This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
fix signed/unsigned mismatch in (M)EXTEND
[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
d6605d24
KW
42possible to have it point to other things For example, it could point
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
326This is best demonstrated by example:
327
328 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
329 SV = PVIV(0x8128450) at 0x81340f0
330 REFCNT = 1
331 FLAGS = (POK,OOK,pPOK)
332 IV = 1 (OFFSET)
333 PV = 0x8135781 ( "1" . ) "2345"\0
334 CUR = 4
335 LEN = 5
336
337Here the number of bytes chopped off (1) is put into IV, and
10e2eb10 338C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
94dde4fb
SC
339portion of the string between the "real" and the "fake" beginnings is
340shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
341the fake beginning, not the real one.
342
fe854a6f 343Something similar to the offset hack is performed on AVs to enable
319cef53
SC
344efficient shifting and splicing off the beginning of the array; while
345C<AvARRAY> points to the first element in the array that is visible from
10e2eb10 346Perl, C<AvALLOC> points to the real start of the C array. These are
319cef53 347usually the same, but a C<shift> operation can be carried out by
6de131f0 348increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
319cef53 349Again, the location of the real start of the C array only comes into
10e2eb10 350play when freeing the array. See C<av_shift> in F<av.c>.
319cef53 351
d1b91892 352=head2 What's Really Stored in an SV?
a0d0e21e
LW
353
354Recall that the usual method of determining the type of scalar you have is
5f05dabc 355to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 356usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
357macros will do the appropriate conversion of string to integer/double or
358integer/double to string.
359
360If you I<really> need to know if you have an integer, double, or string
361pointer in an SV, you can use the following three macros instead:
362
363 SvIOKp(SV*)
364 SvNOKp(SV*)
365 SvPOKp(SV*)
366
367These will tell you if you truly have an integer, double, or string pointer
d1b91892 368stored in your SV. The "p" stands for private.
a0d0e21e 369
da8c5729 370There are various ways in which the private and public flags may differ.
9090718a
FC
371For example, in perl 5.16 and earlier a tied SV may have a valid
372underlying value in the IV slot (so SvIOKp is true), but the data
373should be accessed via the FETCH routine rather than directly,
374so SvIOK is false. (In perl 5.18 onwards, tied scalars use
375the flags the same way as untied scalars.) Another is when
d7f8936a 376numeric conversion has occurred and precision has been lost: only the
10e2eb10 377private flag is set on 'lossy' values. So when an NV is converted to an
9e9796d6
JH
378IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
379
07fa94a1 380In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 381
54310121 382=head2 Working with AVs
a0d0e21e 383
07fa94a1
JO
384There are two ways to create and load an AV. The first method creates an
385empty AV:
a0d0e21e
LW
386
387 AV* newAV();
388
54310121 389The second method both creates the AV and initially populates it with SVs:
a0d0e21e 390
c70927a6 391 AV* av_make(SSize_t num, SV **ptr);
a0d0e21e 392
5f05dabc 393The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 394AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 395
da8c5729 396Once the AV has been created, the following operations are possible on it:
a0d0e21e
LW
397
398 void av_push(AV*, SV*);
399 SV* av_pop(AV*);
400 SV* av_shift(AV*);
c70927a6 401 void av_unshift(AV*, SSize_t num);
a0d0e21e
LW
402
403These should be familiar operations, with the exception of C<av_unshift>.
404This routine adds C<num> elements at the front of the array with the C<undef>
405value. You must then use C<av_store> (described below) to assign values
406to these new elements.
407
408Here are some other functions:
409
c70927a6
FC
410 SSize_t av_top_index(AV*);
411 SV** av_fetch(AV*, SSize_t key, I32 lval);
412 SV** av_store(AV*, SSize_t key, SV* val);
a0d0e21e 413
dab460cd 414The C<av_top_index> function returns the highest index value in an array (just
5f05dabc
PP
415like $#array in Perl). If the array is empty, -1 is returned. The
416C<av_fetch> function returns the value at index C<key>, but if C<lval>
417is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
418The C<av_store> function stores the value C<val> at index C<key>, and does
419not increment the reference count of C<val>. Thus the caller is responsible
420for taking care of that, and if C<av_store> returns NULL, the caller will
421have to decrement the reference count to avoid a memory leak. Note that
422C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
423return value.
d1b91892 424
da8c5729
MH
425A few more:
426
a0d0e21e 427 void av_clear(AV*);
a0d0e21e 428 void av_undef(AV*);
c70927a6 429 void av_extend(AV*, SSize_t key);
5f05dabc
PP
430
431The C<av_clear> function deletes all the elements in the AV* array, but
432does not actually delete the array itself. The C<av_undef> function will
433delete all the elements in the array plus the array itself. The
adc882cf
GS
434C<av_extend> function extends the array so that it contains at least C<key+1>
435elements. If C<key+1> is less than the currently allocated length of the array,
436then nothing is done.
a0d0e21e
LW
437
438If you know the name of an array variable, you can get a pointer to its AV
439by using the following:
440
cbfd0a87 441 AV* get_av("package::varname", 0);
a0d0e21e
LW
442
443This returns NULL if the variable does not exist.
444
04343c6d
GS
445See L<Understanding the Magic of Tied Hashes and Arrays> for more
446information on how to use the array access functions on tied arrays.
447
54310121 448=head2 Working with HVs
a0d0e21e
LW
449
450To create an HV, you use the following routine:
451
452 HV* newHV();
453
da8c5729 454Once the HV has been created, the following operations are possible on it:
a0d0e21e 455
08105a92
GS
456 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
457 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 458
5f05dabc
PP
459The C<klen> parameter is the length of the key being passed in (Note that
460you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
461length of the key). The C<val> argument contains the SV pointer to the
54310121 462scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
463you want C<hv_store> to calculate it for you). The C<lval> parameter
464indicates whether this fetch is actually a part of a store operation, in
465which case a new undefined value will be added to the HV with the supplied
466key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 467
5f05dabc
PP
468Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
469C<SV*>. To access the scalar value, you must first dereference the return
470value. However, you should check to make sure that the return value is
471not NULL before dereferencing it.
a0d0e21e 472
da8c5729
MH
473The first of these two functions checks if a hash table entry exists, and the
474second deletes it.
a0d0e21e 475
08105a92
GS
476 bool hv_exists(HV*, const char* key, U32 klen);
477 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 478
5f05dabc
PP
479If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
480create and return a mortal copy of the deleted value.
481
a0d0e21e
LW
482And more miscellaneous functions:
483
484 void hv_clear(HV*);
a0d0e21e 485 void hv_undef(HV*);
5f05dabc
PP
486
487Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
488table but does not actually delete the hash table. The C<hv_undef> deletes
489both the entries and the hash table itself.
a0d0e21e 490
a9b0660e 491Perl keeps the actual data in a linked list of structures with a typedef of HE.
d1b91892
AD
492These contain the actual key and value pointers (plus extra administrative
493overhead). The key is a string pointer; the value is an C<SV*>. However,
494once you have an C<HE*>, to get the actual key and value, use the routines
495specified below.
496
a0d0e21e
LW
497 I32 hv_iterinit(HV*);
498 /* Prepares starting point to traverse hash table */
499 HE* hv_iternext(HV*);
500 /* Get the next entry, and return a pointer to a
501 structure that has both the key and value */
502 char* hv_iterkey(HE* entry, I32* retlen);
503 /* Get the key from an HE structure and also return
504 the length of the key string */
cb1a09d0 505 SV* hv_iterval(HV*, HE* entry);
d1be9408 506 /* Return an SV pointer to the value of the HE
a0d0e21e 507 structure */
cb1a09d0 508 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
509 /* This convenience routine combines hv_iternext,
510 hv_iterkey, and hv_iterval. The key and retlen
511 arguments are return values for the key and its
512 length. The value is returned in the SV* argument */
a0d0e21e
LW
513
514If you know the name of a hash variable, you can get a pointer to its HV
515by using the following:
516
6673a63c 517 HV* get_hv("package::varname", 0);
a0d0e21e
LW
518
519This returns NULL if the variable does not exist.
520
a43e7901 521The hash algorithm is defined in the C<PERL_HASH> macro:
a0d0e21e 522
a43e7901 523 PERL_HASH(hash, key, klen)
ab192400 524
a43e7901
YO
525The exact implementation of this macro varies by architecture and version
526of perl, and the return value may change per invocation, so the value
527is only valid for the duration of a single perl process.
a0d0e21e 528
04343c6d
GS
529See L<Understanding the Magic of Tied Hashes and Arrays> for more
530information on how to use the hash access functions on tied hashes.
531
1e422769
PP
532=head2 Hash API Extensions
533
534Beginning with version 5.004, the following functions are also supported:
535
536 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
537 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 538
1e422769
PP
539 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
540 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 541
1e422769
PP
542 SV* hv_iterkeysv (HE* entry);
543
544Note that these functions take C<SV*> keys, which simplifies writing
545of extension code that deals with hash structures. These functions
546also allow passing of C<SV*> keys to C<tie> functions without forcing
547you to stringify the keys (unlike the previous set of functions).
548
549They also return and accept whole hash entries (C<HE*>), making their
550use more efficient (since the hash number for a particular string
4a4eefd0
GS
551doesn't have to be recomputed every time). See L<perlapi> for detailed
552descriptions.
1e422769
PP
553
554The following macros must always be used to access the contents of hash
555entries. Note that the arguments to these macros must be simple
556variables, since they may get evaluated more than once. See
4a4eefd0 557L<perlapi> for detailed descriptions of these macros.
1e422769
PP
558
559 HePV(HE* he, STRLEN len)
560 HeVAL(HE* he)
561 HeHASH(HE* he)
562 HeSVKEY(HE* he)
563 HeSVKEY_force(HE* he)
564 HeSVKEY_set(HE* he, SV* sv)
565
566These two lower level macros are defined, but must only be used when
567dealing with keys that are not C<SV*>s:
568
569 HeKEY(HE* he)
570 HeKLEN(HE* he)
571
04343c6d
GS
572Note that both C<hv_store> and C<hv_store_ent> do not increment the
573reference count of the stored C<val>, which is the caller's responsibility.
574If these functions return a NULL value, the caller will usually have to
575decrement the reference count of C<val> to avoid a memory leak.
1e422769 576
a9381218
MHM
577=head2 AVs, HVs and undefined values
578
10e2eb10
FC
579Sometimes you have to store undefined values in AVs or HVs. Although
580this may be a rare case, it can be tricky. That's because you're
a9381218
MHM
581used to using C<&PL_sv_undef> if you need an undefined SV.
582
583For example, intuition tells you that this XS code:
584
585 AV *av = newAV();
586 av_store( av, 0, &PL_sv_undef );
587
588is equivalent to this Perl code:
589
590 my @av;
591 $av[0] = undef;
592
f3c4ec28 593Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
a9381218
MHM
594for indicating that an array element has not yet been initialized.
595Thus, C<exists $av[0]> would be true for the above Perl code, but
f3c4ec28
FC
596false for the array generated by the XS code. In perl 5.20, storing
597&PL_sv_undef will create a read-only element, because the scalar
598&PL_sv_undef itself is stored, not a copy.
a9381218 599
f3c4ec28 600Similar problems can occur when storing C<&PL_sv_undef> in HVs:
a9381218
MHM
601
602 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
603
604This will indeed make the value C<undef>, but if you try to modify
605the value of C<key>, you'll get the following error:
606
607 Modification of non-creatable hash value attempted
608
609In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
10e2eb10 610in restricted hashes. This caused such hash entries not to appear
a9381218
MHM
611when iterating over the hash or when checking for the keys
612with the C<hv_exists> function.
613
8abccac8 614You can run into similar problems when you store C<&PL_sv_yes> or
10e2eb10 615C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
616will give you the following error:
617
618 Modification of a read-only value attempted
619
620To make a long story short, you can use the special variables
8abccac8 621C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
622HVs, but you have to make sure you know what you're doing.
623
624Generally, if you want to store an undefined value in an AV
625or HV, you should not use C<&PL_sv_undef>, but rather create a
626new undefined value using the C<newSV> function, for example:
627
628 av_store( av, 42, newSV(0) );
629 hv_store( hv, "foo", 3, newSV(0), 0 );
630
a0d0e21e
LW
631=head2 References
632
d1b91892 633References are a special type of scalar that point to other data types
a9b0660e 634(including other references).
a0d0e21e 635
07fa94a1 636To create a reference, use either of the following functions:
a0d0e21e 637
5f05dabc
PP
638 SV* newRV_inc((SV*) thing);
639 SV* newRV_noinc((SV*) thing);
a0d0e21e 640
5f05dabc 641The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
642functions are identical except that C<newRV_inc> increments the reference
643count of the C<thing>, while C<newRV_noinc> does not. For historical
644reasons, C<newRV> is a synonym for C<newRV_inc>.
