVERSAL::isa> functionality.
9abd00ed 748
08105a92 749 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 750
00aadd71 751To check if you've got an object derived from a specific class you have
9abd00ed
GS
752to write:
753
754 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 755
5f05dabc 756=head2 Creating New Variables
cb1a09d0 757
5f05dabc
PP
758To create a new Perl variable with an undef value which can be accessed from
759your Perl script, use the following routines, depending on the variable type.
cb1a09d0 760
64ace3f8 761 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 762 AV* get_av("package::varname", GV_ADD);
6673a63c 763 HV* get_hv("package::varname", GV_ADD);
cb1a09d0 764
058a5f6c 765Notice the use of GV_ADD as the second parameter. The new variable can now
cb1a09d0
AD
766be set, using the routines appropriate to the data type.
767
5f05dabc 768There are additional macros whose values may be bitwise OR'ed with the
058a5f6c 769C<GV_ADD> argument to enable certain extra features. Those bits are:
cb1a09d0 770
9a68f1db
SB
771=over
772
773=item GV_ADDMULTI
774
775Marks the variable as multiply defined, thus preventing the:
776
777 Name <varname> used only once: possible typo
778
779warning.
780
9a68f1db
SB
781=item GV_ADDWARN
782
783Issues the warning:
784
785 Had to create <varname> unexpectedly
786
787if the variable did not exist before the function was called.
788
789=back
cb1a09d0 790
07fa94a1
JO
791If you do not specify a package name, the variable is created in the current
792package.
cb1a09d0 793
5f05dabc 794=head2 Reference Counts and Mortality
a0d0e21e 795
10e2eb10 796Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 797AVs, or HVs (xV for short in the following) start their life with a
55497cff 798reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 799then it will be destroyed and its memory made available for reuse.
55497cff
PP
800
801This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
802undef'ed or the last variable holding a reference to it is changed or
803overwritten. At the internal level, however, reference counts can be
55497cff
PP
804manipulated with the following macros:
805
806 int SvREFCNT(SV* sv);
5f05dabc 807 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
808 void SvREFCNT_dec(SV* sv);
809
810However, there is one other function which manipulates the reference
07fa94a1
JO
811count of its argument. The C<newRV_inc> function, you will recall,
812creates a reference to the specified argument. As a side effect,
813it increments the argument's reference count. If this is not what
814you want, use C<newRV_noinc> instead.
815
816For example, imagine you want to return a reference from an XSUB function.
817Inside the XSUB routine, you create an SV which initially has a reference
818count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
819This returns the reference as a new SV, but the reference count of the
820SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
821return the reference from the XSUB routine and forget about the SV.
822But Perl hasn't! Whenever the returned reference is destroyed, the
823reference count of the original SV is decreased to one and nothing happens.
824The SV will hang around without any way to access it until Perl itself
825terminates. This is a memory leak.
5f05dabc
PP
826
827The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
828C<newRV_inc>. Then, if and when the last reference is destroyed,
829the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 830stopping any memory leak.
55497cff 831
5f05dabc 832There are some convenience functions available that can help with the
54310121 833destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
834An xV that is mortal has had its reference count marked to be decremented,
835but not actually decremented, until "a short time later". Generally the
836term "short time later" means a single Perl statement, such as a call to
54310121 837an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
838reference count decremented depends on two macros, SAVETMPS and FREETMPS.
839See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
840
841"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
842However, if you mortalize a variable twice, the reference count will
843later be decremented twice.
844
00aadd71
NIS
845"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
846For example an SV which is created just to pass a number to a called sub
06f6df17 847is made mortal to have it cleaned up automatically when it's popped off
10e2eb10 848the stack. Similarly, results returned by XSUBs (which are pushed on the
06f6df17 849stack) are often made mortal.
a0d0e21e
LW
850
851To create a mortal variable, use the functions:
852
853 SV* sv_newmortal()
854 SV* sv_2mortal(SV*)
855 SV* sv_mortalcopy(SV*)
856
00aadd71 857The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
858SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
859third creates a mortal copy of an existing SV.
da8c5729 860Because C<sv_newmortal> gives the new SV no value, it must normally be given one
9a68f1db 861via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
862
863 SV *tmp = sv_newmortal();
864 sv_setiv(tmp, an_integer);
865
866As that is multiple C statements it is quite common so see this idiom instead:
867
868 SV *tmp = sv_2mortal(newSViv(an_integer));
869
870
871You should be careful about creating mortal variables. Strange things
872can happen if you make the same value mortal within multiple contexts,
10e2eb10
FC
873or if you make a variable mortal multiple
874times. Thinking of "Mortalization"
00aadd71 875as deferred C<SvREFCNT_dec> should help to minimize such problems.
da8c5729 876For example if you are passing an SV which you I<know> has a high enough REFCNT
00aadd71
NIS
877to survive its use on the stack you need not do any mortalization.
878If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
879making a C<sv_mortalcopy> is safer.
a0d0e21e 880
ac036724 881The mortal routines are not just for SVs; AVs and HVs can be
faed5253 882made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 883C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 884
5f05dabc 885=head2 Stashes and Globs
a0d0e21e 886
06f6df17
RGS
887A B<stash> is a hash that contains all variables that are defined
888within a package. Each key of the stash is a symbol
aa689395
PP
889name (shared by all the different types of objects that have the same
890name), and each value in the hash table is a GV (Glob Value). This GV
891in turn contains references to the various objects of that name,
892including (but not limited to) the following:
cb1a09d0 893
a0d0e21e
LW
894 Scalar Value
895 Array Value
896 Hash Value
a3cb178b 897 I/O Handle
a0d0e21e
LW
898 Format
899 Subroutine
900
06f6df17
RGS
901There is a single stash called C<PL_defstash> that holds the items that exist
902in the C<main> package. To get at the items in other packages, append the
903string "::" to the package name. The items in the C<Foo> package are in
904the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
905in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 906
d1b91892 907To get the stash pointer for a particular package, use the function:
a0d0e21e 908
da51bb9b
NC
909 HV* gv_stashpv(const char* name, I32 flags)
910 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
911
912The first function takes a literal string, the second uses the string stored
d1b91892 913in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 914C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
915
916The name that C<gv_stash*v> wants is the name of the package whose symbol table
917you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
918packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
919language itself.
a0d0e21e
LW
920
921Alternately, if you have an SV that is a blessed reference, you can find
922out the stash pointer by using:
923
924 HV* SvSTASH(SvRV(SV*));
925
926then use the following to get the package name itself:
927
928 char* HvNAME(HV* stash);
929
5f05dabc
PP
930If you need to bless or re-bless an object you can use the following
931function:
a0d0e21e
LW
932
933 SV* sv_bless(SV*, HV* stash)
934
935where the first argument, an C<SV*>, must be a reference, and the second
936argument is a stash. The returned C<SV*> can now be used in the same way
937as any other SV.
938
d1b91892
AD
939For more information on references and blessings, consult L<perlref>.
940
54310121 941=head2 Double-Typed SVs
0a753a76
PP
942
943Scalar variables normally contain only one type of value, an integer,
944double, pointer, or reference. Perl will automatically convert the
945actual scalar data from the stored type into the requested type.
946
947Some scalar variables contain more than one type of scalar data. For
948example, the variable C<$!> contains either the numeric value of C<errno>
949or its string equivalent from either C<strerror> or C<sys_errlist[]>.
950
951To force multiple data values into an SV, you must do two things: use the
952C<sv_set*v> routines to add the additional scalar type, then set a flag
953so that Perl will believe it contains more than one type of data. The
954four macros to set the flags are:
955
956 SvIOK_on
957 SvNOK_on
958 SvPOK_on
959 SvROK_on
960
961The particular macro you must use depends on which C<sv_set*v> routine
962you called first. This is because every C<sv_set*v> routine turns on
963only the bit for the particular type of data being set, and turns off
964all the rest.
965
966For example, to create a new Perl variable called "dberror" that contains
967both the numeric and descriptive string error values, you could use the
968following code:
969
970 extern int dberror;
971 extern char *dberror_list;
972
64ace3f8 973 SV* sv = get_sv("dberror", GV_ADD);
0a753a76
PP
974 sv_setiv(sv, (IV) dberror);
975 sv_setpv(sv, dberror_list[dberror]);
976 SvIOK_on(sv);
977
978If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
979macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
980
4f4531b8
FC
981=head2 Read-Only Values
982
983In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
984flag bit with read-only scalars. So the only way to test whether
985C<sv_setsv>, etc., will raise a "Modification of a read-only value" error
986in those versions is:
987
988 SvREADONLY(sv) && !SvIsCOW(sv)
989
990Under Perl 5.18 and later, SvREADONLY only applies to read-only variables,
991and, under 5.20, copy-on-write scalars can also be read-only, so the above
992check is incorrect. You just want:
993
994 SvREADONLY(sv)
995
996If you need to do this check often, define your own macro like this:
997
998 #if PERL_VERSION >= 18
999 # define SvTRULYREADONLY(sv) SvREADONLY(sv)
1000 #else
1001 # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
1002 #endif
1003
1004=head2 Copy on Write
1005
1006Perl implements a copy-on-write (COW) mechanism for scalars, in which
1007string copies are not immediately made when requested, but are deferred
1008until made necessary by one or the other scalar changing. This is mostly
1009transparent, but one must take care not to modify string buffers that are
1010shared by multiple SVs.
1011
1012You can test whether an SV is using copy-on-write with C<SvIsCOW(sv)>.
1013
1014You can force an SV to make its own copy of its string buffer by calling C<sv_force_normal(sv)> or SvPV_force_nolen(sv).
1015
1016If you want to make the SV drop its string buffer, use
1017C<sv_force_normal_flags(sv, SV_COW_DROP_PV)> or simply
1018C<sv_setsv(sv, NULL)>.
1019
1020All of these functions will croak on read-only scalars (see the previous
1021section for more on those).
1022
1023To test that your code is behaving correctly and not modifying COW buffers,
1024on systems that support L<mmap(2)> (i.e., Unix) you can configure perl with
1025C<-Accflags=-DPERL_DEBUG_READONLY_COW> and it will turn buffer violations
1026into crashes. You will find it to be marvellously slow, so you may want to
1027skip perl's own tests.
1028
0a753a76 1029=head2 Magic Variables
a0d0e21e 1030
d1b91892
AD
1031[This section still under construction. Ignore everything here. Post no
1032bills. Everything not permitted is forbidden.]
1033
d1b91892
AD
1034Any SV may be magical, that is, it has special features that a normal
1035SV does not have. These features are stored in the SV structure in a
5f05dabc 1036linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
1037
1038 struct magic {
1039 MAGIC* mg_moremagic;
1040 MGVTBL* mg_virtual;
1041 U16 mg_private;
1042 char mg_type;
1043 U8 mg_flags;
b205eb13 1044 I32 mg_len;
d1b91892
AD
1045 SV* mg_obj;
1046 char* mg_ptr;
d1b91892
AD
1047 };
1048
1049Note this is current as of patchlevel 0, and could change at any time.
1050
1051=head2 Assigning Magic
1052
1053Perl adds magic to an SV using the sv_magic function:
1054
a9b0660e 1055 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
1056
1057The C<sv> argument is a pointer to the SV that is to acquire a new magical
1058feature.
1059
1060If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
10e2eb10
FC
1061convert C<sv> to type C<SVt_PVMG>.