645
646Once you have a reference, you can use the following macro to dereference
647the reference:
a0d0e21e
LW
648
649 SvRV(SV*)
650
651then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 652C<AV*> or C<HV*>, if required.
a0d0e21e 653
d1b91892 654To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
655
656 SvROK(SV*)
657
07fa94a1
JO
658To discover what type of value the reference refers to, use the following
659macro and then check the return value.
d1b91892
AD
660
661 SvTYPE(SvRV(SV*))
662
663The most useful types that will be returned are:
664
a5e62da0
FC
665 < SVt_PVAV Scalar
666 SVt_PVAV Array
667 SVt_PVHV Hash
668 SVt_PVCV Code
669 SVt_PVGV Glob (possibly a file handle)
670
671See L<perlapi/svtype> for more details.
d1b91892 672
cb1a09d0
AD
673=head2 Blessed References and Class Objects
674
06f6df17 675References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
676OO lexicon, an object is simply a reference that has been blessed into a
677package (or class). Once blessed, the programmer may now use the reference
678to access the various methods in the class.
679
680A reference can be blessed into a package with the following function:
681
682 SV* sv_bless(SV* sv, HV* stash);
683
06f6df17
RGS
684The C<sv> argument must be a reference value. The C<stash> argument
685specifies which class the reference will belong to. See
2ae324a7 686L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
687
688/* Still under construction */
689
ddd2cc91
DM
690The following function upgrades rv to reference if not already one.
691Creates a new SV for rv to point to. If C<classname> is non-null, the SV
692is blessed into the specified class. SV is returned.
cb1a09d0 693
08105a92 694 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 695
ddd2cc91
DM
696The following three functions copy integer, unsigned integer or double
697into an SV whose reference is C<rv>. SV is blessed if C<classname> is
698non-null.
cb1a09d0 699
08105a92 700 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 701 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 702 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 703
ddd2cc91
DM
704The following function copies the pointer value (I<the address, not the
705string!>) into an SV whose reference is rv. SV is blessed if C<classname>
706is non-null.
cb1a09d0 707
ddd2cc91 708 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
cb1a09d0 709
a9b0660e 710The following function copies a string into an SV whose reference is C<rv>.
ddd2cc91
DM
711Set length to 0 to let Perl calculate the string length. SV is blessed if
712C<classname> is non-null.
cb1a09d0 713
a9b0660e
KW
714 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
715 STRLEN length);
cb1a09d0 716
ddd2cc91
DM
717The following function tests whether the SV is blessed into the specified
718class. It does not check inheritance relationships.
9abd00ed 719
08105a92 720 int sv_isa(SV* sv, const char* name);
9abd00ed 721
ddd2cc91 722The following function tests whether the SV is a reference to a blessed object.
9abd00ed
GS
723
724 int sv_isobject(SV* sv);
725
ddd2cc91 726The following function tests whether the SV is derived from the specified
10e2eb10
FC
727class. SV can be either a reference to a blessed object or a string
728containing a class name. This is the function implementing the
ddd2cc91 729C<UNIVERSAL::isa> functionality.
9abd00ed 730
08105a92 731 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 732
00aadd71 733To check if you've got an object derived from a specific class you have
9abd00ed
GS
734to write:
735
736 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 737
5f05dabc 738=head2 Creating New Variables
cb1a09d0 739
5f05dabc
PP
740To create a new Perl variable with an undef value which can be accessed from
741your Perl script, use the following routines, depending on the variable type.
cb1a09d0 742
64ace3f8 743 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 744 AV* get_av("package::varname", GV_ADD);
6673a63c 745 HV* get_hv("package::varname", GV_ADD);
cb1a09d0 746
058a5f6c 747Notice the use of GV_ADD as the second parameter. The new variable can now
cb1a09d0
AD
748be set, using the routines appropriate to the data type.
749
5f05dabc 750There are additional macros whose values may be bitwise OR'ed with the
058a5f6c 751C<GV_ADD> argument to enable certain extra features. Those bits are:
cb1a09d0 752
9a68f1db
SB
753=over
754
755=item GV_ADDMULTI
756
757Marks the variable as multiply defined, thus preventing the:
758
759 Name <varname> used only once: possible typo
760
761warning.
762
9a68f1db
SB
763=item GV_ADDWARN
764
765Issues the warning:
766
767 Had to create <varname> unexpectedly
768
769if the variable did not exist before the function was called.
770
771=back
cb1a09d0 772
07fa94a1
JO
773If you do not specify a package name, the variable is created in the current
774package.
cb1a09d0 775
5f05dabc 776=head2 Reference Counts and Mortality
a0d0e21e 777
10e2eb10 778Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 779AVs, or HVs (xV for short in the following) start their life with a
55497cff 780reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 781then it will be destroyed and its memory made available for reuse.
55497cff
PP
782
783This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
784undef'ed or the last variable holding a reference to it is changed or
785overwritten. At the internal level, however, reference counts can be
55497cff
PP
786manipulated with the following macros:
787
788 int SvREFCNT(SV* sv);
5f05dabc 789 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
790 void SvREFCNT_dec(SV* sv);
791
792However, there is one other function which manipulates the reference
07fa94a1
JO
793count of its argument. The C<newRV_inc> function, you will recall,
794creates a reference to the specified argument. As a side effect,
795it increments the argument's reference count. If this is not what
796you want, use C<newRV_noinc> instead.
797
798For example, imagine you want to return a reference from an XSUB function.
799Inside the XSUB routine, you create an SV which initially has a reference
800count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
801This returns the reference as a new SV, but the reference count of the
802SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
803return the reference from the XSUB routine and forget about the SV.
804But Perl hasn't! Whenever the returned reference is destroyed, the
805reference count of the original SV is decreased to one and nothing happens.
806The SV will hang around without any way to access it until Perl itself
807terminates. This is a memory leak.
5f05dabc
PP
808
809The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
810C<newRV_inc>. Then, if and when the last reference is destroyed,
811the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 812stopping any memory leak.
55497cff 813
5f05dabc 814There are some convenience functions available that can help with the
54310121 815destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
816An xV that is mortal has had its reference count marked to be decremented,
817but not actually decremented, until "a short time later". Generally the
818term "short time later" means a single Perl statement, such as a call to
54310121 819an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
820reference count decremented depends on two macros, SAVETMPS and FREETMPS.
821See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
822
823"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
824However, if you mortalize a variable twice, the reference count will
825later be decremented twice.
826
00aadd71
NIS
827"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
828For example an SV which is created just to pass a number to a called sub
06f6df17 829is made mortal to have it cleaned up automatically when it's popped off
10e2eb10 830the stack. Similarly, results returned by XSUBs (which are pushed on the
06f6df17 831stack) are often made mortal.
a0d0e21e
LW
832
833To create a mortal variable, use the functions:
834
835 SV* sv_newmortal()
836 SV* sv_2mortal(SV*)
837 SV* sv_mortalcopy(SV*)
838
00aadd71 839The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
840SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
841third creates a mortal copy of an existing SV.
da8c5729 842Because C<sv_newmortal> gives the new SV no value, it must normally be given one
9a68f1db 843via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
844
845 SV *tmp = sv_newmortal();
846 sv_setiv(tmp, an_integer);
847
848As that is multiple C statements it is quite common so see this idiom instead:
849
850 SV *tmp = sv_2mortal(newSViv(an_integer));
851
852
853You should be careful about creating mortal variables. Strange things
854can happen if you make the same value mortal within multiple contexts,
10e2eb10
FC
855or if you make a variable mortal multiple
856times. Thinking of "Mortalization"
00aadd71 857as deferred C<SvREFCNT_dec> should help to minimize such problems.
da8c5729 858For example if you are passing an SV which you I<know> has a high enough REFCNT
00aadd71
NIS
859to survive its use on the stack you need not do any mortalization.
860If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
861making a C<sv_mortalcopy> is safer.
a0d0e21e 862
ac036724 863The mortal routines are not just for SVs; AVs and HVs can be
faed5253 864made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 865C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 866
5f05dabc 867=head2 Stashes and Globs
a0d0e21e 868
06f6df17
RGS
869A B<stash> is a hash that contains all variables that are defined
870within a package. Each key of the stash is a symbol
aa689395
PP
871name (shared by all the different types of objects that have the same
872name), and each value in the hash table is a GV (Glob Value). This GV
873in turn contains references to the various objects of that name,
874including (but not limited to) the following:
cb1a09d0 875
a0d0e21e
LW
876 Scalar Value
877 Array Value
878 Hash Value
a3cb178b 879 I/O Handle
a0d0e21e
LW
880 Format
881 Subroutine
882
06f6df17
RGS
883There is a single stash called C<PL_defstash> that holds the items that exist
884in the C<main> package. To get at the items in other packages, append the
885string "::" to the package name. The items in the C<Foo> package are in
886the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
887in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 888
d1b91892 889To get the stash pointer for a particular package, use the function:
a0d0e21e 890
da51bb9b
NC
891 HV* gv_stashpv(const char* name, I32 flags)
892 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
893
894The first function takes a literal string, the second uses the string stored
d1b91892 895in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 896C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
897
898The name that C<gv_stash*v> wants is the name of the package whose symbol table
899you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
900packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
901language itself.
a0d0e21e
LW
902
903Alternately, if you have an SV that is a blessed reference, you can find
904out the stash pointer by using:
905
906 HV* SvSTASH(SvRV(SV*));
907
908then use the following to get the package name itself:
909
910 char* HvNAME(HV* stash);
911
5f05dabc
PP
912If you need to bless or re-bless an object you can use the following
913function:
a0d0e21e
LW
914
915 SV* sv_bless(SV*, HV* stash)
916
917where the first argument, an C<SV*>, must be a reference, and the second
918argument is a stash. The returned C<SV*> can now be used in the same way
919as any other SV.
920
d1b91892
AD
921For more information on references and blessings, consult L<perlref>.
922
54310121 923=head2 Double-Typed SVs
0a753a76
PP
924
925Scalar variables normally contain only one type of value, an integer,
926double, pointer, or reference. Perl will automatically convert the
927actual scalar data from the stored type into the requested type.
928
929Some scalar variables contain more than one type of scalar data. For
930example, the variable C<$!> contains either the numeric value of C<errno>
931or its string equivalent from either C<strerror> or C<sys_errlist[]>.
932
933To force multiple data values into an SV, you must do two things: use the
934C<sv_set*v> routines to add the additional scalar type, then set a flag
935so that Perl will believe it contains more than one type of data. The
936four macros to set the flags are:
937
938 SvIOK_on
939 SvNOK_on
940 SvPOK_on
941 SvROK_on
942
943The particular macro you must use depends on which C<sv_set*v> routine
944you called first. This is because every C<sv_set*v> routine turns on
945only the bit for the particular type of data being set, and turns off
946all the rest.
947
948For example, to create a new Perl variable called "dberror" that contains
949both the numeric and descriptive string error values, you could use the
950following code:
951
952 extern int dberror;
953 extern char *dberror_list;
954
64ace3f8 955 SV* sv = get_sv("dberror", GV_ADD);
0a753a76
PP
956 sv_setiv(sv, (IV) dberror);
957 sv_setpv(sv, dberror_list[dberror]);
958 SvIOK_on(sv);
959
960If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
961macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
962
4f4531b8
FC
963=head2 Read-Only Values
964
965In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
966flag bit with read-only scalars. So the only way to test whether
967C<sv_setsv>, etc., will raise a "Modification of a read-only value" error
968in those versions is:
969
970 SvREADONLY(sv) && !SvIsCOW(sv)
971
972Under Perl 5.18 and later, SvREADONLY only applies to read-only variables,
973and, under 5.20, copy-on-write scalars can also be read-only, so the above
974check is incorrect. You just want:
975
976 SvREADONLY(sv)
977
978If you need to do this check often, define your own macro like this:
979
980 #if PERL_VERSION >= 18
981 # define SvTRULYREADONLY(sv) SvREADONLY(sv)
982 #else
983 # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
984 #endif
985
986=head2 Copy on Write
987
988Perl implements a copy-on-write (COW) mechanism for scalars, in which
989string copies are not immediately made when requested, but are deferred
990until made necessary by one or the other scalar changing. This is mostly
991transparent, but one must take care not to modify string buffers that are
992shared by multiple SVs.
993
994You can test whether an SV is using copy-on-write with C<SvIsCOW(sv)>.
995
996You 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).
997
998If you want to make the SV drop its string buffer, use
999C<sv_force_normal_flags(sv, SV_COW_DROP_PV)> or simply
1000C<sv_setsv(sv, NULL)>.
1001
1002All of these functions will croak on read-only scalars (see the previous
1003section for more on those).
1004
1005To test that your code is behaving correctly and not modifying COW buffers,
1006on systems that support L<mmap(2)> (i.e., Unix) you can configure perl with
1007C<-Accflags=-DPERL_DEBUG_READONLY_COW> and it will turn buffer violations
1008into crashes. You will find it to be marvellously slow, so you may want to
1009skip perl's own tests.
1010
0a753a76 1011=head2 Magic Variables
a0d0e21e 1012
d1b91892
AD
1013[This section still under construction. Ignore everything here. Post no
1014bills. Everything not permitted is forbidden.]
1015
d1b91892
AD
1016Any SV may be magical, that is, it has special features that a normal
1017SV does not have. These features are stored in the SV structure in a
5f05dabc 1018linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
1019
1020 struct magic {
1021 MAGIC* mg_moremagic;
1022 MGVTBL* mg_virtual;
1023 U16 mg_private;
1024 char mg_type;
1025 U8 mg_flags;
b205eb13 1026 I32 mg_len;
d1b91892
AD
1027 SV* mg_obj;
1028 char* mg_ptr;
d1b91892
AD
1029 };
1030
1031Note this is current as of patchlevel 0, and could change at any time.
1032
1033=head2 Assigning Magic
1034
1035Perl adds magic to an SV using the sv_magic function:
1036
a9b0660e 1037 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
1038
1039The C<sv> argument is a pointer to the SV that is to acquire a new magical
1040feature.
1041
1042If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
10e2eb10
FC
1043convert C<sv> to type C<SVt_PVMG>.
1044Perl then continues by adding new magic
645c22ef
DM
1045to the beginning of the linked list of magical features. Any prior entry
1046of the same type of magic is deleted. Note that this can be overridden,
1047and multiple instances of the same type of magic can be associated with an
1048SV.
d1b91892 1049
54310121 1050The C<name> and C<namlen> arguments are used to associate a string with
10e2eb10 1051the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
1052C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
1053C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
1054whether C<namlen> is greater than zero or equal to zero respectively. As a
1055special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
1056to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
1057
1058The sv_magic function uses C<how> to determine which, if any, predefined
1059"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 1060See the L<Magic Virtual Tables> section below. The C<how> argument is also
10e2eb10
FC
1061stored in the C<mg_type> field. The value of
1062C<how> should be chosen from the set of macros
1063C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 1064these macros were added, Perl internals used to directly use character
14befaf4 1065literals, so you may occasionally come across old code or documentation
75d0f26d 1066referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
1067
1068The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
1069structure. If it is not the same as the C<sv> argument, the reference
1070count of the C<obj> object is incremented. If it is the same, or if
645c22ef 1071the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 1072then C<obj> is merely stored, without the reference count being incremented.
d1b91892 1073
2d8d5d5a
SH
1074See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
1075to an SV.
1076
cb1a09d0
AD
1077There is also a function to add magic to an C<HV>:
1078
1079 void hv_magic(HV *hv, GV *gv, int how);
1080
1081This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
1082
1083To remove the magic from an SV, call the function sv_unmagic:
1084
70a53b35 1085 int sv_unmagic(SV *sv, int type);
cb1a09d0
AD
1086
1087The C<type> argument should be equal to the C<how> value when the C<SV>
1088was initially made magical.
1089
f6ee7b17 1090However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
10e2eb10
FC
1091C<SV>. If you want to remove only certain
1092magic of a C<type> based on the magic
f6ee7b17
FR
1093virtual table, use C<sv_unmagicext> instead:
1094
1095 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1096
d1b91892
AD
1097=head2 Magic Virtual Tables
1098
d1be9408 1099The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
1100C<MGVTBL>, which is a structure of function pointers and stands for
1101"Magic Virtual Table" to handle the various operations that might be
1102applied to that variable.
1103
301cb7e8
DM
1104The C<MGVTBL> has five (or sometimes eight) pointers to the following
1105routine types:
d1b91892
AD
1106
1107 int (*svt_get)(SV* sv, MAGIC* mg);
1108 int (*svt_set)(SV* sv, MAGIC* mg);
1109 U32 (*svt_len)(SV* sv, MAGIC* mg);
1110 int (*svt_clear)(SV* sv, MAGIC* mg);
1111 int (*svt_free)(SV* sv, MAGIC* mg);
1112
a9b0660e
KW
1113 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1114 const char *name, I32 namlen);
301cb7e8
DM
1115 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1116 int (*svt_local)(SV *nsv, MAGIC *mg);
1117
1118
06f6df17 1119This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
SM
1120currently 32 types. These different structures contain pointers to various
1121routines that perform additional actions depending on which function is
1122being called.
d1b91892 1123
a9b0660e
KW
1124 Function pointer Action taken
1125 ---------------- ------------
1126 svt_get Do something before the value of the SV is
1127 retrieved.
1128 svt_set Do something after the SV is assigned a value.
1129 svt_len Report on the SV's length.
1130 svt_clear Clear something the SV represents.
1131 svt_free Free any extra storage associated with the SV.
d1b91892 1132
a9b0660e
KW
1133 svt_copy copy tied variable magic to a tied element
1134 svt_dup duplicate a magic structure during thread cloning
1135 svt_local copy magic to local value during 'local'
301cb7e8 1136
d1b91892 1137For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1138to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1139
1140 { magic_get, magic_set, magic_len, 0, 0 }
1141
14befaf4
DM
1142Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1143if a get operation is being performed, the routine C<magic_get> is
1144called. All the various routines for the various magical types begin
1145with C<magic_>. NOTE: the magic routines are not considered part of
1146the Perl API, and may not be exported by the Perl library.
d1b91892 1147
301cb7e8
DM
1148The last three slots are a recent addition, and for source code
1149compatibility they are only checked for if one of the three flags
10e2eb10
FC
1150MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1151This means that most code can continue declaring
1152a vtable as a 5-element value. These three are
301cb7e8
DM
1153currently used exclusively by the threading code, and are highly subject
1154to change.
1155
d1b91892
AD
1156The current kinds of Magic Virtual Tables are:
1157
f1f5ddd7
FC
1158=for comment
1159This table is generated by regen/mg_vtable.pl. Any changes made here
1160will be lost.
1161
1162=for mg_vtable.pl begin
1163
a9b0660e 1164 mg_type
bd6e6c12
FC
1165 (old-style char and macro) MGVTBL Type of magic
1166 -------------------------- ------ -------------
1167 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1168 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
e5e1ee61 1169 % PERL_MAGIC_rhash (none) Extra data for restricted
bd6e6c12 1170 hashes
a6d69523
TC
1171 * PERL_MAGIC_debugvar vtbl_debugvar $DB::single, signal, trace
1172 vars
bd6e6c12 1173 . PERL_MAGIC_pos vtbl_pos pos() lvalue
e5e1ee61 1174 : PERL_MAGIC_symtab (none) Extra data for symbol
bd6e6c12 1175 tables
e5e1ee61
FC
1176 < PERL_MAGIC_backref vtbl_backref For weak ref data
1177 @ PERL_MAGIC_arylen_p (none) To move arylen out of XPVAV
bd6e6c12
FC
1178 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1179 (fast string search)
1180 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1181 (AMT) on stash
1182 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1183 (@+ and @- vars)
1184 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1185 element
1186 E PERL_MAGIC_env vtbl_env %ENV hash
1187 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
eccba044 1188 f PERL_MAGIC_fm vtbl_regexp Formline
bd6e6c12 1189 ('compiled' format)
bd6e6c12
FC
1190 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1191 H PERL_MAGIC_hints vtbl_hints %^H hash
1192 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1193 I PERL_MAGIC_isa vtbl_isa @ISA array
1194 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1195 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1196 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1197 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1198 element
1199 N PERL_MAGIC_shared (none) Shared between threads
1200 n PERL_MAGIC_shared_scalar (none) Shared between threads
1201 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1202 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1203 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1204 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
e5e1ee61 1205 r PERL_MAGIC_qr vtbl_regexp Precompiled qr// regex
bd6e6c12
FC
1206 S PERL_MAGIC_sig (none) %SIG hash
1207 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1208 t PERL_MAGIC_taint vtbl_taint Taintedness
1209 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1210 extensions
1211 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1212 extensions
4499db73 1213 V PERL_MAGIC_vstring (none) SV was vstring literal
bd6e6c12
FC
1214 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1215 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1216 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1217 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1218 variable / smart parameter
1219 vivification
baabe3fb
FC
1220 \ PERL_MAGIC_lvref vtbl_lvref Lvalue reference
1221 constructor
e5e1ee61 1222 ] PERL_MAGIC_checkcall vtbl_checkcall Inlining/mutation of call
bd6e6c12
FC
1223 to this CV
1224 ~ PERL_MAGIC_ext (none) Available for use by
1225 extensions
0cbee0a4 1226
f1f5ddd7 1227=for mg_vtable.pl end
d1b91892 1228
68dc0745 1229When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1230uppercase letter is typically used to represent some kind of composite type
1231(a list or a hash), and the lowercase letter is used to represent an element
10e2eb10 1232of that composite type. Some internals code makes use of this case
92f0c265 1233relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1234
1235The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1236specifically for use by extensions and will not be used by perl itself.
1237Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1238to variables (typically objects). This is especially useful because
1239there is no way for normal perl code to corrupt this private information
1240(unlike using extra elements of a hash object).
1241
1242Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1243C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1244C<mg_ptr> field points to a C<ufuncs> structure:
1245
1246 struct ufuncs {
a9402793
AB
1247 I32 (*uf_val)(pTHX_ IV, SV*);
1248 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1249 IV uf_index;
1250 };
1251
1252When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1253function will be called with C<uf_index> as the first arg and a pointer to
1254the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1255magic is shown below. Note that the ufuncs structure is copied by
1256sv_magic, so you can safely allocate it on the stack.
1257
1258 void
1259 Umagic(sv)
1260 SV *sv;
1261 PREINIT:
1262 struct ufuncs uf;
1263 CODE:
1264 uf.uf_val = &my_get_fn;
1265 uf.uf_set = &my_set_fn;
1266 uf.uf_index = 0;
14befaf4 1267 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1268
1e73acc8
AS
1269Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1270
1271For hashes there is a specialized hook that gives control over hash
1272keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1273if the "set" function in the C<ufuncs> structure is NULL. The hook
1274is activated whenever the hash is accessed with a key specified as
1275an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1276C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1277through the functions without the C<..._ent> suffix circumvents the
4509d391 1278hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1e73acc8 1279
14befaf4
DM
1280Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1281or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1282extra care to avoid conflict. Typically only using the magic on
1283objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1284For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1285C<MGVTBL>, even if all its fields will be C<0>, so that individual
1286C<MAGIC> pointers can be identified as a particular kind of magic
10e2eb10 1287using their magic virtual table. C<mg_findext> provides an easy way
f6ee7b17 1288to do that:
2f07f21a
FR
1289
1290 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1291
1292 MAGIC *mg;
f6ee7b17
FR
1293 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1294 /* this is really ours, not another module's PERL_MAGIC_ext */
1295 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1296 ...
2f07f21a 1297 }
5f05dabc 1298
ef50df4b
GS
1299Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1300earlier do B<not> invoke 'set' magic on their targets. This must
1301be done by the user either by calling the C<SvSETMAGIC()> macro after
1302calling these functions, or by using one of the C<sv_set*_mg()> or
1303C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1304C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1305obtained from external sources in functions that don't handle magic.
4a4eefd0 1306See L<perlapi> for a description of these functions.
189b2af5
GS
1307For example, calls to the C<sv_cat*()> functions typically need to be
1308followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1309since their implementation handles 'get' magic.
1310
d1b91892
AD
1311=head2 Finding Magic
1312
a9b0660e
KW
1313 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1314 * type */
f6ee7b17
FR
1315
1316This routine returns a pointer to a C<MAGIC> structure stored in the SV.
10e2eb10
FC
1317If the SV does not have that magical
1318feature, C<NULL> is returned. If the
f6ee7b17 1319SV has multiple instances of that magical feature, the first one will be
10e2eb10
FC
1320returned. C<mg_findext> can be used
1321to find a C<MAGIC> structure of an SV
da8c5729 1322based on both its magic type and its magic virtual table:
f6ee7b17
FR
1323
1324 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
d1b91892 1325
f6ee7b17
FR
1326Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1327SVt_PVMG, Perl may core dump.
d1b91892 1328
08105a92 1329 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1330
1331This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1332field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1333the mg_type field is changed to be the lowercase letter.
a0d0e21e 1334
04343c6d
GS
1335=head2 Understanding the Magic of Tied Hashes and Arrays
1336
14befaf4
DM
1337Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1338magic type.
9edb2b46
GS
1339
1340WARNING: As of the 5.004 release, proper usage of the array and hash
1341access functions requires understanding a few caveats. Some
1342of these caveats are actually considered bugs in the API, to be fixed
10e2eb10 1343in later releases, and are bracketed with [MAYCHANGE] below. If
9edb2b46
GS
1344you find yourself actually applying such information in this section, be
1345aware that the behavior may change in the future, umm, without warning.
04343c6d 1346
1526ead6 1347The perl tie function associates a variable with an object that implements
9a68f1db 1348the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1349tie function from an XSUB, you must mimic this behaviour. The code below
1350carries out the necessary steps - firstly it creates a new hash, and then
1351creates a second hash which it blesses into the class which will implement
10e2eb10 1352the tie methods. Lastly it ties the two hashes together, and returns a
1526ead6
AB
1353reference to the new tied hash. Note that the code below does NOT call the
1354TIEHASH method in the MyTie class -
1355see L<Calling Perl Routines from within C Programs> for details on how
1356to do this.