1062Perl then continues by adding new magic
645c22ef
DM
1063to the beginning of the linked list of magical features. Any prior entry
1064of the same type of magic is deleted. Note that this can be overridden,
1065and multiple instances of the same type of magic can be associated with an
1066SV.
d1b91892 1067
54310121 1068The C<name> and C<namlen> arguments are used to associate a string with
10e2eb10 1069the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
1070C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
1071C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
1072whether C<namlen> is greater than zero or equal to zero respectively. As a
1073special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
1074to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
1075
1076The sv_magic function uses C<how> to determine which, if any, predefined
1077"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 1078See the L<Magic Virtual Tables> section below. The C<how> argument is also
10e2eb10
FC
1079stored in the C<mg_type> field. The value of
1080C<how> should be chosen from the set of macros
1081C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 1082these macros were added, Perl internals used to directly use character
14befaf4 1083literals, so you may occasionally come across old code or documentation
75d0f26d 1084referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
1085
1086The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
1087structure. If it is not the same as the C<sv> argument, the reference
1088count of the C<obj> object is incremented. If it is the same, or if
645c22ef 1089the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 1090then C<obj> is merely stored, without the reference count being incremented.
d1b91892 1091
2d8d5d5a
SH
1092See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
1093to an SV.
1094
cb1a09d0
AD
1095There is also a function to add magic to an C<HV>:
1096
1097 void hv_magic(HV *hv, GV *gv, int how);
1098
1099This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
1100
1101To remove the magic from an SV, call the function sv_unmagic:
1102
70a53b35 1103 int sv_unmagic(SV *sv, int type);
cb1a09d0
AD
1104
1105The C<type> argument should be equal to the C<how> value when the C<SV>
1106was initially made magical.
1107
f6ee7b17 1108However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
10e2eb10
FC
1109C<SV>. If you want to remove only certain
1110magic of a C<type> based on the magic
f6ee7b17
FR
1111virtual table, use C<sv_unmagicext> instead:
1112
1113 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1114
d1b91892
AD
1115=head2 Magic Virtual Tables
1116
d1be9408 1117The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
1118C<MGVTBL>, which is a structure of function pointers and stands for
1119"Magic Virtual Table" to handle the various operations that might be
1120applied to that variable.
1121
301cb7e8
DM
1122The C<MGVTBL> has five (or sometimes eight) pointers to the following
1123routine types:
d1b91892
AD
1124
1125 int (*svt_get)(SV* sv, MAGIC* mg);
1126 int (*svt_set)(SV* sv, MAGIC* mg);
1127 U32 (*svt_len)(SV* sv, MAGIC* mg);
1128 int (*svt_clear)(SV* sv, MAGIC* mg);
1129 int (*svt_free)(SV* sv, MAGIC* mg);
1130
a9b0660e
KW
1131 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1132 const char *name, I32 namlen);
301cb7e8
DM
1133 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1134 int (*svt_local)(SV *nsv, MAGIC *mg);
1135
1136
06f6df17 1137This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
SM
1138currently 32 types. These different structures contain pointers to various
1139routines that perform additional actions depending on which function is
1140being called.
d1b91892 1141
a9b0660e
KW
1142 Function pointer Action taken
1143 ---------------- ------------
1144 svt_get Do something before the value of the SV is
1145 retrieved.
1146 svt_set Do something after the SV is assigned a value.
1147 svt_len Report on the SV's length.
1148 svt_clear Clear something the SV represents.
1149 svt_free Free any extra storage associated with the SV.
d1b91892 1150
a9b0660e
KW
1151 svt_copy copy tied variable magic to a tied element
1152 svt_dup duplicate a magic structure during thread cloning
1153 svt_local copy magic to local value during 'local'
301cb7e8 1154
d1b91892 1155For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1156to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1157
1158 { magic_get, magic_set, magic_len, 0, 0 }
1159
14befaf4
DM
1160Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1161if a get operation is being performed, the routine C<magic_get> is
1162called. All the various routines for the various magical types begin
1163with C<magic_>. NOTE: the magic routines are not considered part of
1164the Perl API, and may not be exported by the Perl library.
d1b91892 1165
301cb7e8
DM
1166The last three slots are a recent addition, and for source code
1167compatibility they are only checked for if one of the three flags
10e2eb10
FC
1168MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1169This means that most code can continue declaring
1170a vtable as a 5-element value. These three are
301cb7e8
DM
1171currently used exclusively by the threading code, and are highly subject
1172to change.
1173
d1b91892
AD
1174The current kinds of Magic Virtual Tables are:
1175
f1f5ddd7
FC
1176=for comment
1177This table is generated by regen/mg_vtable.pl. Any changes made here
1178will be lost.
1179
1180=for mg_vtable.pl begin
1181
a9b0660e 1182 mg_type
bd6e6c12
FC
1183 (old-style char and macro) MGVTBL Type of magic
1184 -------------------------- ------ -------------
1185 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1186 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
e5e1ee61 1187 % PERL_MAGIC_rhash (none) Extra data for restricted
bd6e6c12 1188 hashes
a6d69523
TC
1189 * PERL_MAGIC_debugvar vtbl_debugvar $DB::single, signal, trace
1190 vars
bd6e6c12 1191 . PERL_MAGIC_pos vtbl_pos pos() lvalue
e5e1ee61 1192 : PERL_MAGIC_symtab (none) Extra data for symbol
bd6e6c12 1193 tables
e5e1ee61
FC
1194 < PERL_MAGIC_backref vtbl_backref For weak ref data
1195 @ PERL_MAGIC_arylen_p (none) To move arylen out of XPVAV
bd6e6c12
FC
1196 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1197 (fast string search)
1198 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1199 (AMT) on stash
1200 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1201 (@+ and @- vars)
1202 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1203 element
1204 E PERL_MAGIC_env vtbl_env %ENV hash
1205 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
eccba044 1206 f PERL_MAGIC_fm vtbl_regexp Formline
bd6e6c12 1207 ('compiled' format)
bd6e6c12
FC
1208 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1209 H PERL_MAGIC_hints vtbl_hints %^H hash
1210 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1211 I PERL_MAGIC_isa vtbl_isa @ISA array
1212 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1213 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1214 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1215 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1216 element
1217 N PERL_MAGIC_shared (none) Shared between threads
1218 n PERL_MAGIC_shared_scalar (none) Shared between threads
1219 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1220 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1221 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1222 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
e5e1ee61 1223 r PERL_MAGIC_qr vtbl_regexp Precompiled qr// regex
bd6e6c12
FC
1224 S PERL_MAGIC_sig (none) %SIG hash
1225 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1226 t PERL_MAGIC_taint vtbl_taint Taintedness
1227 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1228 extensions
1229 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1230 extensions
4499db73 1231 V PERL_MAGIC_vstring (none) SV was vstring literal
bd6e6c12
FC
1232 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1233 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1234 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1235 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1236 variable / smart parameter
1237 vivification
baabe3fb
FC
1238 \ PERL_MAGIC_lvref vtbl_lvref Lvalue reference
1239 constructor
e5e1ee61 1240 ] PERL_MAGIC_checkcall vtbl_checkcall Inlining/mutation of call
bd6e6c12
FC
1241 to this CV
1242 ~ PERL_MAGIC_ext (none) Available for use by
1243 extensions
0cbee0a4 1244
f1f5ddd7 1245=for mg_vtable.pl end
d1b91892 1246
68dc0745 1247When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1248uppercase letter is typically used to represent some kind of composite type
1249(a list or a hash), and the lowercase letter is used to represent an element
10e2eb10 1250of that composite type. Some internals code makes use of this case
92f0c265 1251relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1252
1253The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1254specifically for use by extensions and will not be used by perl itself.
1255Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1256to variables (typically objects). This is especially useful because
1257there is no way for normal perl code to corrupt this private information
1258(unlike using extra elements of a hash object).
1259
1260Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1261C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1262C<mg_ptr> field points to a C<ufuncs> structure:
1263
1264 struct ufuncs {
a9402793
AB
1265 I32 (*uf_val)(pTHX_ IV, SV*);
1266 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1267 IV uf_index;
1268 };
1269
1270When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1271function will be called with C<uf_index> as the first arg and a pointer to
1272the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1273magic is shown below. Note that the ufuncs structure is copied by
1274sv_magic, so you can safely allocate it on the stack.
1275
1276 void
1277 Umagic(sv)
1278 SV *sv;
1279 PREINIT:
1280 struct ufuncs uf;
1281 CODE:
1282 uf.uf_val = &my_get_fn;
1283 uf.uf_set = &my_set_fn;
1284 uf.uf_index = 0;
14befaf4 1285 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1286
1e73acc8
AS
1287Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1288
1289For hashes there is a specialized hook that gives control over hash
1290keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1291if the "set" function in the C<ufuncs> structure is NULL. The hook
1292is activated whenever the hash is accessed with a key specified as
1293an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1294C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1295through the functions without the C<..._ent> suffix circumvents the
4509d391 1296hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1e73acc8 1297
14befaf4
DM
1298Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1299or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1300extra care to avoid conflict. Typically only using the magic on
1301objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1302For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1303C<MGVTBL>, even if all its fields will be C<0>, so that individual
1304C<MAGIC> pointers can be identified as a particular kind of magic
10e2eb10 1305using their magic virtual table. C<mg_findext> provides an easy way
f6ee7b17 1306to do that:
2f07f21a
FR
1307
1308 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1309
1310 MAGIC *mg;
f6ee7b17
FR
1311 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1312 /* this is really ours, not another module's PERL_MAGIC_ext */
1313 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1314 ...
2f07f21a 1315 }
5f05dabc 1316
ef50df4b
GS
1317Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1318earlier do B<not> invoke 'set' magic on their targets. This must
1319be done by the user either by calling the C<SvSETMAGIC()> macro after
1320calling these functions, or by using one of the C<sv_set*_mg()> or
1321C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1322C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1323obtained from external sources in functions that don't handle magic.
4a4eefd0 1324See L<perlapi> for a description of these functions.
189b2af5
GS
1325For example, calls to the C<sv_cat*()> functions typically need to be
1326followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1327since their implementation handles 'get' magic.
1328
d1b91892
AD
1329=head2 Finding Magic
1330
a9b0660e
KW
1331 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1332 * type */
f6ee7b17
FR
1333
1334This routine returns a pointer to a C<MAGIC> structure stored in the SV.
10e2eb10
FC
1335If the SV does not have that magical
1336feature, C<NULL> is returned. If the
f6ee7b17 1337SV has multiple instances of that magical feature, the first one will be
10e2eb10
FC
1338returned. C<mg_findext> can be used
1339to find a C<MAGIC> structure of an SV
da8c5729 1340based on both its magic type and its magic virtual table:
f6ee7b17
FR
1341
1342 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
d1b91892 1343
f6ee7b17
FR
1344Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1345SVt_PVMG, Perl may core dump.
d1b91892 1346
08105a92 1347 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1348
1349This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1350field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1351the mg_type field is changed to be the lowercase letter.
a0d0e21e 1352
04343c6d
GS
1353=head2 Understanding the Magic of Tied Hashes and Arrays
1354
14befaf4
DM
1355Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1356magic type.
9edb2b46
GS
1357
1358WARNING: As of the 5.004 release, proper usage of the array and hash
1359access functions requires understanding a few caveats. Some
1360of these caveats are actually considered bugs in the API, to be fixed
10e2eb10 1361in later releases, and are bracketed with [MAYCHANGE] below. If
9edb2b46
GS
1362you find yourself actually applying such information in this section, be
1363aware that the behavior may change in the future, umm, without warning.
04343c6d 1364
1526ead6 1365The perl tie function associates a variable with an object that implements
9a68f1db 1366the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6 1367tie function from an XSUB, you must mimic this behaviour. The code below
61ad4b94 1368carries out the necessary steps -- firstly it creates a new hash, and then
1526ead6 1369creates a second hash which it blesses into the class which will implement
10e2eb10 1370the tie methods. Lastly it ties the two hashes together, and returns a
1526ead6
AB
1371reference to the new tied hash. Note that the code below does NOT call the
1372TIEHASH method in the MyTie class -
1373see L<Calling Perl Routines from within C Programs> for details on how
1374to do this.
1375
1376 SV*
1377 mytie()
1378 PREINIT:
1379 HV *hash;
1380 HV *stash;
1381 SV *tie;
1382 CODE:
1383 hash = newHV();
1384 tie = newRV_noinc((SV*)newHV());
da51bb9b 1385 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1386 sv_bless(tie, stash);
899e16d0 1387 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1388 RETVAL = newRV_noinc(hash);
1389 OUTPUT:
1390 RETVAL
1391
04343c6d
GS
1392The C<av_store> function, when given a tied array argument, merely
1393copies the magic of the array onto the value to be "stored", using
1394C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1395actually need to be stored in the array. [MAYCHANGE] After a call to
1396C<av_store> on a tied array, the caller will usually need to call
1397C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1398TIEARRAY object. If C<av_store> did return NULL, a call to
1399C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1400leak. [/MAYCHANGE]
04343c6d
GS
1401
1402The previous paragraph is applicable verbatim to tied hash access using the
1403C<hv_store> and C<hv_store_ent> functions as well.
1404
1405C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1406C<hv_fetch_ent> actually return an undefined mortal value whose magic
1407has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1408need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1409need to call C<mg_get()> on the returned value in order to actually invoke
1410the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1411you may also call C<mg_set()> on the return value after possibly assigning
1412a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1413method on the TIE object. [/MAYCHANGE]
04343c6d 1414
9edb2b46 1415[MAYCHANGE]
04343c6d
GS
1416In other words, the array or hash fetch/store functions don't really
1417fetch and store actual values in the case of tied arrays and hashes. They
1418merely call C<mg_copy> to attach magic to the values that were meant to be
1419"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1420do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1421the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1422and hashes.
1423
1424Currently (as of perl version 5.004), use of the hash and array access
1425functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1426"normal" hashes and arrays, or on their tied variants. The API may be
1427changed to provide more transparent access to both tied and normal data
1428types in future versions.
1429[/MAYCHANGE]
04343c6d
GS
1430
1431You would do well to understand that the TIEARRAY and TIEHASH interfaces
1432are mere sugar to invoke some perl method calls while using the uniform hash
1433and array syntax. The use of this sugar imposes some overhead (typically
1434about two to four extra opcodes per FETCH/STORE operation, in addition to
1435the creation of all the mortal variables required to invoke the methods).
1436This overhead will be comparatively small if the TIE methods are themselves
1437substantial, but if they are only a few statements long, the overhead
1438will not be insignificant.
1439
d1c897a1
IZ
1440=head2 Localizing changes
1441
1442Perl has a very handy construction
1443
1444 {
1445 local $var = 2;
1446 ...
1447 }
1448
1449This construction is I<approximately> equivalent to
1450
1451 {
1452 my $oldvar = $var;
1453 $var = 2;
1454 ...
1455 $var = $oldvar;
1456 }
1457
1458The biggest difference is that the first construction would
1459reinstate the initial value of $var, irrespective of how control exits
10e2eb10 1460the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1461more efficient as well.
1462
1463There is a way to achieve a similar task from C via Perl API: create a
1464I<pseudo-block>, and arrange for some changes to be automatically
1465undone at the end of it, either explicit, or via a non-local exit (via
10e2eb10 1466die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1467C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1468Such a construct may be created specially for some important localized
1469task, or an existing one (like boundaries of enclosing Perl
1470subroutine/block, or an existing pair for freeing TMPs) may be
10e2eb10
FC
1471used. (In the second case the overhead of additional localization must
1472be almost negligible.) Note that any XSUB is automatically enclosed in
b687b08b 1473an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1474
1475Inside such a I<pseudo-block> the following service is available:
1476
13a2d996 1477=over 4
d1c897a1
IZ
1478
1479=item C<SAVEINT(int i)>
1480
1481=item C<SAVEIV(IV i)>
1482
1483=item C<SAVEI32(I32 i)>
1484
1485=item C<SAVELONG(long i)>
1486
1487These macros arrange things to restore the value of integer variable
1488C<i> at the end of enclosing I<pseudo-block>.
1489
1490=item C<SAVESPTR(s)>
1491
1492=item C<SAVEPPTR(p)>
1493
1494These macros arrange things to restore the value of pointers C<s> and
10e2eb10 1495C<p>. C<s> must be a pointer of a type which survives conversion to
d1c897a1
IZ
1496C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1497and back.
1498
1499=item C<SAVEFREESV(SV *sv)>
1500
1501The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1502I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1503mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1504extends the lifetime of C<sv> until the beginning of the next statement,
1505C<SAVEFREESV> extends it until the end of the enclosing scope. These
1506lifetimes can be wildly different.
1507
1508Also compare C<SAVEMORTALIZESV>.
1509
1510=item C<SAVEMORTALIZESV(SV *sv)>
1511
1512Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1513scope instead of decrementing its reference count. This usually has the
1514effect of keeping C<sv> alive until the statement that called the currently
1515live scope has finished executing.
d1c897a1
IZ
1516
1517=item C<SAVEFREEOP(OP *op)>
1518
1519The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1520
1521=item C<SAVEFREEPV(p)>
1522
1523The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1524end of I<pseudo-block>.
1525
1526=item C<SAVECLEARSV(SV *sv)>
1527
1528Clears a slot in the current scratchpad which corresponds to C<sv> at
1529the end of I<pseudo-block>.
1530
1531=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1532
10e2eb10 1533The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
d1c897a1
IZ
1534string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1535short-lived storage, the corresponding string may be reallocated like
1536this:
1537
9cde0e7f 1538 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1539
c76ac1ee 1540=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1541
1542At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1543only argument C<p>.
1544
1545=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1546
1547At the end of I<pseudo-block> the function C<f> is called with the
1548implicit context argument (if any), and C<p>.
d1c897a1
IZ
1549
1550=item C<SAVESTACK_POS()>
1551
1552The current offset on the Perl internal stack (cf. C<SP>) is restored
1553at the end of I<pseudo-block>.
1554
1555=back
1556
1557The following API list contains functions, thus one needs to
1558provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1559or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1560function takes C<int *>.
1561
13a2d996 1562=over 4
d1c897a1
IZ
1563
1564=item C<SV* save_scalar(GV *gv)>
1565
1566Equivalent to Perl code C<local $gv>.
1567
1568=item C<AV* save_ary(GV *gv)>
1569
1570=item C<HV* save_hash(GV *gv)>
1571
1572Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1573
1574=item C<void save_item(SV *item)>
1575
1576Duplicates the current value of C<SV>, on the exit from the current
1577C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
10e2eb10 1578using the stored value. It doesn't handle magic. Use C<save_scalar> if
038fcae3 1579magic is affected.
d1c897a1
IZ
1580
1581=item C<void save_list(SV **sarg, I32 maxsarg)>
1582
1583A variant of C<save_item> which takes multiple arguments via an array
1584C<sarg> of C<SV*> of length C<maxsarg>.
1585
1586=item C<SV* save_svref(SV **sptr)>
1587
d1be9408 1588Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1589
1590=item C<void save_aptr(AV **aptr)>
1591
1592=item C<void save_hptr(HV **hptr)>
1593
1594Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1595
1596=back
1597
1598The C<Alias> module implements localization of the basic types within the
1599I<caller's scope>. People who are interested in how to localize things in
1600the containing scope should take a look there too.
1601
0a753a76 1602=head1 Subroutines
a0d0e21e 1603
68dc0745 1604=head2 XSUBs and the Argument Stack
5f05dabc
PP
1605
1606The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1607An XSUB routine will have a stack that contains the arguments from the Perl
1608program, and a way to map from the Perl data structures to a C equivalent.
1609
1610The stack arguments are accessible through the C<ST(n)> macro, which returns
1611the C<n>'th stack argument. Argument 0 is the first argument passed in the
1612Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1613an C<SV*> is used.
1614
1615Most of the time, output from the C routine can be handled through use of
1616the RETVAL and OUTPUT directives. However, there are some cases where the
1617argument stack is not already long enough to handle all the return values.
1618An example is the POSIX tzname() call, which takes no arguments, but returns
1619two, the local time zone's standard and summer time abbreviations.
1620
1621To handle this situation, the PPCODE directive is used and the stack is
1622extended using the macro:
1623
924508f0 1624 EXTEND(SP, num);
5f05dabc 1625
924508f0
GS
1626where C<SP> is the macro that represents the local copy of the stack pointer,
1627and C<num> is the number of elements the stack should be extended by.
5f05dabc 1628
00aadd71 1629Now that there is room on the stack, values can be pushed on it using C<PUSHs>
10e2eb10 1630macro. The pushed values will often need to be "mortal" (See
d82b684c 1631L</Reference Counts and Mortality>):
5f05dabc 1632
00aadd71 1633 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1634 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1635 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1636 PUSHs(sv_2mortal(newSVpv("Some String",0)))
a9b0660e
KW
1637 /* Although the last example is better written as the more
1638 * efficient: */
a3179684 1639 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
5f05dabc
PP
1640
1641And now the Perl program calling C<tzname>, the two values will be assigned
1642as in:
1643
1644 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1645
1646An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1647to use the macro:
5f05dabc 1648
5f05dabc
PP
1649 XPUSHs(SV*)
1650
da8c5729 1651This macro automatically adjusts the stack for you, if needed. Thus, you
5f05dabc 1652do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1653
1654Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1655C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1656For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1657C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc
PP
1658
1659For more information, consult L<perlxs> and L<perlxstut>.
1660
5b36e945
FC
1661=head2 Autoloading with XSUBs
1662
1663If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1664fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1665of the XSUB's package.
1666
1667But it also puts the same information in certain fields of the XSUB itself:
1668
1669 HV *stash = CvSTASH(cv);
1670 const char *subname = SvPVX(cv);
1671 STRLEN name_length = SvCUR(cv); /* in bytes */
1672 U32 is_utf8 = SvUTF8(cv);
f703fc96 1673
5b36e945 1674C<SvPVX(cv)> contains just the sub name itself, not including the package.
d8893903
FC
1675For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1676C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
5b36e945
FC
1677
1678B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1679XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1680XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1681to support 5.8-5.14, use the XSUB's fields.
1682
5f05dabc 1683=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1684
1685There are four routines that can be used to call a Perl subroutine from
1686within a C program. These four are:
1687
954c1994
GS
1688 I32 call_sv(SV*, I32);
1689 I32 call_pv(const char*, I32);
1690 I32 call_method(const char*, I32);
5aaab254 1691 I32 call_argv(const char*, I32, char**);
a0d0e21e 1692
954c1994 1693The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1694contains either the name of the Perl subroutine to be called, or a
1695reference to the subroutine. The second argument consists of flags
1696that control the context in which the subroutine is called, whether
1697or not the subroutine is being passed arguments, how errors should be
1698trapped, and how to treat return values.
a0d0e21e
LW
1699
1700All four routines return the number of arguments that the subroutine returned
1701on the Perl stack.
1702
9a68f1db 1703These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1704but those names are now deprecated; macros of the same name are provided for
1705compatibility.