1357
1358 SV*
1359 mytie()
1360 PREINIT:
1361 HV *hash;
1362 HV *stash;
1363 SV *tie;
1364 CODE:
1365 hash = newHV();
1366 tie = newRV_noinc((SV*)newHV());
da51bb9b 1367 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1368 sv_bless(tie, stash);
899e16d0 1369 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1370 RETVAL = newRV_noinc(hash);
1371 OUTPUT:
1372 RETVAL
1373
04343c6d
GS
1374The C<av_store> function, when given a tied array argument, merely
1375copies the magic of the array onto the value to be "stored", using
1376C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1377actually need to be stored in the array. [MAYCHANGE] After a call to
1378C<av_store> on a tied array, the caller will usually need to call
1379C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1380TIEARRAY object. If C<av_store> did return NULL, a call to
1381C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1382leak. [/MAYCHANGE]
04343c6d
GS
1383
1384The previous paragraph is applicable verbatim to tied hash access using the
1385C<hv_store> and C<hv_store_ent> functions as well.
1386
1387C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1388C<hv_fetch_ent> actually return an undefined mortal value whose magic
1389has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1390need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1391need to call C<mg_get()> on the returned value in order to actually invoke
1392the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1393you may also call C<mg_set()> on the return value after possibly assigning
1394a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1395method on the TIE object. [/MAYCHANGE]
04343c6d 1396
9edb2b46 1397[MAYCHANGE]
04343c6d
GS
1398In other words, the array or hash fetch/store functions don't really
1399fetch and store actual values in the case of tied arrays and hashes. They
1400merely call C<mg_copy> to attach magic to the values that were meant to be
1401"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1402do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1403the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1404and hashes.
1405
1406Currently (as of perl version 5.004), use of the hash and array access
1407functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1408"normal" hashes and arrays, or on their tied variants. The API may be
1409changed to provide more transparent access to both tied and normal data
1410types in future versions.
1411[/MAYCHANGE]
04343c6d
GS
1412
1413You would do well to understand that the TIEARRAY and TIEHASH interfaces
1414are mere sugar to invoke some perl method calls while using the uniform hash
1415and array syntax. The use of this sugar imposes some overhead (typically
1416about two to four extra opcodes per FETCH/STORE operation, in addition to
1417the creation of all the mortal variables required to invoke the methods).
1418This overhead will be comparatively small if the TIE methods are themselves
1419substantial, but if they are only a few statements long, the overhead
1420will not be insignificant.
1421
d1c897a1
IZ
1422=head2 Localizing changes
1423
1424Perl has a very handy construction
1425
1426 {
1427 local $var = 2;
1428 ...
1429 }
1430
1431This construction is I<approximately> equivalent to
1432
1433 {
1434 my $oldvar = $var;
1435 $var = 2;
1436 ...
1437 $var = $oldvar;
1438 }
1439
1440The biggest difference is that the first construction would
1441reinstate the initial value of $var, irrespective of how control exits
10e2eb10 1442the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1443more efficient as well.
1444
1445There is a way to achieve a similar task from C via Perl API: create a
1446I<pseudo-block>, and arrange for some changes to be automatically
1447undone at the end of it, either explicit, or via a non-local exit (via
10e2eb10 1448die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1449C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1450Such a construct may be created specially for some important localized
1451task, or an existing one (like boundaries of enclosing Perl
1452subroutine/block, or an existing pair for freeing TMPs) may be
10e2eb10
FC
1453used. (In the second case the overhead of additional localization must
1454be almost negligible.) Note that any XSUB is automatically enclosed in
b687b08b 1455an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1456
1457Inside such a I<pseudo-block> the following service is available:
1458
13a2d996 1459=over 4
d1c897a1
IZ
1460
1461=item C<SAVEINT(int i)>
1462
1463=item C<SAVEIV(IV i)>
1464
1465=item C<SAVEI32(I32 i)>
1466
1467=item C<SAVELONG(long i)>
1468
1469These macros arrange things to restore the value of integer variable
1470C<i> at the end of enclosing I<pseudo-block>.
1471
1472=item C<SAVESPTR(s)>
1473
1474=item C<SAVEPPTR(p)>
1475
1476These macros arrange things to restore the value of pointers C<s> and
10e2eb10 1477C<p>. C<s> must be a pointer of a type which survives conversion to
d1c897a1
IZ
1478C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1479and back.
1480
1481=item C<SAVEFREESV(SV *sv)>
1482
1483The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1484I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1485mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1486extends the lifetime of C<sv> until the beginning of the next statement,
1487C<SAVEFREESV> extends it until the end of the enclosing scope. These
1488lifetimes can be wildly different.
1489
1490Also compare C<SAVEMORTALIZESV>.
1491
1492=item C<SAVEMORTALIZESV(SV *sv)>
1493
1494Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1495scope instead of decrementing its reference count. This usually has the
1496effect of keeping C<sv> alive until the statement that called the currently
1497live scope has finished executing.
d1c897a1
IZ
1498
1499=item C<SAVEFREEOP(OP *op)>
1500
1501The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1502
1503=item C<SAVEFREEPV(p)>
1504
1505The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1506end of I<pseudo-block>.
1507
1508=item C<SAVECLEARSV(SV *sv)>
1509
1510Clears a slot in the current scratchpad which corresponds to C<sv> at
1511the end of I<pseudo-block>.
1512
1513=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1514
10e2eb10 1515The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
d1c897a1
IZ
1516string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1517short-lived storage, the corresponding string may be reallocated like
1518this:
1519
9cde0e7f 1520 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1521
c76ac1ee 1522=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1523
1524At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1525only argument C<p>.
1526
1527=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1528
1529At the end of I<pseudo-block> the function C<f> is called with the
1530implicit context argument (if any), and C<p>.
d1c897a1
IZ
1531
1532=item C<SAVESTACK_POS()>
1533
1534The current offset on the Perl internal stack (cf. C<SP>) is restored
1535at the end of I<pseudo-block>.
1536
1537=back
1538
1539The following API list contains functions, thus one needs to
1540provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1541or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1542function takes C<int *>.
1543
13a2d996 1544=over 4
d1c897a1
IZ
1545
1546=item C<SV* save_scalar(GV *gv)>
1547
1548Equivalent to Perl code C<local $gv>.
1549
1550=item C<AV* save_ary(GV *gv)>
1551
1552=item C<HV* save_hash(GV *gv)>
1553
1554Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1555
1556=item C<void save_item(SV *item)>
1557
1558Duplicates the current value of C<SV>, on the exit from the current
1559C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
10e2eb10 1560using the stored value. It doesn't handle magic. Use C<save_scalar> if
038fcae3 1561magic is affected.
d1c897a1
IZ
1562
1563=item C<void save_list(SV **sarg, I32 maxsarg)>
1564
1565A variant of C<save_item> which takes multiple arguments via an array
1566C<sarg> of C<SV*> of length C<maxsarg>.
1567
1568=item C<SV* save_svref(SV **sptr)>
1569
d1be9408 1570Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1571
1572=item C<void save_aptr(AV **aptr)>
1573
1574=item C<void save_hptr(HV **hptr)>
1575
1576Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1577
1578=back
1579
1580The C<Alias> module implements localization of the basic types within the
1581I<caller's scope>. People who are interested in how to localize things in
1582the containing scope should take a look there too.
1583
0a753a76 1584=head1 Subroutines
a0d0e21e 1585
68dc0745 1586=head2 XSUBs and the Argument Stack
5f05dabc
PP
1587
1588The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1589An XSUB routine will have a stack that contains the arguments from the Perl
1590program, and a way to map from the Perl data structures to a C equivalent.
1591
1592The stack arguments are accessible through the C<ST(n)> macro, which returns
1593the C<n>'th stack argument. Argument 0 is the first argument passed in the
1594Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1595an C<SV*> is used.
1596
1597Most of the time, output from the C routine can be handled through use of
1598the RETVAL and OUTPUT directives. However, there are some cases where the
1599argument stack is not already long enough to handle all the return values.
1600An example is the POSIX tzname() call, which takes no arguments, but returns
1601two, the local time zone's standard and summer time abbreviations.
1602
1603To handle this situation, the PPCODE directive is used and the stack is
1604extended using the macro:
1605
924508f0 1606 EXTEND(SP, num);
5f05dabc 1607
924508f0
GS
1608where C<SP> is the macro that represents the local copy of the stack pointer,
1609and C<num> is the number of elements the stack should be extended by.
5f05dabc 1610
00aadd71 1611Now that there is room on the stack, values can be pushed on it using C<PUSHs>
10e2eb10 1612macro. The pushed values will often need to be "mortal" (See
d82b684c 1613L</Reference Counts and Mortality>):
5f05dabc 1614
00aadd71 1615 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1616 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1617 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1618 PUSHs(sv_2mortal(newSVpv("Some String",0)))
a9b0660e
KW
1619 /* Although the last example is better written as the more
1620 * efficient: */
a3179684 1621 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
5f05dabc
PP
1622
1623And now the Perl program calling C<tzname>, the two values will be assigned
1624as in:
1625
1626 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1627
1628An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1629to use the macro:
5f05dabc 1630
5f05dabc
PP
1631 XPUSHs(SV*)
1632
da8c5729 1633This macro automatically adjusts the stack for you, if needed. Thus, you
5f05dabc 1634do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1635
1636Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1637C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1638For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1639C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc
PP
1640
1641For more information, consult L<perlxs> and L<perlxstut>.
1642
5b36e945
FC
1643=head2 Autoloading with XSUBs
1644
1645If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1646fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1647of the XSUB's package.
1648
1649But it also puts the same information in certain fields of the XSUB itself:
1650
1651 HV *stash = CvSTASH(cv);
1652 const char *subname = SvPVX(cv);
1653 STRLEN name_length = SvCUR(cv); /* in bytes */
1654 U32 is_utf8 = SvUTF8(cv);
f703fc96 1655
5b36e945 1656C<SvPVX(cv)> contains just the sub name itself, not including the package.
d8893903
FC
1657For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1658C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
5b36e945
FC
1659
1660B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1661XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1662XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1663to support 5.8-5.14, use the XSUB's fields.
1664
5f05dabc 1665=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1666
1667There are four routines that can be used to call a Perl subroutine from
1668within a C program. These four are:
1669
954c1994
GS
1670 I32 call_sv(SV*, I32);
1671 I32 call_pv(const char*, I32);
1672 I32 call_method(const char*, I32);
5aaab254 1673 I32 call_argv(const char*, I32, char**);
a0d0e21e 1674
954c1994 1675The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1676contains either the name of the Perl subroutine to be called, or a
1677reference to the subroutine. The second argument consists of flags
1678that control the context in which the subroutine is called, whether
1679or not the subroutine is being passed arguments, how errors should be
1680trapped, and how to treat return values.
a0d0e21e
LW
1681
1682All four routines return the number of arguments that the subroutine returned
1683on the Perl stack.
1684
9a68f1db 1685These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1686but those names are now deprecated; macros of the same name are provided for
1687compatibility.
1688
1689When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1690must manipulate the Perl stack. These include the following macros and
1691functions:
a0d0e21e
LW
1692
1693 dSP
924508f0 1694 SP
a0d0e21e
LW
1695 PUSHMARK()
1696 PUTBACK
1697 SPAGAIN
1698 ENTER
1699 SAVETMPS
1700 FREETMPS
1701 LEAVE
1702 XPUSH*()
cb1a09d0 1703 POP*()
a0d0e21e 1704
5f05dabc
PP
1705For a detailed description of calling conventions from C to Perl,
1706consult L<perlcall>.
a0d0e21e 1707
8ebc5c01 1708=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1709
1710A lot of opcodes (this is an elementary operation in the internal perl
10e2eb10
FC
1711stack machine) put an SV* on the stack. However, as an optimization
1712the corresponding SV is (usually) not recreated each time. The opcodes
ce3d39e2
IZ
1713reuse specially assigned SVs (I<target>s) which are (as a corollary)
1714not constantly freed/created.
1715
0a753a76 1716Each of the targets is created only once (but see
ce3d39e2
IZ
1717L<Scratchpads and recursion> below), and when an opcode needs to put
1718an integer, a double, or a string on stack, it just sets the
1719corresponding parts of its I<target> and puts the I<target> on stack.
1720
1721The macro to put this target on stack is C<PUSHTARG>, and it is
1722directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1723others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1724
1bd1c0d5 1725Because the target is reused, you must be careful when pushing multiple
10e2eb10 1726values on the stack. The following code will not do what you think:
1bd1c0d5
SC
1727
1728 XPUSHi(10);
1729 XPUSHi(20);
1730
1731This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1732the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1733At the end of the operation, the stack does not contain the values 10
1734and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1735to 20.
1bd1c0d5 1736
d82b684c
SH
1737If you need to push multiple different values then you should either use
1738the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1739none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1740SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1741will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1742this a little easier to achieve by creating a new mortal for you (via
1743C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1744in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1745Thus, instead of writing this to "fix" the example above:
1746
1747 XPUSHs(sv_2mortal(newSViv(10)))
1748 XPUSHs(sv_2mortal(newSViv(20)))
1749
1750you can simply write:
1751
1752 mXPUSHi(10)
1753 mXPUSHi(20)
1754
1755On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1756need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1757macros can make use of the local variable C<TARG>. See also C<dTARGET>
1758and C<dXSTARG>.
1bd1c0d5 1759
8ebc5c01 1760=head2 Scratchpads
ce3d39e2 1761
54310121 1762The question remains on when the SVs which are I<target>s for opcodes
10e2eb10 1763are created. The answer is that they are created when the current
ac036724 1764unit--a subroutine or a file (for opcodes for statements outside of
10e2eb10 1765subroutines)--is compiled. During this time a special anonymous Perl
ac036724 1766array is created, which is called a scratchpad for the current unit.
ce3d39e2 1767
54310121 1768A scratchpad keeps SVs which are lexicals for the current unit and are
d777b41a
FC
1769targets for opcodes. A previous version of this document
1770stated that one can deduce that an SV lives on a scratchpad
ce3d39e2 1771by looking on its flags: lexicals have C<SVs_PADMY> set, and
eee3e302 1772I<target>s have C<SVs_PADTMP> set. But this has never been fully true.
d777b41a
FC
1773C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
1774While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
eed77337
FC
1775that have never resided in a pad, but nonetheless act like I<target>s. As
1776of perl 5.21.5, the C<SVs_PADMY> flag is no longer used and is defined as
17770. C<SvPADMY()> now returns true for anything without C<SVs_PADTMP>.
ce3d39e2 1778
10e2eb10 1779The correspondence between OPs and I<target>s is not 1-to-1. Different
54310121 1780OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1781would not conflict with the expected life of the temporary.
1782
2ae324a7 1783=head2 Scratchpads and recursion
ce3d39e2
IZ
1784
1785In fact it is not 100% true that a compiled unit contains a pointer to
10e2eb10
FC
1786the scratchpad AV. In fact it contains a pointer to an AV of
1787(initially) one element, and this element is the scratchpad AV. Why do
ce3d39e2
IZ
1788we need an extra level of indirection?
1789
10e2eb10 1790The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2 1791these can create several execution pointers going into the same
10e2eb10 1792subroutine. For the subroutine-child not write over the temporaries
ce3d39e2
IZ
1793for the subroutine-parent (lifespan of which covers the call to the
1794child), the parent and the child should have different
10e2eb10 1795scratchpads. (I<And> the lexicals should be separate anyway!)
ce3d39e2 1796
5f05dabc
PP
1797So each subroutine is born with an array of scratchpads (of length 1).
1798On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1799depth of the recursion is not more than the length of this array, and
1800if it is, new scratchpad is created and pushed into the array.
1801
1802The I<target>s on this scratchpad are C<undef>s, but they are already
1803marked with correct flags.
1804
22d36020
FC
1805=head1 Memory Allocation
1806
1807=head2 Allocation
1808
1809All memory meant to be used with the Perl API functions should be manipulated
1810using the macros described in this section. The macros provide the necessary
1811transparency between differences in the actual malloc implementation that is
1812used within perl.
1813
1814It is suggested that you enable the version of malloc that is distributed
1815with Perl. It keeps pools of various sizes of unallocated memory in
1816order to satisfy allocation requests more quickly. However, on some
1817platforms, it may cause spurious malloc or free errors.
1818
1819The following three macros are used to initially allocate memory :
1820
1821 Newx(pointer, number, type);
1822 Newxc(pointer, number, type, cast);
1823 Newxz(pointer, number, type);
1824
1825The first argument C<pointer> should be the name of a variable that will
1826point to the newly allocated memory.
1827
1828The second and third arguments C<number> and C<type> specify how many of
1829the specified type of data structure should be allocated. The argument
1830C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
1831should be used if the C<pointer> argument is different from the C<type>
1832argument.
1833
1834Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
1835to zero out all the newly allocated memory.
1836
1837=head2 Reallocation
1838
1839 Renew(pointer, number, type);
1840 Renewc(pointer, number, type, cast);
1841 Safefree(pointer)
1842
1843These three macros are used to change a memory buffer size or to free a
1844piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1845match those of C<New> and C<Newc> with the exception of not needing the
1846"magic cookie" argument.
1847
1848=head2 Moving
1849
1850 Move(source, dest, number, type);
1851 Copy(source, dest, number, type);
1852 Zero(dest, number, type);
1853
1854These three macros are used to move, copy, or zero out previously allocated
1855memory. The C<source> and C<dest> arguments point to the source and
1856destination starting points. Perl will move, copy, or zero out C<number>
1857instances of the size of the C<type> data structure (using the C<sizeof>
1858function).
1859
1860=head1 PerlIO
1861
1862The most recent development releases of Perl have been experimenting with
1863removing Perl's dependency on the "normal" standard I/O suite and allowing
1864other stdio implementations to be used. This involves creating a new
1865abstraction layer that then calls whichever implementation of stdio Perl
1866was compiled with. All XSUBs should now use the functions in the PerlIO
1867abstraction layer and not make any assumptions about what kind of stdio
1868is being used.
1869
1870For a complete description of the PerlIO abstraction, consult L<perlapio>.
1871
0a753a76
PP
1872=head1 Compiled code
1873
1874=head2 Code tree
1875
1876Here we describe the internal form your code is converted to by
10e2eb10 1877Perl. Start with a simple example:
0a753a76
PP
1878
1879 $a = $b + $c;
1880
1881This is converted to a tree similar to this one:
1882
1883 assign-to
1884 / \
1885 + $a
1886 / \
1887 $b $c
1888
7b8d334a 1889(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1890parsed your code, but has nothing to do with the execution order.
1891There is an additional "thread" going through the nodes of the tree
1892which shows the order of execution of the nodes. In our simplified
1893example above it looks like:
1894
1895 $b ---> $c ---> + ---> $a ---> assign-to
1896
1897But with the actual compile tree for C<$a = $b + $c> it is different:
1898some nodes I<optimized away>. As a corollary, though the actual tree
1899contains more nodes than our simplified example, the execution order
1900is the same as in our example.
1901
1902=head2 Examining the tree
1903
06f6df17
RGS
1904If you have your perl compiled for debugging (usually done with
1905C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1906compiled tree by specifying C<-Dx> on the Perl command line. The
1907output takes several lines per node, and for C<$b+$c> it looks like
1908this:
1909
1910 5 TYPE = add ===> 6
1911 TARG = 1
1912 FLAGS = (SCALAR,KIDS)
1913 {
1914 TYPE = null ===> (4)
1915 (was rv2sv)
1916 FLAGS = (SCALAR,KIDS)
1917 {
1918 3 TYPE = gvsv ===> 4
1919 FLAGS = (SCALAR)
1920 GV = main::b
1921 }
1922 }
1923 {
1924 TYPE = null ===> (5)
1925 (was rv2sv)
1926 FLAGS = (SCALAR,KIDS)
1927 {
1928 4 TYPE = gvsv ===> 5
1929 FLAGS = (SCALAR)
1930 GV = main::c
1931 }
1932 }
1933
1934This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1935not optimized away (one per number in the left column). The immediate
1936children of the given node correspond to C<{}> pairs on the same level
1937of indentation, thus this listing corresponds to the tree:
1938
1939 add
1940 / \
1941 null null
1942 | |
1943 gvsv gvsv
1944
1945The execution order is indicated by C<===E<gt>> marks, thus it is C<3
19464 5 6> (node C<6> is not included into above listing), i.e.,
1947C<gvsv gvsv add whatever>.
1948
9afa14e3 1949Each of these nodes represents an op, a fundamental operation inside the
10e2eb10 1950Perl core. The code which implements each operation can be found in the
9afa14e3 1951F<pp*.c> files; the function which implements the op with type C<gvsv>
10e2eb10 1952is C<pp_gvsv>, and so on. As the tree above shows, different ops have
9afa14e3 1953different numbers of children: C<add> is a binary operator, as one would
10e2eb10 1954expect, and so has two children. To accommodate the various different
9afa14e3
SC
1955numbers of children, there are various types of op data structure, and
1956they link together in different ways.
1957
10e2eb10 1958The simplest type of op structure is C<OP>: this has no children. Unary
9afa14e3 1959operators, C<UNOP>s, have one child, and this is pointed to by the
10e2eb10
FC
1960C<op_first> field. Binary operators (C<BINOP>s) have not only an
1961C<op_first> field but also an C<op_last> field. The most complex type of
1962op is a C<LISTOP>, which has any number of children. In this case, the
9afa14e3 1963first child is pointed to by C<op_first> and the last child by
10e2eb10 1964C<op_last>. The children in between can be found by iteratively
3ef130ce 1965following the C<op_sibling> pointer from the first child to the last (but
29e61fd9 1966see below).
9afa14e3 1967
29e61fd9 1968There are also some other op types: a C<PMOP> holds a regular expression,
10e2eb10
FC
1969and has no children, and a C<LOOP> may or may not have children. If the
1970C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
9afa14e3
SC
1971complicate matters, if a C<UNOP> is actually a C<null> op after
1972optimization (see L</Compile pass 2: context propagation>) it will still
1973have children in accordance with its former type.
1974
29e61fd9
DM
1975Finally, there is a C<LOGOP>, or logic op. Like a C<LISTOP>, this has one
1976or more children, but it doesn't have an C<op_last> field: so you have to
1977follow C<op_first> and then the C<op_sibling> chain itself to find the
1978last child. Instead it has an C<op_other> field, which is comparable to
1979the C<op_next> field described below, and represents an alternate
1980execution path. Operators like C<and>, C<or> and C<?> are C<LOGOP>s. Note
1981that in general, C<op_other> may not point to any of the direct children
1982of the C<LOGOP>.
1983
1984Starting in version 5.21.2, perls built with the experimental
1985define C<-DPERL_OP_PARENT> add an extra boolean flag for each op,
1986C<op_lastsib>. When set, this indicates that this is the last op in an
1987C<op_sibling> chain. This frees up the C<op_sibling> field on the last
e6dae479 1988sibling to point back to the parent op. The macro C<OpSIBLING(o)> wraps
29e61fd9
DM
1989this special behaviour, and always returns NULL on the last sibling.
1990With this build the C<op_parent(o)> function can be used to find the
3ef130ce
DM
1991parent of any op. Thus for forward compatibility, you should always use
1992the C<OpSIBLING(o)> macro rather than accessing C<op_sibling> directly.
29e61fd9 1993
06f6df17
RGS
1994Another way to examine the tree is to use a compiler back-end module, such
1995as L<B::Concise>.
1996
0a753a76
PP
1997=head2 Compile pass 1: check routines
1998
8870b5c7 1999The tree is created by the compiler while I<yacc> code feeds it
10e2eb10 2000the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
2001the first pass of perl compilation.
2002
2003What makes this pass interesting for perl developers is that some
2004optimization may be performed on this pass. This is optimization by
8870b5c7 2005so-called "check routines". The correspondence between node names
0a753a76
PP
2006and corresponding check routines is described in F<opcode.pl> (do not
2007forget to run C<make regen_headers> if you modify this file).
2008
2009A check routine is called when the node is fully constructed except
7b8d334a 2010for the execution-order thread. Since at this time there are no
0a753a76
PP
2011back-links to the currently constructed node, one can do most any
2012operation to the top-level node, including freeing it and/or creating
2013new nodes above/below it.
2014
2015The check routine returns the node which should be inserted into the
2016tree (if the top-level node was not modified, check routine returns
2017its argument).
2018
10e2eb10 2019By convention, check routines have names C<ck_*>. They are usually
0a753a76
PP
2020called from C<new*OP> subroutines (or C<convert>) (which in turn are
2021called from F<perly.y>).
2022
2023=head2 Compile pass 1a: constant folding
2024
2025Immediately after the check routine is called the returned node is
2026checked for being compile-time executable. If it is (the value is
2027judged to be constant) it is immediately executed, and a I<constant>
2028node with the "return value" of the corresponding subtree is
2029substituted instead. The subtree is deleted.
2030
2031If constant folding was not performed, the execution-order thread is
2032created.
2033
2034=head2 Compile pass 2: context propagation
2035
2036When a context for a part of compile tree is known, it is propagated
a3cb178b 2037down through the tree. At this time the context can have 5 values
0a753a76
PP
2038(instead of 2 for runtime context): void, boolean, scalar, list, and
2039lvalue. In contrast with the pass 1 this pass is processed from top
2040to bottom: a node's context determines the context for its children.
2041
2042Additional context-dependent optimizations are performed at this time.