1706
1707When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1708must manipulate the Perl stack. These include the following macros and
1709functions:
a0d0e21e
LW
1710
1711 dSP
924508f0 1712 SP
a0d0e21e
LW
1713 PUSHMARK()
1714 PUTBACK
1715 SPAGAIN
1716 ENTER
1717 SAVETMPS
1718 FREETMPS
1719 LEAVE
1720 XPUSH*()
cb1a09d0 1721 POP*()
a0d0e21e 1722
5f05dabc
PP
1723For a detailed description of calling conventions from C to Perl,
1724consult L<perlcall>.
a0d0e21e 1725
8ebc5c01 1726=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1727
1728A lot of opcodes (this is an elementary operation in the internal perl
10e2eb10
FC
1729stack machine) put an SV* on the stack. However, as an optimization
1730the corresponding SV is (usually) not recreated each time. The opcodes
ce3d39e2
IZ
1731reuse specially assigned SVs (I<target>s) which are (as a corollary)
1732not constantly freed/created.
1733
0a753a76 1734Each of the targets is created only once (but see
ce3d39e2
IZ
1735L<Scratchpads and recursion> below), and when an opcode needs to put
1736an integer, a double, or a string on stack, it just sets the
1737corresponding parts of its I<target> and puts the I<target> on stack.
1738
1739The macro to put this target on stack is C<PUSHTARG>, and it is
1740directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1741others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1742
1bd1c0d5 1743Because the target is reused, you must be careful when pushing multiple
10e2eb10 1744values on the stack. The following code will not do what you think:
1bd1c0d5
SC
1745
1746 XPUSHi(10);
1747 XPUSHi(20);
1748
1749This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1750the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1751At the end of the operation, the stack does not contain the values 10
1752and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1753to 20.
1bd1c0d5 1754
d82b684c
SH
1755If you need to push multiple different values then you should either use
1756the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1757none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1758SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1759will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1760this a little easier to achieve by creating a new mortal for you (via
1761C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1762in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1763Thus, instead of writing this to "fix" the example above:
1764
1765 XPUSHs(sv_2mortal(newSViv(10)))
1766 XPUSHs(sv_2mortal(newSViv(20)))
1767
1768you can simply write:
1769
1770 mXPUSHi(10)
1771 mXPUSHi(20)
1772
1773On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1774need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1775macros can make use of the local variable C<TARG>. See also C<dTARGET>
1776and C<dXSTARG>.
1bd1c0d5 1777
8ebc5c01 1778=head2 Scratchpads
ce3d39e2 1779
54310121 1780The question remains on when the SVs which are I<target>s for opcodes
10e2eb10 1781are created. The answer is that they are created when the current
ac036724 1782unit--a subroutine or a file (for opcodes for statements outside of
10e2eb10 1783subroutines)--is compiled. During this time a special anonymous Perl
ac036724 1784array is created, which is called a scratchpad for the current unit.
ce3d39e2 1785
54310121 1786A scratchpad keeps SVs which are lexicals for the current unit and are
d777b41a
FC
1787targets for opcodes. A previous version of this document
1788stated that one can deduce that an SV lives on a scratchpad
ce3d39e2 1789by looking on its flags: lexicals have C<SVs_PADMY> set, and
eee3e302 1790I<target>s have C<SVs_PADTMP> set. But this has never been fully true.
d777b41a
FC
1791C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
1792While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
eed77337
FC
1793that have never resided in a pad, but nonetheless act like I<target>s. As
1794of perl 5.21.5, the C<SVs_PADMY> flag is no longer used and is defined as
17950. C<SvPADMY()> now returns true for anything without C<SVs_PADTMP>.
ce3d39e2 1796
10e2eb10 1797The correspondence between OPs and I<target>s is not 1-to-1. Different
54310121 1798OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1799would not conflict with the expected life of the temporary.
1800
2ae324a7 1801=head2 Scratchpads and recursion
ce3d39e2
IZ
1802
1803In fact it is not 100% true that a compiled unit contains a pointer to
10e2eb10
FC
1804the scratchpad AV. In fact it contains a pointer to an AV of
1805(initially) one element, and this element is the scratchpad AV. Why do
ce3d39e2
IZ
1806we need an extra level of indirection?
1807
10e2eb10 1808The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2 1809these can create several execution pointers going into the same
10e2eb10 1810subroutine. For the subroutine-child not write over the temporaries
ce3d39e2
IZ
1811for the subroutine-parent (lifespan of which covers the call to the
1812child), the parent and the child should have different
10e2eb10 1813scratchpads. (I<And> the lexicals should be separate anyway!)
ce3d39e2 1814
5f05dabc
PP
1815So each subroutine is born with an array of scratchpads (of length 1).
1816On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1817depth of the recursion is not more than the length of this array, and
1818if it is, new scratchpad is created and pushed into the array.
1819
1820The I<target>s on this scratchpad are C<undef>s, but they are already
1821marked with correct flags.
1822
22d36020
FC
1823=head1 Memory Allocation
1824
1825=head2 Allocation
1826
1827All memory meant to be used with the Perl API functions should be manipulated
1828using the macros described in this section. The macros provide the necessary
1829transparency between differences in the actual malloc implementation that is
1830used within perl.
1831
1832It is suggested that you enable the version of malloc that is distributed
1833with Perl. It keeps pools of various sizes of unallocated memory in
1834order to satisfy allocation requests more quickly. However, on some
1835platforms, it may cause spurious malloc or free errors.
1836
1837The following three macros are used to initially allocate memory :
1838
1839 Newx(pointer, number, type);
1840 Newxc(pointer, number, type, cast);
1841 Newxz(pointer, number, type);
1842
1843The first argument C<pointer> should be the name of a variable that will
1844point to the newly allocated memory.
1845
1846The second and third arguments C<number> and C<type> specify how many of
1847the specified type of data structure should be allocated. The argument
1848C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
1849should be used if the C<pointer> argument is different from the C<type>
1850argument.
1851
1852Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
1853to zero out all the newly allocated memory.
1854
1855=head2 Reallocation
1856
1857 Renew(pointer, number, type);
1858 Renewc(pointer, number, type, cast);
1859 Safefree(pointer)
1860
1861These three macros are used to change a memory buffer size or to free a
1862piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1863match those of C<New> and C<Newc> with the exception of not needing the
1864"magic cookie" argument.
1865
1866=head2 Moving
1867
1868 Move(source, dest, number, type);
1869 Copy(source, dest, number, type);
1870 Zero(dest, number, type);
1871
1872These three macros are used to move, copy, or zero out previously allocated
1873memory. The C<source> and C<dest> arguments point to the source and
1874destination starting points. Perl will move, copy, or zero out C<number>
1875instances of the size of the C<type> data structure (using the C<sizeof>
1876function).
1877
1878=head1 PerlIO
1879
1880The most recent development releases of Perl have been experimenting with
1881removing Perl's dependency on the "normal" standard I/O suite and allowing
1882other stdio implementations to be used. This involves creating a new
1883abstraction layer that then calls whichever implementation of stdio Perl
1884was compiled with. All XSUBs should now use the functions in the PerlIO
1885abstraction layer and not make any assumptions about what kind of stdio
1886is being used.
1887
1888For a complete description of the PerlIO abstraction, consult L<perlapio>.
1889
0a753a76
PP
1890=head1 Compiled code
1891
1892=head2 Code tree
1893
1894Here we describe the internal form your code is converted to by
10e2eb10 1895Perl. Start with a simple example:
0a753a76
PP
1896
1897 $a = $b + $c;
1898
1899This is converted to a tree similar to this one:
1900
1901 assign-to
1902 / \
1903 + $a
1904 / \
1905 $b $c
1906
7b8d334a 1907(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1908parsed your code, but has nothing to do with the execution order.
1909There is an additional "thread" going through the nodes of the tree
1910which shows the order of execution of the nodes. In our simplified
1911example above it looks like:
1912
1913 $b ---> $c ---> + ---> $a ---> assign-to
1914
1915But with the actual compile tree for C<$a = $b + $c> it is different:
1916some nodes I<optimized away>. As a corollary, though the actual tree
1917contains more nodes than our simplified example, the execution order
1918is the same as in our example.
1919
1920=head2 Examining the tree
1921
06f6df17
RGS
1922If you have your perl compiled for debugging (usually done with
1923C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1924compiled tree by specifying C<-Dx> on the Perl command line. The
1925output takes several lines per node, and for C<$b+$c> it looks like
1926this:
1927
1928 5 TYPE = add ===> 6
1929 TARG = 1
1930 FLAGS = (SCALAR,KIDS)
1931 {
1932 TYPE = null ===> (4)
1933 (was rv2sv)
1934 FLAGS = (SCALAR,KIDS)
1935 {
1936 3 TYPE = gvsv ===> 4
1937 FLAGS = (SCALAR)
1938 GV = main::b
1939 }
1940 }
1941 {
1942 TYPE = null ===> (5)
1943 (was rv2sv)
1944 FLAGS = (SCALAR,KIDS)
1945 {
1946 4 TYPE = gvsv ===> 5
1947 FLAGS = (SCALAR)
1948 GV = main::c
1949 }
1950 }
1951
1952This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1953not optimized away (one per number in the left column). The immediate
1954children of the given node correspond to C<{}> pairs on the same level
1955of indentation, thus this listing corresponds to the tree:
1956
1957 add
1958 / \
1959 null null
1960 | |
1961 gvsv gvsv
1962
1963The execution order is indicated by C<===E<gt>> marks, thus it is C<3
19644 5 6> (node C<6> is not included into above listing), i.e.,
1965C<gvsv gvsv add whatever>.
1966
9afa14e3 1967Each of these nodes represents an op, a fundamental operation inside the
10e2eb10 1968Perl core. The code which implements each operation can be found in the
9afa14e3 1969F<pp*.c> files; the function which implements the op with type C<gvsv>
10e2eb10 1970is C<pp_gvsv>, and so on. As the tree above shows, different ops have
9afa14e3 1971different numbers of children: C<add> is a binary operator, as one would
10e2eb10 1972expect, and so has two children. To accommodate the various different
9afa14e3
SC
1973numbers of children, there are various types of op data structure, and
1974they link together in different ways.
1975
10e2eb10 1976The simplest type of op structure is C<OP>: this has no children. Unary
9afa14e3 1977operators, C<UNOP>s, have one child, and this is pointed to by the
10e2eb10
FC
1978C<op_first> field. Binary operators (C<BINOP>s) have not only an
1979C<op_first> field but also an C<op_last> field. The most complex type of
1980op is a C<LISTOP>, which has any number of children. In this case, the
9afa14e3 1981first child is pointed to by C<op_first> and the last child by
10e2eb10 1982C<op_last>. The children in between can be found by iteratively
86cd3a13 1983following the C<OpSIBLING> pointer from the first child to the last (but
29e61fd9 1984see below).
9afa14e3 1985
29e61fd9 1986There are also some other op types: a C<PMOP> holds a regular expression,
10e2eb10
FC
1987and has no children, and a C<LOOP> may or may not have children. If the
1988C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
9afa14e3
SC
1989complicate matters, if a C<UNOP> is actually a C<null> op after
1990optimization (see L</Compile pass 2: context propagation>) it will still
1991have children in accordance with its former type.
1992
29e61fd9
DM
1993Finally, there is a C<LOGOP>, or logic op. Like a C<LISTOP>, this has one
1994or more children, but it doesn't have an C<op_last> field: so you have to
86cd3a13 1995follow C<op_first> and then the C<OpSIBLING> chain itself to find the
29e61fd9
DM
1996last child. Instead it has an C<op_other> field, which is comparable to
1997the C<op_next> field described below, and represents an alternate
1998execution path. Operators like C<and>, C<or> and C<?> are C<LOGOP>s. Note
1999that in general, C<op_other> may not point to any of the direct children
2000of the C<LOGOP>.
2001
2002Starting in version 5.21.2, perls built with the experimental
2003define C<-DPERL_OP_PARENT> add an extra boolean flag for each op,
87b5a8b9 2004C<op_moresib>. When not set, this indicates that this is the last op in an
86cd3a13
DM
2005C<OpSIBLING> chain. This frees up the C<op_sibling> field on the last
2006sibling to point back to the parent op. Under this build, that field is
2007also renamed C<op_sibparent> to reflect its joint role. The macro
2008C<OpSIBLING(o)> wraps this special behaviour, and always returns NULL on
2009the last sibling. With this build the C<op_parent(o)> function can be
2010used to find the parent of any op. Thus for forward compatibility, you
2011should always use the C<OpSIBLING(o)> macro rather than accessing
2012C<op_sibling> directly.
29e61fd9 2013
06f6df17
RGS
2014Another way to examine the tree is to use a compiler back-end module, such
2015as L<B::Concise>.
2016
0a753a76
PP
2017=head2 Compile pass 1: check routines
2018
8870b5c7 2019The tree is created by the compiler while I<yacc> code feeds it
10e2eb10 2020the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
2021the first pass of perl compilation.
2022
2023What makes this pass interesting for perl developers is that some
2024optimization may be performed on this pass. This is optimization by
8870b5c7 2025so-called "check routines". The correspondence between node names
0a753a76
PP
2026and corresponding check routines is described in F<opcode.pl> (do not
2027forget to run C<make regen_headers> if you modify this file).
2028
2029A check routine is called when the node is fully constructed except
7b8d334a 2030for the execution-order thread. Since at this time there are no
0a753a76
PP
2031back-links to the currently constructed node, one can do most any
2032operation to the top-level node, including freeing it and/or creating
2033new nodes above/below it.
2034
2035The check routine returns the node which should be inserted into the
2036tree (if the top-level node was not modified, check routine returns
2037its argument).
2038
10e2eb10 2039By convention, check routines have names C<ck_*>. They are usually
0a753a76
PP
2040called from C<new*OP> subroutines (or C<convert>) (which in turn are
2041called from F<perly.y>).
2042
2043=head2 Compile pass 1a: constant folding
2044
2045Immediately after the check routine is called the returned node is
2046checked for being compile-time executable. If it is (the value is
2047judged to be constant) it is immediately executed, and a I<constant>
2048node with the "return value" of the corresponding subtree is
2049substituted instead. The subtree is deleted.
2050
2051If constant folding was not performed, the execution-order thread is
2052created.
2053
2054=head2 Compile pass 2: context propagation
2055
2056When a context for a part of compile tree is known, it is propagated
a3cb178b 2057down through the tree. At this time the context can have 5 values
0a753a76
PP
2058(instead of 2 for runtime context): void, boolean, scalar, list, and
2059lvalue. In contrast with the pass 1 this pass is processed from top
2060to bottom: a node's context determines the context for its children.
2061
2062Additional context-dependent optimizations are performed at this time.
2063Since at this moment the compile tree contains back-references (via
2064"thread" pointers), nodes cannot be free()d now. To allow
2065optimized-away nodes at this stage, such nodes are null()ified instead
2066of free()ing (i.e. their type is changed to OP_NULL).
2067
2068=head2 Compile pass 3: peephole optimization
2069
2070After the compile tree for a subroutine (or for an C<eval> or a file)
10e2eb10 2071is created, an additional pass over the code is performed. This pass
0a753a76 2072is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
2073additional complications for conditionals). Optimizations performed
2074at this stage are subject to the same restrictions as in the pass 2.
2075
2076Peephole optimizations are done by calling the function pointed to
2077by the global variable C<PL_peepp>. By default, C<PL_peepp> just
2078calls the function pointed to by the global variable C<PL_rpeepp>.
2079By default, that performs some basic op fixups and optimisations along
2080the execution-order op chain, and recursively calls C<PL_rpeepp> for
2081each side chain of ops (resulting from conditionals). Extensions may
2082provide additional optimisations or fixups, hooking into either the
2083per-subroutine or recursive stage, like this:
2084
2085 static peep_t prev_peepp;
2086 static void my_peep(pTHX_ OP *o)
2087 {
2088 /* custom per-subroutine optimisation goes here */
f0358462 2089 prev_peepp(aTHX_ o);
9ea12537
Z
2090 /* custom per-subroutine optimisation may also go here */
2091 }
2092 BOOT:
2093 prev_peepp = PL_peepp;
2094 PL_peepp = my_peep;
2095
2096 static peep_t prev_rpeepp;
2097 static void my_rpeep(pTHX_ OP *o)
2098 {
2099 OP *orig_o = o;
2100 for(; o; o = o->op_next) {
2101 /* custom per-op optimisation goes here */
2102 }
f0358462 2103 prev_rpeepp(aTHX_ orig_o);
9ea12537
Z
2104 }
2105 BOOT:
2106 prev_rpeepp = PL_rpeepp;
2107 PL_rpeepp = my_rpeep;
0a753a76 2108
1ba7f851
PJ
2109=head2 Pluggable runops
2110
2111The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
2112functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
2113with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
2114control over the execution of the compile tree it is possible to provide
2115your own runops function.
1ba7f851
PJ
2116
2117It's probably best to copy one of the existing runops functions and
2118change it to suit your needs. Then, in the BOOT section of your XS
2119file, add the line:
2120
2121 PL_runops = my_runops;
2122
2123This function should be as efficient as possible to keep your programs
2124running as fast as possible.
2125
fd85fad2
BM
2126=head2 Compile-time scope hooks
2127
2128As of perl 5.14 it is possible to hook into the compile-time lexical
10e2eb10 2129scope mechanism using C<Perl_blockhook_register>. This is used like
fd85fad2
BM
2130this:
2131
2132 STATIC void my_start_hook(pTHX_ int full);
2133 STATIC BHK my_hooks;
2134
2135 BOOT:
a88d97bf 2136 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
2137 Perl_blockhook_register(aTHX_ &my_hooks);
2138
2139This will arrange to have C<my_start_hook> called at the start of
10e2eb10 2140compiling every lexical scope. The available hooks are:
fd85fad2
BM
2141
2142=over 4
2143
a88d97bf 2144=item C<void bhk_start(pTHX_ int full)>
fd85fad2 2145
10e2eb10 2146This is called just after starting a new lexical scope. Note that Perl
fd85fad2
BM
2147code like
2148
2149 if ($x) { ... }
2150
2151creates two scopes: the first starts at the C<(> and has C<full == 1>,
10e2eb10 2152the second starts at the C<{> and has C<full == 0>. Both end at the
f185f654 2153C<}>, so calls to C<start> and C<pre>/C<post_end> will match. Anything
fd85fad2
BM
2154pushed onto the save stack by this hook will be popped just before the
2155scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2156
a88d97bf 2157=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
2158
2159This is called at the end of a lexical scope, just before unwinding the
10e2eb10 2160stack. I<o> is the root of the optree representing the scope; it is a
fd85fad2
BM
2161double pointer so you can replace the OP if you need to.
2162
a88d97bf 2163=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
2164
2165This is called at the end of a lexical scope, just after unwinding the
10e2eb10 2166stack. I<o> is as above. Note that it is possible for calls to C<pre_>
fd85fad2
BM
2167and C<post_end> to nest, if there is something on the save stack that
2168calls string eval.
2169
a88d97bf 2170=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
2171
2172This is called just before starting to compile an C<eval STRING>, C<do
10e2eb10 2173FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
fd85fad2
BM
2174OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2175C<OP_DOFILE> or C<OP_REQUIRE>.
2176
2177=back
2178
2179Once you have your hook functions, you need a C<BHK> structure to put
10e2eb10
FC
2180them in. It's best to allocate it statically, since there is no way to
2181free it once it's registered. The function pointers should be inserted
fd85fad2 2182into this structure using the C<BhkENTRY_set> macro, which will also set
10e2eb10 2183flags indicating which entries are valid. If you do need to allocate
fd85fad2
BM
2184your C<BHK> dynamically for some reason, be sure to zero it before you
2185start.
2186
2187Once registered, there is no mechanism to switch these hooks off, so if
10e2eb10 2188that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
2189is probably the best way, so the effect is lexically scoped; however it
2190is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
10e2eb10 2191temporarily switch entries on and off. You should also be aware that
a3e07c87 2192generally speaking at least one scope will have opened before your
f185f654 2193extension is loaded, so you will see some C<pre>/C<post_end> pairs that
a3e07c87 2194didn't have a matching C<start>.
fd85fad2 2195
9afa14e3
SC
2196=head1 Examining internal data structures with the C<dump> functions
2197
2198To aid debugging, the source file F<dump.c> contains a number of
2199functions which produce formatted output of internal data structures.
2200
2201The most commonly used of these functions is C<Perl_sv_dump>; it's used
10e2eb10 2202for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
9afa14e3 2203C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 2204module should already be familiar with its format.
9afa14e3
SC
2205
2206C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 2207derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
2208C<Perl_dump_eval> will dump the main root of the code being evaluated,
2209exactly like C<-Dx>.
2210
2211Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2212op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2213subroutines in a package like so: (Thankfully, these are all xsubs, so
2214there is no op tree)
2215
2216 (gdb) print Perl_dump_packsubs(PL_defstash)
2217
2218 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2219
2220 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2221
2222 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2223
2224 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2225
2226 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2227
2228and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2229the op tree of the main root.
2230
954c1994 2231=head1 How multiple interpreters and concurrency are supported
ee072b34 2232
ee072b34
GS
2233=head2 Background and PERL_IMPLICIT_CONTEXT
2234
2235The Perl interpreter can be regarded as a closed box: it has an API
2236for feeding it code or otherwise making it do things, but it also has
2237functions for its own use. This smells a lot like an object, and
2238there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2239interpreters, with one interpreter represented either as a C structure,
2240or inside a thread-specific structure. These structures contain all
2241the context, the state of that interpreter.
2242
10e2eb10 2243One macro controls the major Perl build flavor: MULTIPLICITY. The
7b52221d 2244MULTIPLICITY build has a C structure that packages all the interpreter
10e2eb10 2245state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
7b52221d 2246normally defined, and enables the support for passing in a "hidden" first
10e2eb10 2247argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2248multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2249to the macro USE_ITHREADS.)
54aff467 2250
27da23d5
JH
2251Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2252PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2253former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2254internal variables of Perl to be wrapped inside a single global struct,
2255struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2256the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2257one step further, there is still a single struct (allocated in main()
2258either from heap or from stack) but there are no global data symbols
3bf17896 2259pointing to it. In either case the global struct should be initialized
27da23d5
JH
2260as the very first thing in main() using Perl_init_global_struct() and
2261correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2262please see F<miniperlmain.c> for usage details. You may also need
2263to use C<dVAR> in your coding to "declare the global variables"
2264when you are using them. dTHX does this for you automatically.
2265
9aa97215
JH
2266To see whether you have non-const data you can use a BSD (or GNU)
2267compatible C<nm>:
bc028b6b
JH
2268
2269 nm libperl.a | grep -v ' [TURtr] '
2270
9aa97215
JH
2271If this displays any C<D> or C<d> symbols (or possibly C<C> or C<c>),
2272you have non-const data. The symbols the C<grep> removed are as follows:
2273C<Tt> are I<text>, or code, the C<Rr> are I<read-only> (const) data,
2274and the C<U> is <undefined>, external symbols referred to.