2043Since at this moment the compile tree contains back-references (via
2044"thread" pointers), nodes cannot be free()d now. To allow
2045optimized-away nodes at this stage, such nodes are null()ified instead
2046of free()ing (i.e. their type is changed to OP_NULL).
2047
2048=head2 Compile pass 3: peephole optimization
2049
2050After the compile tree for a subroutine (or for an C<eval> or a file)
10e2eb10 2051is created, an additional pass over the code is performed. This pass
0a753a76 2052is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
2053additional complications for conditionals). Optimizations performed
2054at this stage are subject to the same restrictions as in the pass 2.
2055
2056Peephole optimizations are done by calling the function pointed to
2057by the global variable C<PL_peepp>. By default, C<PL_peepp> just
2058calls the function pointed to by the global variable C<PL_rpeepp>.
2059By default, that performs some basic op fixups and optimisations along
2060the execution-order op chain, and recursively calls C<PL_rpeepp> for
2061each side chain of ops (resulting from conditionals). Extensions may
2062provide additional optimisations or fixups, hooking into either the
2063per-subroutine or recursive stage, like this:
2064
2065 static peep_t prev_peepp;
2066 static void my_peep(pTHX_ OP *o)
2067 {
2068 /* custom per-subroutine optimisation goes here */
f0358462 2069 prev_peepp(aTHX_ o);
9ea12537
Z
2070 /* custom per-subroutine optimisation may also go here */
2071 }
2072 BOOT:
2073 prev_peepp = PL_peepp;
2074 PL_peepp = my_peep;
2075
2076 static peep_t prev_rpeepp;
2077 static void my_rpeep(pTHX_ OP *o)
2078 {
2079 OP *orig_o = o;
2080 for(; o; o = o->op_next) {
2081 /* custom per-op optimisation goes here */
2082 }
f0358462 2083 prev_rpeepp(aTHX_ orig_o);
9ea12537
Z
2084 }
2085 BOOT:
2086 prev_rpeepp = PL_rpeepp;
2087 PL_rpeepp = my_rpeep;
0a753a76 2088
1ba7f851
PJ
2089=head2 Pluggable runops
2090
2091The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
2092functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
2093with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
2094control over the execution of the compile tree it is possible to provide
2095your own runops function.
1ba7f851
PJ
2096
2097It's probably best to copy one of the existing runops functions and
2098change it to suit your needs. Then, in the BOOT section of your XS
2099file, add the line:
2100
2101 PL_runops = my_runops;
2102
2103This function should be as efficient as possible to keep your programs
2104running as fast as possible.
2105
fd85fad2
BM
2106=head2 Compile-time scope hooks
2107
2108As of perl 5.14 it is possible to hook into the compile-time lexical
10e2eb10 2109scope mechanism using C<Perl_blockhook_register>. This is used like
fd85fad2
BM
2110this:
2111
2112 STATIC void my_start_hook(pTHX_ int full);
2113 STATIC BHK my_hooks;
2114
2115 BOOT:
a88d97bf 2116 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
2117 Perl_blockhook_register(aTHX_ &my_hooks);
2118
2119This will arrange to have C<my_start_hook> called at the start of
10e2eb10 2120compiling every lexical scope. The available hooks are:
fd85fad2
BM
2121
2122=over 4
2123
a88d97bf 2124=item C<void bhk_start(pTHX_ int full)>
fd85fad2 2125
10e2eb10 2126This is called just after starting a new lexical scope. Note that Perl
fd85fad2
BM
2127code like
2128
2129 if ($x) { ... }
2130
2131creates two scopes: the first starts at the C<(> and has C<full == 1>,
10e2eb10
FC
2132the second starts at the C<{> and has C<full == 0>. Both end at the
2133C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
fd85fad2
BM
2134pushed onto the save stack by this hook will be popped just before the
2135scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2136
a88d97bf 2137=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
2138
2139This is called at the end of a lexical scope, just before unwinding the
10e2eb10 2140stack. I<o> is the root of the optree representing the scope; it is a
fd85fad2
BM
2141double pointer so you can replace the OP if you need to.
2142
a88d97bf 2143=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
2144
2145This is called at the end of a lexical scope, just after unwinding the
10e2eb10 2146stack. I<o> is as above. Note that it is possible for calls to C<pre_>
fd85fad2
BM
2147and C<post_end> to nest, if there is something on the save stack that
2148calls string eval.
2149
a88d97bf 2150=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
2151
2152This is called just before starting to compile an C<eval STRING>, C<do
10e2eb10 2153FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
fd85fad2
BM
2154OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2155C<OP_DOFILE> or C<OP_REQUIRE>.
2156
2157=back
2158
2159Once you have your hook functions, you need a C<BHK> structure to put
10e2eb10
FC
2160them in. It's best to allocate it statically, since there is no way to
2161free it once it's registered. The function pointers should be inserted
fd85fad2 2162into this structure using the C<BhkENTRY_set> macro, which will also set
10e2eb10 2163flags indicating which entries are valid. If you do need to allocate
fd85fad2
BM
2164your C<BHK> dynamically for some reason, be sure to zero it before you
2165start.
2166
2167Once registered, there is no mechanism to switch these hooks off, so if
10e2eb10 2168that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
2169is probably the best way, so the effect is lexically scoped; however it
2170is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
10e2eb10 2171temporarily switch entries on and off. You should also be aware that
a3e07c87
BM
2172generally speaking at least one scope will have opened before your
2173extension is loaded, so you will see some C<pre/post_end> pairs that
2174didn't have a matching C<start>.
fd85fad2 2175
9afa14e3
SC
2176=head1 Examining internal data structures with the C<dump> functions
2177
2178To aid debugging, the source file F<dump.c> contains a number of
2179functions which produce formatted output of internal data structures.
2180
2181The most commonly used of these functions is C<Perl_sv_dump>; it's used
10e2eb10 2182for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
9afa14e3 2183C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 2184module should already be familiar with its format.
9afa14e3
SC
2185
2186C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 2187derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
2188C<Perl_dump_eval> will dump the main root of the code being evaluated,
2189exactly like C<-Dx>.
2190
2191Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2192op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2193subroutines in a package like so: (Thankfully, these are all xsubs, so
2194there is no op tree)
2195
2196 (gdb) print Perl_dump_packsubs(PL_defstash)
2197
2198 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2199
2200 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2201
2202 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2203
2204 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2205
2206 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2207
2208and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2209the op tree of the main root.
2210
954c1994 2211=head1 How multiple interpreters and concurrency are supported
ee072b34 2212
ee072b34
GS
2213=head2 Background and PERL_IMPLICIT_CONTEXT
2214
2215The Perl interpreter can be regarded as a closed box: it has an API
2216for feeding it code or otherwise making it do things, but it also has
2217functions for its own use. This smells a lot like an object, and
2218there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2219interpreters, with one interpreter represented either as a C structure,
2220or inside a thread-specific structure. These structures contain all
2221the context, the state of that interpreter.
2222
10e2eb10 2223One macro controls the major Perl build flavor: MULTIPLICITY. The
7b52221d 2224MULTIPLICITY build has a C structure that packages all the interpreter
10e2eb10 2225state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
7b52221d 2226normally defined, and enables the support for passing in a "hidden" first
10e2eb10 2227argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2228multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2229to the macro USE_ITHREADS.)
54aff467 2230
27da23d5
JH
2231Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2232PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2233former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2234internal variables of Perl to be wrapped inside a single global struct,
2235struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2236the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2237one step further, there is still a single struct (allocated in main()
2238either from heap or from stack) but there are no global data symbols
3bf17896 2239pointing to it. In either case the global struct should be initialized
27da23d5
JH
2240as the very first thing in main() using Perl_init_global_struct() and
2241correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2242please see F<miniperlmain.c> for usage details. You may also need
2243to use C<dVAR> in your coding to "declare the global variables"
2244when you are using them. dTHX does this for you automatically.
2245
9aa97215
JH
2246To see whether you have non-const data you can use a BSD (or GNU)
2247compatible C<nm>:
bc028b6b
JH
2248
2249 nm libperl.a | grep -v ' [TURtr] '
2250
9aa97215
JH
2251If this displays any C<D> or C<d> symbols (or possibly C<C> or C<c>),
2252you have non-const data. The symbols the C<grep> removed are as follows:
2253C<Tt> are I<text>, or code, the C<Rr> are I<read-only> (const) data,
2254and the C<U> is <undefined>, external symbols referred to.
2255
2256The test F<t/porting/libperl.t> does this kind of symbol sanity
2257checking on C<libperl.a>.
bc028b6b 2258
27da23d5
JH
2259For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2260doesn't actually hide all symbols inside a big global struct: some
2261PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2262then hides everything (see how the PERLIO_FUNCS_DECL is used).
2263
54aff467 2264All this obviously requires a way for the Perl internal functions to be
acfe0abc 2265either subroutines taking some kind of structure as the first
ee072b34 2266argument, or subroutines taking nothing as the first argument. To
acfe0abc 2267enable these two very different ways of building the interpreter,
ee072b34
GS
2268the Perl source (as it does in so many other situations) makes heavy
2269use of macros and subroutine naming conventions.
2270
54aff467 2271First problem: deciding which functions will be public API functions and
00aadd71 2272which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2273(think "S" for "secret" or "static"). All other functions begin with
2274"Perl_", but just because a function begins with "Perl_" does not mean it is
10e2eb10
FC
2275part of the API. (See L</Internal
2276Functions>.) The easiest way to be B<sure> a
00aadd71
NIS
2277function is part of the API is to find its entry in L<perlapi>.
2278If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2279think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2280L<perlbug> explaining why you think it should be.
ee072b34
GS
2281
2282Second problem: there must be a syntax so that the same subroutine
2283declarations and calls can pass a structure as their first argument,
2284or pass nothing. To solve this, the subroutines are named and
2285declared in a particular way. Here's a typical start of a static
2286function used within the Perl guts:
2287
2288 STATIC void
2289 S_incline(pTHX_ char *s)
2290
acfe0abc 2291STATIC becomes "static" in C, and may be #define'd to nothing in some
da8c5729 2292configurations in the future.
ee072b34 2293
651a3225
GS
2294A public function (i.e. part of the internal API, but not necessarily
2295sanctioned for use in extensions) begins like this:
ee072b34
GS
2296
2297 void
2307c6d0 2298 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2299
0147cd53 2300C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2301details of the interpreter's context. THX stands for "thread", "this",
2302or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2303The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2304or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2305their variants.
ee072b34 2306
a7486cbb
JH
2307When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2308first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2309in the pTHX_ macro indicates that the macro expansion needs a comma
2310after the context argument because other arguments follow it. If
2311PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2312subroutine is not prototyped to take the extra argument. The form of the
2313macro without the trailing underscore is used when there are no additional
ee072b34
GS
2314explicit arguments.
2315
54aff467 2316When a core function calls another, it must pass the context. This
2307c6d0 2317is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2318something like this:
2319
2307c6d0
SB
2320 #ifdef PERL_IMPLICIT_CONTEXT
2321 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2322 /* can't do this for vararg functions, see below */
2307c6d0
SB
2323 #else
2324 #define sv_setiv Perl_sv_setiv
2325 #endif
ee072b34
GS
2326
2327This works well, and means that XS authors can gleefully write:
2328
2307c6d0 2329 sv_setiv(foo, bar);
ee072b34
GS
2330
2331and still have it work under all the modes Perl could have been
2332compiled with.
2333
ee072b34
GS
2334This doesn't work so cleanly for varargs functions, though, as macros
2335imply that the number of arguments is known in advance. Instead we
2336either need to spell them out fully, passing C<aTHX_> as the first
2337argument (the Perl core tends to do this with functions like
2338Perl_warner), or use a context-free version.
2339
2340The context-free version of Perl_warner is called
2341Perl_warner_nocontext, and does not take the extra argument. Instead
2342it does dTHX; to get the context from thread-local storage. We
2343C<#define warner Perl_warner_nocontext> so that extensions get source
2344compatibility at the expense of performance. (Passing an arg is
2345cheaper than grabbing it from thread-local storage.)
2346
acfe0abc 2347You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2348Those are strictly for use within the core. Extensions and embedders
2349need only be aware of [pad]THX.
2350
a7486cbb
JH
2351=head2 So what happened to dTHR?
2352
2353C<dTHR> was introduced in perl 5.005 to support the older thread model.
2354The older thread model now uses the C<THX> mechanism to pass context
2355pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2356later still have it for backward source compatibility, but it is defined
2357to be a no-op.
2358
ee072b34
GS
2359=head2 How do I use all this in extensions?
2360
2361When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2362any functions in the Perl API will need to pass the initial context
2363argument somehow. The kicker is that you will need to write it in
2364such a way that the extension still compiles when Perl hasn't been
2365built with PERL_IMPLICIT_CONTEXT enabled.
2366
2367There are three ways to do this. First, the easy but inefficient way,
2368which is also the default, in order to maintain source compatibility
0147cd53 2369with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2370and aTHX_ macros to call a function that will return the context.