2275
2276The test F<t/porting/libperl.t> does this kind of symbol sanity
2277checking on C<libperl.a>.
bc028b6b 2278
27da23d5
JH
2279For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2280doesn't actually hide all symbols inside a big global struct: some
2281PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2282then hides everything (see how the PERLIO_FUNCS_DECL is used).
2283
54aff467 2284All this obviously requires a way for the Perl internal functions to be
acfe0abc 2285either subroutines taking some kind of structure as the first
ee072b34 2286argument, or subroutines taking nothing as the first argument. To
acfe0abc 2287enable these two very different ways of building the interpreter,
ee072b34
GS
2288the Perl source (as it does in so many other situations) makes heavy
2289use of macros and subroutine naming conventions.
2290
54aff467 2291First problem: deciding which functions will be public API functions and
00aadd71 2292which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2293(think "S" for "secret" or "static"). All other functions begin with
2294"Perl_", but just because a function begins with "Perl_" does not mean it is
10e2eb10
FC
2295part of the API. (See L</Internal
2296Functions>.) The easiest way to be B<sure> a
00aadd71
NIS
2297function is part of the API is to find its entry in L<perlapi>.
2298If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2299think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2300L<perlbug> explaining why you think it should be.
ee072b34
GS
2301
2302Second problem: there must be a syntax so that the same subroutine
2303declarations and calls can pass a structure as their first argument,
2304or pass nothing. To solve this, the subroutines are named and
2305declared in a particular way. Here's a typical start of a static
2306function used within the Perl guts:
2307
2308 STATIC void
2309 S_incline(pTHX_ char *s)
2310
acfe0abc 2311STATIC becomes "static" in C, and may be #define'd to nothing in some
da8c5729 2312configurations in the future.
ee072b34 2313
651a3225
GS
2314A public function (i.e. part of the internal API, but not necessarily
2315sanctioned for use in extensions) begins like this:
ee072b34
GS
2316
2317 void
2307c6d0 2318 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2319
0147cd53 2320C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2321details of the interpreter's context. THX stands for "thread", "this",
2322or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2323The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2324or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2325their variants.
ee072b34 2326
a7486cbb
JH
2327When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2328first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2329in the pTHX_ macro indicates that the macro expansion needs a comma
2330after the context argument because other arguments follow it. If
2331PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2332subroutine is not prototyped to take the extra argument. The form of the
2333macro without the trailing underscore is used when there are no additional
ee072b34
GS
2334explicit arguments.
2335
54aff467 2336When a core function calls another, it must pass the context. This
2307c6d0 2337is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2338something like this:
2339
2307c6d0
SB
2340 #ifdef PERL_IMPLICIT_CONTEXT
2341 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2342 /* can't do this for vararg functions, see below */
2307c6d0
SB
2343 #else
2344 #define sv_setiv Perl_sv_setiv
2345 #endif
ee072b34
GS
2346
2347This works well, and means that XS authors can gleefully write:
2348
2307c6d0 2349 sv_setiv(foo, bar);
ee072b34
GS
2350
2351and still have it work under all the modes Perl could have been
2352compiled with.
2353
ee072b34
GS
2354This doesn't work so cleanly for varargs functions, though, as macros
2355imply that the number of arguments is known in advance. Instead we
2356either need to spell them out fully, passing C<aTHX_> as the first
2357argument (the Perl core tends to do this with functions like
2358Perl_warner), or use a context-free version.
2359
2360The context-free version of Perl_warner is called
2361Perl_warner_nocontext, and does not take the extra argument. Instead
2362it does dTHX; to get the context from thread-local storage. We
2363C<#define warner Perl_warner_nocontext> so that extensions get source
2364compatibility at the expense of performance. (Passing an arg is
2365cheaper than grabbing it from thread-local storage.)
2366
acfe0abc 2367You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2368Those are strictly for use within the core. Extensions and embedders
2369need only be aware of [pad]THX.
2370
a7486cbb
JH
2371=head2 So what happened to dTHR?
2372
2373C<dTHR> was introduced in perl 5.005 to support the older thread model.
2374The older thread model now uses the C<THX> mechanism to pass context
2375pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2376later still have it for backward source compatibility, but it is defined
2377to be a no-op.
2378
ee072b34
GS
2379=head2 How do I use all this in extensions?
2380
2381When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2382any functions in the Perl API will need to pass the initial context
2383argument somehow. The kicker is that you will need to write it in
2384such a way that the extension still compiles when Perl hasn't been
2385built with PERL_IMPLICIT_CONTEXT enabled.
2386
2387There are three ways to do this. First, the easy but inefficient way,
2388which is also the default, in order to maintain source compatibility
0147cd53 2389with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2390and aTHX_ macros to call a function that will return the context.
2391Thus, something like:
2392
2307c6d0 2393 sv_setiv(sv, num);
ee072b34 2394
4375e838 2395in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2396in effect:
ee072b34 2397
2307c6d0 2398 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2399
54aff467 2400or to this otherwise:
ee072b34 2401
2307c6d0 2402 Perl_sv_setiv(sv, num);
ee072b34 2403
da8c5729 2404You don't have to do anything new in your extension to get this; since
2fa86c13 2405the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2406work.
2407
2408The second, more efficient way is to use the following template for
2409your Foo.xs:
2410
c52f9dcd
JH
2411 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2412 #include "EXTERN.h"
2413 #include "perl.h"
2414 #include "XSUB.h"
ee072b34 2415
fd061412 2416 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2417
fd061412 2418 STATIC void
c52f9dcd
JH
2419 my_private_function(int arg1, int arg2)
2420 {
2421 dTHX; /* fetch context */
2422 ... call many Perl API functions ...
2423 }
ee072b34
GS
2424
2425 [... etc ...]
2426
c52f9dcd 2427 MODULE = Foo PACKAGE = Foo
ee072b34 2428
c52f9dcd 2429 /* typical XSUB */
ee072b34 2430
c52f9dcd
JH
2431 void
2432 my_xsub(arg)
2433 int arg
2434 CODE:
2435 my_private_function(arg, 10);
ee072b34
GS
2436
2437Note that the only two changes from the normal way of writing an
2438extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2439including the Perl headers, followed by a C<dTHX;> declaration at
2440the start of every function that will call the Perl API. (You'll
2441know which functions need this, because the C compiler will complain
2442that there's an undeclared identifier in those functions.) No changes
2443are needed for the XSUBs themselves, because the XS() macro is
2444correctly defined to pass in the implicit context if needed.
2445
2446The third, even more efficient way is to ape how it is done within
2447the Perl guts:
2448
2449
c52f9dcd
JH
2450 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2451 #include "EXTERN.h"
2452 #include "perl.h"
2453 #include "XSUB.h"
ee072b34
GS
2454
2455 /* pTHX_ only needed for functions that call Perl API */
fd061412 2456 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2457
fd061412 2458 STATIC void
c52f9dcd
JH
2459 my_private_function(pTHX_ int arg1, int arg2)
2460 {
2461 /* dTHX; not needed here, because THX is an argument */
2462 ... call Perl API functions ...
2463 }
ee072b34
GS
2464
2465 [... etc ...]
2466
c52f9dcd 2467 MODULE = Foo PACKAGE = Foo
ee072b34 2468
c52f9dcd 2469 /* typical XSUB */
ee072b34 2470
c52f9dcd
JH
2471 void
2472 my_xsub(arg)
2473 int arg
2474 CODE:
2475 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2476
2477This implementation never has to fetch the context using a function
2478call, since it is always passed as an extra argument. Depending on
2479your needs for simplicity or efficiency, you may mix the previous
2480two approaches freely.
2481
651a3225
GS
2482Never add a comma after C<pTHX> yourself--always use the form of the
2483macro with the underscore for functions that take explicit arguments,
2484or the form without the argument for functions with no explicit arguments.
ee072b34 2485
27da23d5
JH
2486If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2487definition is needed if the Perl global variables (see F<perlvars.h>
2488or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2489used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2490the need for C<dVAR> only with the said compile-time define, because
2491otherwise the Perl global variables are visible as-is.
2492
a7486cbb
JH
2493=head2 Should I do anything special if I call perl from multiple threads?
2494
2495If you create interpreters in one thread and then proceed to call them in
2496another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2497initialized correctly in each of those threads.
2498
2499The C<perl_alloc> and C<perl_clone> API functions will automatically set
2500the TLS slot to the interpreter they created, so that there is no need to do
2501anything special if the interpreter is always accessed in the same thread that
2502created it, and that thread did not create or call any other interpreters
2503afterwards. If that is not the case, you have to set the TLS slot of the
2504thread before calling any functions in the Perl API on that particular
2505interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2506thread as the first thing you do:
2507
2508 /* do this before doing anything else with some_perl */
2509 PERL_SET_CONTEXT(some_perl);
2510
2511 ... other Perl API calls on some_perl go here ...
2512
ee072b34
GS
2513=head2 Future Plans and PERL_IMPLICIT_SYS
2514
2515Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2516that the interpreter knows about itself and pass it around, so too are
2517there plans to allow the interpreter to bundle up everything it knows
2518about the environment it's running on. This is enabled with the
7b52221d
RGS
2519PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2520Windows.
ee072b34
GS
2521
2522This allows the ability to provide an extra pointer (called the "host"
2523environment) for all the system calls. This makes it possible for
2524all the system stuff to maintain their own state, broken down into
2525seven C structures. These are thin wrappers around the usual system
0147cd53 2526calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2527more ambitious host (like the one that would do fork() emulation) all
2528the extra work needed to pretend that different interpreters are
2529actually different "processes", would be done here.
2530
2531The Perl engine/interpreter and the host are orthogonal entities.
2532There could be one or more interpreters in a process, and one or
2533more "hosts", with free association between them.
2534
a422fd2d
SC
2535=head1 Internal Functions
2536
2537All of Perl's internal functions which will be exposed to the outside
06f6df17 2538world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d 2539functions or functions used in a program in which Perl is embedded.
10e2eb10 2540Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2541static functions start with C<S_>.)
a422fd2d 2542
0972ecdf
DM
2543Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2544either with or without the C<Perl_> prefix, thanks to a bunch of defines
10e2eb10 2545that live in F<embed.h>. Note that extension code should I<not> set
0972ecdf
DM
2546C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2547breakage of the XS in each new perl release.
2548
2549The file F<embed.h> is generated automatically from
10e2eb10 2550F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
dc9b1d22 2551header files for the internal functions, generates the documentation
10e2eb10 2552and a lot of other bits and pieces. It's important that when you add
dc9b1d22 2553a new function to the core or change an existing one, you change the
10e2eb10 2554data in the table in F<embed.fnc> as well. Here's a sample entry from
dc9b1d22 2555that table:
a422fd2d
SC
2556
2557 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2558
10e2eb10
FC
2559The second column is the return type, the third column the name. Columns
2560after that are the arguments. The first column is a set of flags:
a422fd2d
SC
2561
2562=over 3
2563
2564=item A
2565
10e2eb10
FC
2566This function is a part of the public
2567API. All such functions should also
1aa6ea50 2568have 'd', very few do not.
a422fd2d
SC
2569
2570=item p
2571
1aa6ea50
JC
2572This function has a C<Perl_> prefix; i.e. it is defined as
2573C<Perl_av_fetch>.
a422fd2d
SC
2574
2575=item d
2576
2577This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2578look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2579
2580=back
2581
2582Other available flags are:
2583
2584=over 3
2585
2586=item s
2587
1aa6ea50
JC
2588This is a static function and is defined as C<STATIC S_whatever>, and
2589usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2590
2591=item n
2592
da8c5729 2593This does not need an interpreter context, so the definition has no
1aa6ea50 2594C<pTHX>, and it follows that callers don't use C<aTHX>. (See
d3a43cd8 2595L</Background and PERL_IMPLICIT_CONTEXT>.)
a422fd2d
SC
2596
2597=item r
2598
2599This function never returns; C<croak>, C<exit> and friends.
2600
2601=item f
2602
2603This function takes a variable number of arguments, C<printf> style.
2604The argument list should end with C<...>, like this:
2605
2606 Afprd |void |croak |const char* pat|...
2607
a7486cbb 2608=item M
a422fd2d 2609
00aadd71 2610This function is part of the experimental development API, and may change
a422fd2d
SC
2611or disappear without notice.
2612
2613=item o
2614
2615This function should not have a compatibility macro to define, say,
10e2eb10 2616C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
a422fd2d 2617
a422fd2d
SC
2618=item x
2619
2620This function isn't exported out of the Perl core.
2621
dc9b1d22
MHM
2622=item m
2623
2624This is implemented as a macro.
2625
2626=item X
2627
2628This function is explicitly exported.
2629
2630=item E
2631
2632This function is visible to extensions included in the Perl core.
2633
2634=item b
2635
2636Binary backward compatibility; this function is a macro but also has
2637a C<Perl_> implementation (which is exported).
2638
1aa6ea50
JC
2639=item others
2640
2641See the comments at the top of C<embed.fnc> for others.
2642
a422fd2d
SC
2643=back
2644
dc9b1d22
MHM
2645If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2646C<make regen_headers> to force a rebuild of F<embed.h> and other
2647auto-generated files.
a422fd2d 2648
6b4667fc 2649=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2650
6b4667fc
A
2651If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2652formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2653following macros for portability
9dd9db0b 2654
c52f9dcd
JH
2655 IVdf IV in decimal
2656 UVuf UV in decimal
2657 UVof UV in octal
2658 UVxf UV in hexadecimal
2659 NVef NV %e-like
2660 NVff NV %f-like
2661 NVgf NV %g-like
9dd9db0b 2662
6b4667fc
A
2663These will take care of 64-bit integers and long doubles.
2664For example:
2665
c52f9dcd 2666 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2667
2668The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2669
aacf4ea2
JH
2670Note that there are different "long doubles": Perl will use
2671whatever the compiler has.
2672
8908e76d
JH
2673If you are printing addresses of pointers, use UVxf combined
2674with PTR2UV(), do not use %lx or %p.
2675
2676=head2 Pointer-To-Integer and Integer-To-Pointer
2677
2678Because pointer size does not necessarily equal integer size,
2679use the follow macros to do it right.
2680
c52f9dcd
JH
2681 PTR2UV(pointer)
2682 PTR2IV(pointer)
2683 PTR2NV(pointer)
2684 INT2PTR(pointertotype, integer)
8908e76d
JH
2685
2686For example:
2687
c52f9dcd
JH
2688 IV iv = ...;
2689 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2690
2691and
2692
c52f9dcd
JH
2693 AV *av = ...;
2694 UV uv = PTR2UV(av);
8908e76d 2695
0ca3a874
MHM
2696=head2 Exception Handling
2697
9b5c3821 2698There are a couple of macros to do very basic exception handling in XS
10e2eb10 2699modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
9b5c3821
MHM
2700be able to use these macros:
2701
2702 #define NO_XSLOCKS
2703 #include "XSUB.h"
2704
2705You can use these macros if you call code that may croak, but you need
10e2eb10 2706to do some cleanup before giving control back to Perl. For example:
0ca3a874 2707
d7f8936a 2708 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2709
2710 XCPT_TRY_START {
2711 code_that_may_croak();
2712 } XCPT_TRY_END
2713
2714 XCPT_CATCH
2715 {
2716 /* do cleanup here */
2717 XCPT_RETHROW;
2718 }
2719
2720Note that you always have to rethrow an exception that has been
10e2eb10
FC
2721caught. Using these macros, it is not possible to just catch the
2722exception and ignore it. If you have to ignore the exception, you
0ca3a874
MHM
2723have to use the C<call_*> function.
2724
2725The advantage of using the above macros is that you don't have
2726to setup an extra function for C<call_*>, and that using these
2727macros is faster than using C<call_*>.
2728
a422fd2d
SC
2729=head2 Source Documentation
2730
2731There's an effort going on to document the internal functions and
61ad4b94 2732automatically produce reference manuals from them -- L<perlapi> is one
a422fd2d 2733such manual which details all the functions which are available to XS
10e2eb10 2734writers. L<perlintern> is the autogenerated manual for the functions
a422fd2d
SC
2735which are not part of the API and are supposedly for internal use only.
2736
2737Source documentation is created by putting POD comments into the C
2738source, like this:
2739
2740 /*
2741 =for apidoc sv_setiv
2742
2743 Copies an integer into the given SV. Does not handle 'set' magic. See
2744 C<sv_setiv_mg>.
2745
2746 =cut
2747 */
2748
2749Please try and supply some documentation if you add functions to the
2750Perl core.
2751
0d098d33
MHM
2752=head2 Backwards compatibility
2753
10e2eb10
FC
2754The Perl API changes over time. New functions are
2755added or the interfaces of existing functions are
2756changed. The C<Devel::PPPort> module tries to
0d098d33
MHM
2757provide compatibility code for some of these changes, so XS writers don't
2758have to code it themselves when supporting multiple versions of Perl.
2759
2760C<Devel::PPPort> generates a C header file F<ppport.h> that can also
10e2eb10 2761be run as a Perl script. To generate F<ppport.h>, run:
0d098d33
MHM
2762
2763 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2764
2765Besides checking existing XS code, the script can also be used to retrieve
2766compatibility information for various API calls using the C<--api-info>
10e2eb10 2767command line switch. For example:
0d098d33
MHM
2768
2769 % perl ppport.h --api-info=sv_magicext
2770
2771For details, see C<perldoc ppport.h>.
2772
a422fd2d
SC
2773=head1 Unicode Support
2774
10e2eb10 2775Perl 5.6.0 introduced Unicode support. It's important for porters and XS
a422fd2d
SC
2776writers to understand this support and make sure that the code they
2777write does not corrupt Unicode data.
2778
2779=head2 What B<is> Unicode, anyway?
2780
10e2eb10
FC
2781In the olden, less enlightened times, we all used to use ASCII. Most of
2782us did, anyway. The big problem with ASCII is that it's American. Well,
a422fd2d 2783no, that's not actually the problem; the problem is that it's not
10e2eb10 2784particularly useful for people who don't use the Roman alphabet. What
a422fd2d 2785used to happen was that particular languages would stick their own
10e2eb10 2786alphabet in the upper range of the sequence, between 128 and 255. Of
a422fd2d
SC
2787course, we then ended up with plenty of variants that weren't quite
2788ASCII, and the whole point of it being a standard was lost.
2789
2790Worse still, if you've got a language like Chinese or
2791Japanese that has hundreds or thousands of characters, then you really
2792can't fit them into a mere 256, so they had to forget about ASCII
2793altogether, and build their own systems using pairs of numbers to refer
2794to one character.
2795
2796To fix this, some people formed Unicode, Inc. and
2797produced a new character set containing all the characters you can
10e2eb10
FC
2798possibly think of and more. There are several ways of representing these
2799characters, and the one Perl uses is called UTF-8. UTF-8 uses
2800a variable number of bytes to represent a character. You can learn more
2575c402 2801about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2802
3ad86f0e
KW
2803(On EBCDIC platforms, Perl uses instead UTF-EBCDIC, which is a form of
2804UTF-8 adapted for EBCDIC platforms. Below, we just talk about UTF-8.
2805UTF-EBCDIC is like UTF-8, but the details are different. The macros
2806hide the differences from you, just remember that the particular numbers
2807and bit patterns presented below will differ in UTF-EBCDIC.)
2808
1e54db1a 2809=head2 How can I recognise a UTF-8 string?
a422fd2d 2810
10e2eb10
FC
2811You can't. This is because UTF-8 data is stored in bytes just like
2812non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d 2813capital E with a grave accent, is represented by the two bytes
10e2eb10 2814C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
61ad4b94 2815has that byte sequence as well. So you can't tell just by looking -- this
a422fd2d
SC
2816is what makes Unicode input an interesting problem.
2817
2575c402
JW
2818In general, you either have to know what you're dealing with, or you
2819have to guess. The API function C<is_utf8_string> can help; it'll tell
61ad4b94
KW
2820you if a string contains only valid UTF-8 characters, and the chances
2821of a non-UTF-8 string looking like valid UTF-8 become very small very
2822quickly with increasing string length. On a character-by-character
2823basis, C<isUTF8_CHAR>
2575c402 2824will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2825
1e54db1a 2826=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2827
1e54db1a 2828As mentioned above, UTF-8 uses a variable number of bytes to store a
10e2eb10
FC
2829character. Characters with values 0...127 are stored in one
2830byte, just like good ol' ASCII. Character 128 is stored as
2831C<v194.128>; this continues up to character 191, which is
2832C<v194.191>. Now we've run out of bits (191 is binary
61ad4b94 2833C<10111111>) so we move on; character 192 is C<v195.128>. And
a422fd2d 2834so it goes on, moving to three bytes at character 2048.
6e31cdd1 2835L<perlunicode/Unicode Encodings> has pictures of how this works.
a422fd2d 2836
1e54db1a 2837Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2838how long the first character in it is with the C<UTF8SKIP> macro:
2839
2840 char *utf = "\305\233\340\240\201";
2841 I32 len;
2842
2843 len = UTF8SKIP(utf); /* len is 2 here */
2844 utf += len;
2845 len = UTF8SKIP(utf); /* len is 3 here */
2846
1e54db1a 2847Another way to skip over characters in a UTF-8 string is to use
a422fd2d 2848C<utf8_hop>, which takes a string and a number of characters to skip
10e2eb10 2849over. You're on your own about bounds checking, though, so don't use it
a422fd2d
SC
2850lightly.
2851
1e54db1a 2852All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe 2853so you can test if you need to do something special with this
61ad4b94 2854character like this (the C<UTF8_IS_INVARIANT()> is a macro that tests
9f98c7fe 2855whether the byte is encoded as a single byte even in UTF-8):
a422fd2d 2856
3a2263fe 2857 U8 *utf;
4b88fb76 2858 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
3a2263fe 2859 UV uv; /* Note: a UV, not a U8, not a char */
95701e00 2860 STRLEN len; /* length of character in bytes */
a422fd2d 2861
3a2263fe 2862 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2863 /* Must treat this as UTF-8 */
4b88fb76 2864 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
a422fd2d
SC
2865 else
2866 /* OK to treat this character as a byte */
2867 uv = *utf;
2868
4b88fb76 2869You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
95701e00 2870value of the character; the inverse function C<uvchr_to_utf8> is available
1e54db1a 2871for putting a UV into UTF-8:
a422fd2d 2872
61ad4b94 2873 if (!UVCHR_IS_INVARIANT(uv))
a422fd2d 2874 /* Must treat this as UTF8 */
95701e00 2875 utf8 = uvchr_to_utf8(utf8, uv);
a422fd2d
SC
2876 else
2877 /* OK to treat this character as a byte */
2878 *utf8++ = uv;
2879
2880You B<must> convert characters to UVs using the above functions if
1e54db1a 2881you're ever in a situation where you have to match UTF-8 and non-UTF-8
10e2eb10 2882characters. You may not skip over UTF-8 characters in this case. If you
1e54db1a
JH
2883do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2884for instance, if your UTF-8 string contains C<v196.172>, and you skip
2885that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2886So don't do that!