2371Thus, something like:
2372
2307c6d0 2373 sv_setiv(sv, num);
ee072b34 2374
4375e838 2375in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2376in effect:
ee072b34 2377
2307c6d0 2378 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2379
54aff467 2380or to this otherwise:
ee072b34 2381
2307c6d0 2382 Perl_sv_setiv(sv, num);
ee072b34 2383
da8c5729 2384You don't have to do anything new in your extension to get this; since
2fa86c13 2385the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2386work.
2387
2388The second, more efficient way is to use the following template for
2389your Foo.xs:
2390
c52f9dcd
JH
2391 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2392 #include "EXTERN.h"
2393 #include "perl.h"
2394 #include "XSUB.h"
ee072b34 2395
fd061412 2396 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2397
fd061412 2398 STATIC void
c52f9dcd
JH
2399 my_private_function(int arg1, int arg2)
2400 {
2401 dTHX; /* fetch context */
2402 ... call many Perl API functions ...
2403 }
ee072b34
GS
2404
2405 [... etc ...]
2406
c52f9dcd 2407 MODULE = Foo PACKAGE = Foo
ee072b34 2408
c52f9dcd 2409 /* typical XSUB */
ee072b34 2410
c52f9dcd
JH
2411 void
2412 my_xsub(arg)
2413 int arg
2414 CODE:
2415 my_private_function(arg, 10);
ee072b34
GS
2416
2417Note that the only two changes from the normal way of writing an
2418extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2419including the Perl headers, followed by a C<dTHX;> declaration at
2420the start of every function that will call the Perl API. (You'll
2421know which functions need this, because the C compiler will complain
2422that there's an undeclared identifier in those functions.) No changes
2423are needed for the XSUBs themselves, because the XS() macro is
2424correctly defined to pass in the implicit context if needed.
2425
2426The third, even more efficient way is to ape how it is done within
2427the Perl guts:
2428
2429
c52f9dcd
JH
2430 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2431 #include "EXTERN.h"
2432 #include "perl.h"
2433 #include "XSUB.h"
ee072b34
GS
2434
2435 /* pTHX_ only needed for functions that call Perl API */
fd061412 2436 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2437
fd061412 2438 STATIC void
c52f9dcd
JH
2439 my_private_function(pTHX_ int arg1, int arg2)
2440 {
2441 /* dTHX; not needed here, because THX is an argument */
2442 ... call Perl API functions ...
2443 }
ee072b34
GS
2444
2445 [... etc ...]
2446
c52f9dcd 2447 MODULE = Foo PACKAGE = Foo
ee072b34 2448
c52f9dcd 2449 /* typical XSUB */
ee072b34 2450
c52f9dcd
JH
2451 void
2452 my_xsub(arg)
2453 int arg
2454 CODE:
2455 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2456
2457This implementation never has to fetch the context using a function
2458call, since it is always passed as an extra argument. Depending on
2459your needs for simplicity or efficiency, you may mix the previous
2460two approaches freely.
2461
651a3225
GS
2462Never add a comma after C<pTHX> yourself--always use the form of the
2463macro with the underscore for functions that take explicit arguments,
2464or the form without the argument for functions with no explicit arguments.
ee072b34 2465
27da23d5
JH
2466If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2467definition is needed if the Perl global variables (see F<perlvars.h>
2468or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2469used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2470the need for C<dVAR> only with the said compile-time define, because
2471otherwise the Perl global variables are visible as-is.
2472
a7486cbb
JH
2473=head2 Should I do anything special if I call perl from multiple threads?
2474
2475If you create interpreters in one thread and then proceed to call them in
2476another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2477initialized correctly in each of those threads.
2478
2479The C<perl_alloc> and C<perl_clone> API functions will automatically set
2480the TLS slot to the interpreter they created, so that there is no need to do
2481anything special if the interpreter is always accessed in the same thread that
2482created it, and that thread did not create or call any other interpreters
2483afterwards. If that is not the case, you have to set the TLS slot of the
2484thread before calling any functions in the Perl API on that particular
2485interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2486thread as the first thing you do:
2487
2488 /* do this before doing anything else with some_perl */
2489 PERL_SET_CONTEXT(some_perl);
2490
2491 ... other Perl API calls on some_perl go here ...
2492
ee072b34
GS
2493=head2 Future Plans and PERL_IMPLICIT_SYS
2494
2495Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2496that the interpreter knows about itself and pass it around, so too are
2497there plans to allow the interpreter to bundle up everything it knows
2498about the environment it's running on. This is enabled with the
7b52221d
RGS
2499PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2500Windows.
ee072b34
GS
2501
2502This allows the ability to provide an extra pointer (called the "host"
2503environment) for all the system calls. This makes it possible for
2504all the system stuff to maintain their own state, broken down into
2505seven C structures. These are thin wrappers around the usual system
0147cd53 2506calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2507more ambitious host (like the one that would do fork() emulation) all
2508the extra work needed to pretend that different interpreters are
2509actually different "processes", would be done here.
2510
2511The Perl engine/interpreter and the host are orthogonal entities.
2512There could be one or more interpreters in a process, and one or
2513more "hosts", with free association between them.
2514
a422fd2d
SC
2515=head1 Internal Functions
2516
2517All of Perl's internal functions which will be exposed to the outside
06f6df17 2518world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d 2519functions or functions used in a program in which Perl is embedded.
10e2eb10 2520Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2521static functions start with C<S_>.)
a422fd2d 2522
0972ecdf
DM
2523Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2524either with or without the C<Perl_> prefix, thanks to a bunch of defines
10e2eb10 2525that live in F<embed.h>. Note that extension code should I<not> set
0972ecdf
DM
2526C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2527breakage of the XS in each new perl release.
2528
2529The file F<embed.h> is generated automatically from
10e2eb10 2530F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
dc9b1d22 2531header files for the internal functions, generates the documentation
10e2eb10 2532and a lot of other bits and pieces. It's important that when you add
dc9b1d22 2533a new function to the core or change an existing one, you change the
10e2eb10 2534data in the table in F<embed.fnc> as well. Here's a sample entry from
dc9b1d22 2535that table:
a422fd2d
SC
2536
2537 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2538
10e2eb10
FC
2539The second column is the return type, the third column the name. Columns
2540after that are the arguments. The first column is a set of flags:
a422fd2d
SC
2541
2542=over 3
2543
2544=item A
2545
10e2eb10
FC
2546This function is a part of the public
2547API. All such functions should also
1aa6ea50 2548have 'd', very few do not.
a422fd2d
SC
2549
2550=item p
2551
1aa6ea50
JC
2552This function has a C<Perl_> prefix; i.e. it is defined as
2553C<Perl_av_fetch>.
a422fd2d
SC
2554
2555=item d
2556
2557This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2558look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2559
2560=back
2561
2562Other available flags are:
2563
2564=over 3
2565
2566=item s
2567
1aa6ea50
JC
2568This is a static function and is defined as C<STATIC S_whatever>, and
2569usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2570
2571=item n
2572
da8c5729 2573This does not need an interpreter context, so the definition has no
1aa6ea50 2574C<pTHX>, and it follows that callers don't use C<aTHX>. (See
d3a43cd8 2575L</Background and PERL_IMPLICIT_CONTEXT>.)
a422fd2d
SC
2576
2577=item r
2578
2579This function never returns; C<croak>, C<exit> and friends.
2580
2581=item f
2582
2583This function takes a variable number of arguments, C<printf> style.
2584The argument list should end with C<...>, like this:
2585
2586 Afprd |void |croak |const char* pat|...
2587
a7486cbb 2588=item M
a422fd2d 2589
00aadd71 2590This function is part of the experimental development API, and may change
a422fd2d
SC
2591or disappear without notice.
2592
2593=item o
2594
2595This function should not have a compatibility macro to define, say,
10e2eb10 2596C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
a422fd2d 2597
a422fd2d
SC
2598=item x
2599
2600This function isn't exported out of the Perl core.
2601
dc9b1d22
MHM
2602=item m
2603
2604This is implemented as a macro.
2605
2606=item X
2607
2608This function is explicitly exported.
2609
2610=item E
2611
2612This function is visible to extensions included in the Perl core.
2613
2614=item b
2615
2616Binary backward compatibility; this function is a macro but also has
2617a C<Perl_> implementation (which is exported).
2618
1aa6ea50
JC
2619=item others
2620
2621See the comments at the top of C<embed.fnc> for others.
2622
a422fd2d
SC
2623=back
2624
dc9b1d22
MHM
2625If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2626C<make regen_headers> to force a rebuild of F<embed.h> and other
2627auto-generated files.
a422fd2d 2628
6b4667fc 2629=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2630
6b4667fc
A
2631If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2632formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2633following macros for portability
9dd9db0b 2634
c52f9dcd
JH
2635 IVdf IV in decimal
2636 UVuf UV in decimal
2637 UVof UV in octal
2638 UVxf UV in hexadecimal
2639 NVef NV %e-like
2640 NVff NV %f-like
2641 NVgf NV %g-like
9dd9db0b 2642
6b4667fc
A
2643These will take care of 64-bit integers and long doubles.
2644For example:
2645
c52f9dcd 2646 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2647
2648The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2649
aacf4ea2
JH
2650Note that there are different "long doubles": Perl will use
2651whatever the compiler has.
2652
8908e76d
JH
2653If you are printing addresses of pointers, use UVxf combined
2654with PTR2UV(), do not use %lx or %p.
2655
2656=head2 Pointer-To-Integer and Integer-To-Pointer
2657
2658Because pointer size does not necessarily equal integer size,
2659use the follow macros to do it right.
2660
c52f9dcd
JH
2661 PTR2UV(pointer)
2662 PTR2IV(pointer)
2663 PTR2NV(pointer)
2664 INT2PTR(pointertotype, integer)
8908e76d
JH
2665
2666For example:
2667
c52f9dcd
JH
2668 IV iv = ...;
2669 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2670
2671and
2672
c52f9dcd
JH
2673 AV *av = ...;
2674 UV uv = PTR2UV(av);
8908e76d 2675
0ca3a874
MHM
2676=head2 Exception Handling
2677
9b5c3821 2678There are a couple of macros to do very basic exception handling in XS
10e2eb10 2679modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
9b5c3821
MHM
2680be able to use these macros:
2681
2682 #define NO_XSLOCKS
2683 #include "XSUB.h"
2684
2685You can use these macros if you call code that may croak, but you need
10e2eb10 2686to do some cleanup before giving control back to Perl. For example:
0ca3a874 2687
d7f8936a 2688 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2689
2690 XCPT_TRY_START {
2691 code_that_may_croak();
2692 } XCPT_TRY_END
2693
2694 XCPT_CATCH
2695 {
2696 /* do cleanup here */
2697 XCPT_RETHROW;
2698 }
2699
2700Note that you always have to rethrow an exception that has been
10e2eb10
FC
2701caught. Using these macros, it is not possible to just catch the
2702exception and ignore it. If you have to ignore the exception, you
0ca3a874
MHM
2703have to use the C<call_*> function.
2704
2705The advantage of using the above macros is that you don't have
2706to setup an extra function for C<call_*>, and that using these
2707macros is faster than using C<call_*>.
2708
a422fd2d
SC
2709=head2 Source Documentation
2710
2711There's an effort going on to document the internal functions and
2712automatically produce reference manuals from them - L<perlapi> is one
2713such manual which details all the functions which are available to XS
10e2eb10 2714writers. L<perlintern> is the autogenerated manual for the functions
a422fd2d
SC
2715which are not part of the API and are supposedly for internal use only.
2716
2717Source documentation is created by putting POD comments into the C
2718source, like this:
2719
2720 /*
2721 =for apidoc sv_setiv
2722
2723 Copies an integer into the given SV. Does not handle 'set' magic. See
2724 C<sv_setiv_mg>.
2725
2726 =cut
2727 */
2728
2729Please try and supply some documentation if you add functions to the
2730Perl core.
2731
0d098d33
MHM
2732=head2 Backwards compatibility
2733
10e2eb10
FC
2734The Perl API changes over time. New functions are
2735added or the interfaces of existing functions are
2736changed. The C<Devel::PPPort> module tries to
0d098d33
MHM
2737provide compatibility code for some of these changes, so XS writers don't
2738have to code it themselves when supporting multiple versions of Perl.
2739
2740C<Devel::PPPort> generates a C header file F<ppport.h> that can also
10e2eb10 2741be run as a Perl script. To generate F<ppport.h>, run:
0d098d33
MHM
2742
2743 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2744
2745Besides checking existing XS code, the script can also be used to retrieve
2746compatibility information for various API calls using the C<--api-info>
10e2eb10 2747command line switch. For example:
0d098d33
MHM
2748
2749 % perl ppport.h --api-info=sv_magicext
2750
2751For details, see C<perldoc ppport.h>.
2752
a422fd2d
SC
2753=head1 Unicode Support
2754
10e2eb10 2755Perl 5.6.0 introduced Unicode support. It's important for porters and XS
a422fd2d
SC
2756writers to understand this support and make sure that the code they
2757write does not corrupt Unicode data.