2887
61ad4b94
KW
2888(Note that we don't have to test for invariant characters in the
2889examples above. The functions work on any well-formed UTF-8 input.
2890It's just that its faster to avoid the function overhead when it's not
2891needed.)
2892
1e54db1a 2893=head2 How does Perl store UTF-8 strings?
a422fd2d 2894
61ad4b94 2895Currently, Perl deals with UTF-8 strings and non-UTF-8 strings
10e2eb10
FC
2896slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2897string is internally encoded as UTF-8. Without it, the byte value is the
61ad4b94
KW
2898codepoint number and vice versa. This flag is only meaningful if the SV
2899is C<SvPOK> or immediately after stringification via C<SvPV> or a
2900similar macro. You can check and manipulate this flag with the
2575c402 2901following macros:
a422fd2d
SC
2902
2903 SvUTF8(sv)
2904 SvUTF8_on(sv)
2905 SvUTF8_off(sv)
2906
2907This flag has an important effect on Perl's treatment of the string: if
61ad4b94 2908UTF-8 data is not properly distinguished, regular expressions,
a422fd2d 2909C<length>, C<substr> and other string handling operations will have
61ad4b94 2910undesirable (wrong) results.
a422fd2d
SC
2911
2912The problem comes when you have, for instance, a string that isn't
61ad4b94 2913flagged as UTF-8, and contains a byte sequence that could be UTF-8 --
1e54db1a 2914especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d 2915
61ad4b94
KW
2916Never forget that the C<SVf_UTF8> flag is separate from the PV value; you
2917need to be sure you don't accidentally knock it off while you're
10e2eb10 2918manipulating SVs. More specifically, you cannot expect to do this:
a422fd2d
SC
2919
2920 SV *sv;
2921 SV *nsv;
2922 STRLEN len;
2923 char *p;
2924
2925 p = SvPV(sv, len);
2926 frobnicate(p);
2927 nsv = newSVpvn(p, len);
2928
2929The C<char*> string does not tell you the whole story, and you can't
10e2eb10 2930copy or reconstruct an SV just by copying the string value. Check if the
c31cc9fc
FC
2931old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
2932accordingly:
a422fd2d
SC
2933
2934 p = SvPV(sv, len);
6db25795
KW
2935 is_utf8 = SvUTF8(sv);
2936 frobnicate(p, is_utf8);
a422fd2d 2937 nsv = newSVpvn(p, len);
6db25795 2938 if (is_utf8)
a422fd2d
SC
2939 SvUTF8_on(nsv);
2940
6db25795
KW
2941In the above, your C<frobnicate> function has been changed to be made
2942aware of whether or not it's dealing with UTF-8 data, so that it can
2943handle the string appropriately.
a422fd2d 2944
3a2263fe 2945Since just passing an SV to an XS function and copying the data of
2575c402 2946the SV is not enough to copy the UTF8 flags, even less right is just
61ad4b94 2947passing a S<C<char *>> to an XS function.
3a2263fe 2948
dc83bf8e 2949For full generality, use the L<C<DO_UTF8>|perlapi/DO_UTF8> macro to see if the
6db25795
KW
2950string in an SV is to be I<treated> as UTF-8. This takes into account
2951if the call to the XS function is being made from within the scope of
2952L<S<C<use bytes>>|bytes>. If so, the underlying bytes that comprise the
2953UTF-8 string are to be exposed, rather than the character they
2954represent. But this pragma should only really be used for debugging and
2955perhaps low-level testing at the byte level. Hence most XS code need
2956not concern itself with this, but various areas of the perl core do need
2957to support it.
2958
2959And this isn't the whole story. Starting in Perl v5.12, strings that
2960aren't encoded in UTF-8 may also be treated as Unicode under various
6e31cdd1 2961conditions (see L<perlunicode/ASCII Rules versus Unicode Rules>).
6db25795
KW
2962This is only really a problem for characters whose ordinals are between
2963128 and 255, and their behavior varies under ASCII versus Unicode rules
2964in ways that your code cares about (see L<perlunicode/The "Unicode Bug">).
2965There is no published API for dealing with this, as it is subject to
2966change, but you can look at the code for C<pp_lc> in F<pp.c> for an
2967example as to how it's currently done.
2968
1e54db1a 2969=head2 How do I convert a string to UTF-8?
a422fd2d 2970
2575c402 2971If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
61ad4b94 2972the non-UTF-8 strings to UTF-8. If you've got an SV, the easiest way to do
2575c402 2973this is:
a422fd2d
SC
2974
2975 sv_utf8_upgrade(sv);
2976
2977However, you must not do this, for example:
2978
2979 if (!SvUTF8(left))
2980 sv_utf8_upgrade(left);
2981
2982If you do this in a binary operator, you will actually change one of the
b1866b2d 2983strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2984by the end user, it can cause problems in deficient code.
a422fd2d 2985
1e54db1a 2986Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
10e2eb10
FC
2987string argument. This is useful for having the data available for
2988comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2989C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2990the string contains any characters above 255 that can't be represented
2991in a single byte.
2992
6db25795
KW
2993=head2 How do I compare strings?
2994
2995L<perlapi/sv_cmp> and L<perlapi/sv_cmp_flags> do a lexigraphic
2996comparison of two SV's, and handle UTF-8ness properly. Note, however,
2997that Unicode specifies a much fancier mechanism for collation, available
2998via the L<Unicode::Collate> module.
2999
3000To just compare two strings for equality/non-equality, you can just use
3001L<C<memEQ()>|perlapi/memEQ> and L<C<memNE()>|perlapi/memEQ> as usual,
3002except the strings must be both UTF-8 or not UTF-8 encoded.
3003
3004To compare two strings case-insensitively, use
3005L<C<foldEQ_utf8()>|perlapi/foldEQ_utf8> (the strings don't have to have
3006the same UTF-8ness).
3007
a422fd2d
SC
3008=head2 Is there anything else I need to know?
3009
10e2eb10 3010Not really. Just remember these things:
a422fd2d
SC
3011
3012=over 3
3013
3014=item *
3015
6db25795
KW
3016There's no way to tell if a S<C<char *>> or S<C<U8 *>> string is UTF-8
3017or not. But you can tell if an SV is to be treated as UTF-8 by calling
3018C<DO_UTF8> on it, after stringifying it with C<SvPV> or a similar
3019macro. And, you can tell if SV is actually UTF-8 (even if it is not to
3020be treated as such) by looking at its C<SvUTF8> flag (again after
3021stringifying it). Don't forget to set the flag if something should be
3022UTF-8.
3023Treat the flag as part of the PV, even though it's not -- if you pass on
3024the PV to somewhere, pass on the flag too.
a422fd2d
SC
3025
3026=item *
3027
4b88fb76 3028If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3a2263fe 3029unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
3030
3031=item *
3032
61ad4b94
KW
3033When writing a character UV to a UTF-8 string, B<always> use
3034C<uvchr_to_utf8>, unless C<UVCHR_IS_INVARIANT(uv))> in which case
3a2263fe 3035you can use C<*s = uv>.
a422fd2d
SC
3036
3037=item *
3038
10e2eb10
FC
3039Mixing UTF-8 and non-UTF-8 strings is
3040tricky. Use C<bytes_to_utf8> to get
2bbc8d55 3041a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
3042
3043=back
3044
53e06cf0
SC
3045=head1 Custom Operators
3046
2a0fd0f1 3047Custom operator support is an experimental feature that allows you to
10e2eb10 3048define your own ops. This is primarily to allow the building of
53e06cf0
SC
3049interpreters for other languages in the Perl core, but it also allows
3050optimizations through the creation of "macro-ops" (ops which perform the
3051functions of multiple ops which are usually executed together, such as
1aa6ea50 3052C<gvsv, gvsv, add>.)
53e06cf0 3053
10e2eb10 3054This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0 3055core does not "know" anything special about this op type, and so it will
10e2eb10 3056not be involved in any optimizations. This also means that you can
61ad4b94
KW
3057define your custom ops to be any op structure -- unary, binary, list and
3058so on -- you like.
53e06cf0 3059
10e2eb10
FC
3060It's important to know what custom operators won't do for you. They
3061won't let you add new syntax to Perl, directly. They won't even let you
3062add new keywords, directly. In fact, they won't change the way Perl
3063compiles a program at all. You have to do those changes yourself, after
3064Perl has compiled the program. You do this either by manipulating the op
53e06cf0
SC
3065tree using a C<CHECK> block and the C<B::Generate> module, or by adding
3066a custom peephole optimizer with the C<optimize> module.
3067
3068When you do this, you replace ordinary Perl ops with custom ops by
407f86e1 3069creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
10e2eb10
FC
3070PP function. This should be defined in XS code, and should look like
3071the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
53e06cf0
SC
3072takes the appropriate number of values from the stack, and you are
3073responsible for adding stack marks if necessary.
3074
3075You should also "register" your op with the Perl interpreter so that it
10e2eb10 3076can produce sensible error and warning messages. Since it is possible to
53e06cf0 3077have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
9733086d 3078Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
10e2eb10 3079it is dealing with. You should create an C<XOP> structure for each
9733086d
BM
3080ppaddr you use, set the properties of the custom op with
3081C<XopENTRY_set>, and register the structure against the ppaddr using
10e2eb10 3082C<Perl_custom_op_register>. A trivial example might look like:
9733086d
BM
3083
3084 static XOP my_xop;
3085 static OP *my_pp(pTHX);
3086
3087 BOOT:
3088 XopENTRY_set(&my_xop, xop_name, "myxop");
3089 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
3090 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
3091
3092The available fields in the structure are:
3093
3094=over 4
3095
3096=item xop_name
3097
10e2eb10 3098A short name for your op. This will be included in some error messages,
9733086d
BM
3099and will also be returned as C<< $op->name >> by the L<B|B> module, so
3100it will appear in the output of module like L<B::Concise|B::Concise>.
3101
3102=item xop_desc
3103
3104A short description of the function of the op.
3105
3106=item xop_class
3107
10e2eb10 3108Which of the various C<*OP> structures this op uses. This should be one of
9733086d
BM
3109the C<OA_*> constants from F<op.h>, namely
3110
3111=over 4
3112
3113=item OA_BASEOP
3114
3115=item OA_UNOP
3116
3117=item OA_BINOP
3118
3119=item OA_LOGOP
3120
3121=item OA_LISTOP
3122
3123=item OA_PMOP
3124
3125=item OA_SVOP
3126
3127=item OA_PADOP
3128
3129=item OA_PVOP_OR_SVOP
3130
10e2eb10 3131This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
9733086d
BM
3132the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
3133
3134=item OA_LOOP
3135
3136=item OA_COP
3137
3138=back
3139
3140The other C<OA_*> constants should not be used.
3141
3142=item xop_peep
3143
3144This member is of type C<Perl_cpeep_t>, which expands to C<void
10e2eb10 3145(*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
9733086d 3146will be called from C<Perl_rpeep> when ops of this type are encountered
10e2eb10 3147by the peephole optimizer. I<o> is the OP that needs optimizing;
9733086d
BM
3148I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
3149
3150=back
53e06cf0 3151
e7d4c058 3152C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 3153
954c1994 3154=head1 AUTHORS
e89caa19 3155
954c1994 3156Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
3157E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
3158itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 3159
954c1994
GS
3160With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
3161Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
3162Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
3163Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 3164
954c1994 3165=head1 SEE ALSO
cb1a09d0 3166
ba555bf5 3167L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>