2758
2759=head2 What B<is> Unicode, anyway?
2760
10e2eb10
FC
2761In the olden, less enlightened times, we all used to use ASCII. Most of
2762us did, anyway. The big problem with ASCII is that it's American. Well,
a422fd2d 2763no, that's not actually the problem; the problem is that it's not
10e2eb10 2764particularly useful for people who don't use the Roman alphabet. What
a422fd2d 2765used to happen was that particular languages would stick their own
10e2eb10 2766alphabet in the upper range of the sequence, between 128 and 255. Of
a422fd2d
SC
2767course, we then ended up with plenty of variants that weren't quite
2768ASCII, and the whole point of it being a standard was lost.
2769
2770Worse still, if you've got a language like Chinese or
2771Japanese that has hundreds or thousands of characters, then you really
2772can't fit them into a mere 256, so they had to forget about ASCII
2773altogether, and build their own systems using pairs of numbers to refer
2774to one character.
2775
2776To fix this, some people formed Unicode, Inc. and
2777produced a new character set containing all the characters you can
10e2eb10
FC
2778possibly think of and more. There are several ways of representing these
2779characters, and the one Perl uses is called UTF-8. UTF-8 uses
2780a variable number of bytes to represent a character. You can learn more
2575c402 2781about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2782
1e54db1a 2783=head2 How can I recognise a UTF-8 string?
a422fd2d 2784
10e2eb10
FC
2785You can't. This is because UTF-8 data is stored in bytes just like
2786non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d 2787capital E with a grave accent, is represented by the two bytes
10e2eb10
FC
2788C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2789has that byte sequence as well. So you can't tell just by looking - this
a422fd2d
SC
2790is what makes Unicode input an interesting problem.
2791
2575c402
JW
2792In general, you either have to know what you're dealing with, or you
2793have to guess. The API function C<is_utf8_string> can help; it'll tell
10e2eb10
FC
2794you if a string contains only valid UTF-8 characters. However, it can't
2795do the work for you. On a character-by-character basis,
6302f837 2796C<isUTF8_CHAR>
2575c402 2797will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2798
1e54db1a 2799=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2800
1e54db1a 2801As mentioned above, UTF-8 uses a variable number of bytes to store a
10e2eb10
FC
2802character. Characters with values 0...127 are stored in one
2803byte, just like good ol' ASCII. Character 128 is stored as
2804C<v194.128>; this continues up to character 191, which is
2805C<v194.191>. Now we've run out of bits (191 is binary
2806C<10111111>) so we move on; 192 is C<v195.128>. And
a422fd2d
SC
2807so it goes on, moving to three bytes at character 2048.
2808
1e54db1a 2809Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2810how long the first character in it is with the C<UTF8SKIP> macro:
2811
2812 char *utf = "\305\233\340\240\201";
2813 I32 len;
2814
2815 len = UTF8SKIP(utf); /* len is 2 here */
2816 utf += len;
2817 len = UTF8SKIP(utf); /* len is 3 here */
2818
1e54db1a 2819Another way to skip over characters in a UTF-8 string is to use
a422fd2d 2820C<utf8_hop>, which takes a string and a number of characters to skip
10e2eb10 2821over. You're on your own about bounds checking, though, so don't use it
a422fd2d
SC
2822lightly.
2823
1e54db1a 2824All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2825so you can test if you need to do something special with this
2826character like this (the UTF8_IS_INVARIANT() is a macro that tests
9f98c7fe 2827whether the byte is encoded as a single byte even in UTF-8):
a422fd2d 2828
3a2263fe 2829 U8 *utf;
4b88fb76 2830 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
3a2263fe 2831 UV uv; /* Note: a UV, not a U8, not a char */
95701e00 2832 STRLEN len; /* length of character in bytes */
a422fd2d 2833
3a2263fe 2834 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2835 /* Must treat this as UTF-8 */
4b88fb76 2836 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
a422fd2d
SC
2837 else
2838 /* OK to treat this character as a byte */
2839 uv = *utf;
2840
4b88fb76 2841You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
95701e00 2842value of the character; the inverse function C<uvchr_to_utf8> is available
1e54db1a 2843for putting a UV into UTF-8:
a422fd2d 2844
3a2263fe 2845 if (!UTF8_IS_INVARIANT(uv))
a422fd2d 2846 /* Must treat this as UTF8 */
95701e00 2847 utf8 = uvchr_to_utf8(utf8, uv);
a422fd2d
SC
2848 else
2849 /* OK to treat this character as a byte */
2850 *utf8++ = uv;
2851
2852You B<must> convert characters to UVs using the above functions if
1e54db1a 2853you're ever in a situation where you have to match UTF-8 and non-UTF-8
10e2eb10 2854characters. You may not skip over UTF-8 characters in this case. If you
1e54db1a
JH
2855do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2856for instance, if your UTF-8 string contains C<v196.172>, and you skip
2857that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2858So don't do that!
2859
1e54db1a 2860=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2861
2862Currently, Perl deals with Unicode strings and non-Unicode strings
10e2eb10
FC
2863slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2864string is internally encoded as UTF-8. Without it, the byte value is the
2575c402 2865codepoint number and vice versa (in other words, the string is encoded
e1b711da 2866as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
c31cc9fc
FC
2867semantics). This flag is only meaningful if the SV is C<SvPOK>
2868or immediately after stringification via C<SvPV> or a similar
2869macro. You can check and manipulate this flag with the
2575c402 2870following macros:
a422fd2d
SC
2871
2872 SvUTF8(sv)
2873 SvUTF8_on(sv)
2874 SvUTF8_off(sv)
2875
2876This flag has an important effect on Perl's treatment of the string: if
2877Unicode data is not properly distinguished, regular expressions,
2878C<length>, C<substr> and other string handling operations will have
2879undesirable results.
2880
2881The problem comes when you have, for instance, a string that isn't
2575c402 2882flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
1e54db1a 2883especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2884
2885Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2886need be sure you don't accidentally knock it off while you're
10e2eb10 2887manipulating SVs. More specifically, you cannot expect to do this:
a422fd2d
SC
2888
2889 SV *sv;
2890 SV *nsv;
2891 STRLEN len;
2892 char *p;
2893
2894 p = SvPV(sv, len);
2895 frobnicate(p);
2896 nsv = newSVpvn(p, len);
2897
2898The C<char*> string does not tell you the whole story, and you can't
10e2eb10 2899copy or reconstruct an SV just by copying the string value. Check if the
c31cc9fc
FC
2900old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
2901accordingly:
a422fd2d
SC
2902
2903 p = SvPV(sv, len);
2904 frobnicate(p);
2905 nsv = newSVpvn(p, len);
2906 if (SvUTF8(sv))
2907 SvUTF8_on(nsv);
2908
2909In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2910not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2911appropriately.
2912
3a2263fe 2913Since just passing an SV to an XS function and copying the data of
2575c402 2914the SV is not enough to copy the UTF8 flags, even less right is just
3a2263fe
RGS
2915passing a C<char *> to an XS function.
2916
1e54db1a 2917=head2 How do I convert a string to UTF-8?
a422fd2d 2918
2575c402 2919If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
10e2eb10 2920one of the strings to UTF-8. If you've got an SV, the easiest way to do
2575c402 2921this is:
a422fd2d
SC
2922
2923 sv_utf8_upgrade(sv);
2924
2925However, you must not do this, for example:
2926
2927 if (!SvUTF8(left))
2928 sv_utf8_upgrade(left);
2929
2930If you do this in a binary operator, you will actually change one of the
b1866b2d 2931strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2932by the end user, it can cause problems in deficient code.
a422fd2d 2933
1e54db1a 2934Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
10e2eb10
FC
2935string argument. This is useful for having the data available for
2936comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2937C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2938the string contains any characters above 255 that can't be represented
2939in a single byte.
2940
2941=head2 Is there anything else I need to know?
2942
10e2eb10 2943Not really. Just remember these things:
a422fd2d
SC
2944
2945=over 3
2946
2947=item *
2948
10e2eb10 2949There's no way to tell if a string is UTF-8 or not. You can tell if an SV
c31cc9fc
FC
2950is UTF-8 by looking at its C<SvUTF8> flag after stringifying it
2951with C<SvPV> or a similar macro. Don't forget to set the flag if
10e2eb10 2952something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2953it's not - if you pass on the PV to somewhere, pass on the flag too.
2954
2955=item *
2956
4b88fb76 2957If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3a2263fe 2958unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2959
2960=item *
2961
1e54db1a 2962When writing a character C<uv> to a UTF-8 string, B<always> use
95701e00 2963C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
3a2263fe 2964you can use C<*s = uv>.
a422fd2d
SC
2965
2966=item *
2967
10e2eb10
FC
2968Mixing UTF-8 and non-UTF-8 strings is
2969tricky. Use C<bytes_to_utf8> to get
2bbc8d55 2970a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
2971
2972=back
2973
53e06cf0
SC
2974=head1 Custom Operators
2975
2a0fd0f1 2976Custom operator support is an experimental feature that allows you to
10e2eb10 2977define your own ops. This is primarily to allow the building of
53e06cf0
SC
2978interpreters for other languages in the Perl core, but it also allows
2979optimizations through the creation of "macro-ops" (ops which perform the
2980functions of multiple ops which are usually executed together, such as
1aa6ea50 2981C<gvsv, gvsv, add>.)
53e06cf0 2982
10e2eb10 2983This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0 2984core does not "know" anything special about this op type, and so it will
10e2eb10 2985not be involved in any optimizations. This also means that you can
53e06cf0
SC
2986define your custom ops to be any op structure - unary, binary, list and
2987so on - you like.
2988
10e2eb10
FC
2989It's important to know what custom operators won't do for you. They
2990won't let you add new syntax to Perl, directly. They won't even let you
2991add new keywords, directly. In fact, they won't change the way Perl
2992compiles a program at all. You have to do those changes yourself, after
2993Perl has compiled the program. You do this either by manipulating the op
53e06cf0
SC
2994tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2995a custom peephole optimizer with the C<optimize> module.
2996
2997When you do this, you replace ordinary Perl ops with custom ops by
407f86e1 2998creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
10e2eb10
FC
2999PP function. This should be defined in XS code, and should look like
3000the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
53e06cf0
SC
3001takes the appropriate number of values from the stack, and you are
3002responsible for adding stack marks if necessary.
3003
3004You should also "register" your op with the Perl interpreter so that it
10e2eb10 3005can produce sensible error and warning messages. Since it is possible to
53e06cf0 3006have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
9733086d 3007Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
10e2eb10 3008it is dealing with. You should create an C<XOP> structure for each
9733086d
BM
3009ppaddr you use, set the properties of the custom op with
3010C<XopENTRY_set>, and register the structure against the ppaddr using
10e2eb10 3011C<Perl_custom_op_register>. A trivial example might look like:
9733086d
BM
3012
3013 static XOP my_xop;
3014 static OP *my_pp(pTHX);
3015
3016 BOOT:
3017 XopENTRY_set(&my_xop, xop_name, "myxop");
3018 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
3019 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
3020
3021The available fields in the structure are:
3022
3023=over 4
3024
3025=item xop_name
3026
10e2eb10 3027A short name for your op. This will be included in some error messages,
9733086d
BM
3028and will also be returned as C<< $op->name >> by the L<B|B> module, so
3029it will appear in the output of module like L<B::Concise|B::Concise>.
3030
3031=item xop_desc
3032
3033A short description of the function of the op.
3034
3035=item xop_class
3036
10e2eb10 3037Which of the various C<*OP> structures this op uses. This should be one of
9733086d
BM
3038the C<OA_*> constants from F<op.h>, namely
3039
3040=over 4
3041
3042=item OA_BASEOP
3043
3044=item OA_UNOP
3045
3046=item OA_BINOP
3047
3048=item OA_LOGOP
3049
3050=item OA_LISTOP
3051
3052=item OA_PMOP
3053
3054=item OA_SVOP
3055
3056=item OA_PADOP
3057
3058=item OA_PVOP_OR_SVOP
3059
10e2eb10 3060This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
9733086d
BM
3061the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
3062
3063=item OA_LOOP
3064
3065=item OA_COP
3066
3067=back
3068
3069The other C<OA_*> constants should not be used.
3070
3071=item xop_peep
3072
3073This member is of type C<Perl_cpeep_t>, which expands to C<void
10e2eb10 3074(*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
9733086d 3075will be called from C<Perl_rpeep> when ops of this type are encountered
10e2eb10 3076by the peephole optimizer. I<o> is the OP that needs optimizing;
9733086d
BM
3077I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
3078
3079=back
53e06cf0 3080
e7d4c058 3081C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 3082
954c1994 3083=head1 AUTHORS
e89caa19 3084
954c1994 3085Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
3086E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
3087itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 3088
954c1994
GS
3089With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
3090Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
3091Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
3092Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 3093
954c1994 3094=head1 SEE ALSO
cb1a09d0 3095
ba555bf5 3096L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>