This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Consistent spaces after dots in perlintern.pod
[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
46example, a trailing NUL is tacked on automatically. The non-string use
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
66the 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
a9b0660e
KW
90string terminating with a NUL character, and not otherwise containing
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
a3cb178b
GS
108All SVs that contain strings should be terminated with a NUL character.
109If it is not NUL-terminated there is a risk of
5f05dabc
PP
110core dumps and corruptions from code which passes the string to C
111functions or system calls which expect a NUL-terminated string.
112Perl's own functions typically add a trailing NUL for this reason.
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
134might not be terminated by a 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
a9b0660e 159add space for the trailing NUL byte (perl's own string functions typically do
8ebc5c01 160C<SvGROW(sv, len + 1)>).
a0d0e21e
LW
161
162If you have an SV and want to know what kind of data Perl thinks is stored
163in it, you can use the following macros to check the type of SV you have.
164
165 SvIOK(SV*)
166 SvNOK(SV*)
167 SvPOK(SV*)
168
169You can get and set the current length of the string stored in an SV with
170the following macros:
171
172 SvCUR(SV*)
173 SvCUR_set(SV*, I32 val)
174
cb1a09d0
AD
175You can also get a pointer to the end of the string stored in the SV
176with the macro:
177
178 SvEND(SV*)
179
180But note that these last three macros are valid only if C<SvPOK()> is true.
a0d0e21e 181
d1b91892
AD
182If you want to append something to the end of string stored in an C<SV*>,
183you can use the following functions:
184
08105a92 185 void sv_catpv(SV*, const char*);
e65f3abd 186 void sv_catpvn(SV*, const char*, STRLEN);
46fc3d4c 187 void sv_catpvf(SV*, const char*, ...);
a9b0660e
KW
188 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
189 I32, bool);
d1b91892
AD
190 void sv_catsv(SV*, SV*);
191
192The first function calculates the length of the string to be appended by
193using C<strlen>. In the second, you specify the length of the string
46fc3d4c 194yourself. The third function processes its arguments like C<sprintf> and
9abd00ed
GS
195appends the formatted output. The fourth function works like C<vsprintf>.
196You can specify the address and length of an array of SVs instead of the
10e2eb10
FC
197va_list argument. The fifth function
198extends the string stored in the first
9abd00ed
GS
199SV with the string stored in the second SV. It also forces the second SV
200to be interpreted as a string.
201
202The C<sv_cat*()> functions are not generic enough to operate on values that
203have "magic". See L<Magic Virtual Tables> later in this document.
d1b91892 204
a0d0e21e
LW
205If you know the name of a scalar variable, you can get a pointer to its SV
206by using the following:
207
64ace3f8 208 SV* get_sv("package::varname", 0);
a0d0e21e
LW
209
210This returns NULL if the variable does not exist.
211
d1b91892 212If you want to know if this variable (or any other SV) is actually C<defined>,
a0d0e21e
LW
213you can call:
214
215 SvOK(SV*)
216
06f6df17 217The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
9adebda4 218
10e2eb10
FC
219Its address can be used whenever an C<SV*> is needed. Make sure that
220you don't try to compare a random sv with C<&PL_sv_undef>. For example
9adebda4
SB
221when interfacing Perl code, it'll work correctly for:
222
223 foo(undef);
224
225But won't work when called as:
226
227 $x = undef;
228 foo($x);
229
230So to repeat always use SvOK() to check whether an sv is defined.
231
232Also you have to be careful when using C<&PL_sv_undef> as a value in
233AVs or HVs (see L<AVs, HVs and undefined values>).
a0d0e21e 234
06f6df17
RGS
235There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
236boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
237addresses can be used whenever an C<SV*> is needed.
a0d0e21e 238
9cde0e7f 239Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
a0d0e21e
LW
240Take this code:
241
242 SV* sv = (SV*) 0;
243 if (I-am-to-return-a-real-value) {
244 sv = sv_2mortal(newSViv(42));
245 }
246 sv_setsv(ST(0), sv);
247
248This code tries to return a new SV (which contains the value 42) if it should
04343c6d 249return a real value, or undef otherwise. Instead it has returned a NULL
a0d0e21e 250pointer which, somewhere down the line, will cause a segmentation violation,
06f6df17
RGS
251bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
252first line and all will be well.
a0d0e21e
LW
253
254To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
3fe9a6f1 255call is not necessary (see L<Reference Counts and Mortality>).
a0d0e21e 256
94dde4fb
SC
257=head2 Offsets
258
259Perl provides the function C<sv_chop> to efficiently remove characters
260from the beginning of a string; you give it an SV and a pointer to
da75cd15 261somewhere inside the PV, and it discards everything before the
10e2eb10 262pointer. The efficiency comes by means of a little hack: instead of
94dde4fb
SC
263actually removing the characters, C<sv_chop> sets the flag C<OOK>
264(offset OK) to signal to other functions that the offset hack is in
883bb8c0
KW
265effect, and it moves the PV pointer (called C<SvPVX>) forward
266by the number of bytes chopped off, and adjusts C<SvCUR> and C<SvLEN>
267accordingly. (A portion of the space between the old and new PV
268pointers is used to store the count of chopped bytes.)
94dde4fb
SC
269
270Hence, at this point, the start of the buffer that we allocated lives
271at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
272into the middle of this allocated storage.
273
274This is best demonstrated by example:
275
276 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
277 SV = PVIV(0x8128450) at 0x81340f0
278 REFCNT = 1
279 FLAGS = (POK,OOK,pPOK)
280 IV = 1 (OFFSET)
281 PV = 0x8135781 ( "1" . ) "2345"\0
282 CUR = 4
283 LEN = 5
284
285Here the number of bytes chopped off (1) is put into IV, and
10e2eb10 286C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
94dde4fb
SC
287portion of the string between the "real" and the "fake" beginnings is
288shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
289the fake beginning, not the real one.
290
fe854a6f 291Something similar to the offset hack is performed on AVs to enable
319cef53
SC
292efficient shifting and splicing off the beginning of the array; while
293C<AvARRAY> points to the first element in the array that is visible from
10e2eb10 294Perl, C<AvALLOC> points to the real start of the C array. These are
319cef53 295usually the same, but a C<shift> operation can be carried out by
6de131f0 296increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
319cef53 297Again, the location of the real start of the C array only comes into
10e2eb10 298play when freeing the array. See C<av_shift> in F<av.c>.
319cef53 299
d1b91892 300=head2 What's Really Stored in an SV?
a0d0e21e
LW
301
302Recall that the usual method of determining the type of scalar you have is
5f05dabc 303to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 304usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
305macros will do the appropriate conversion of string to integer/double or
306integer/double to string.
307
308If you I<really> need to know if you have an integer, double, or string
309pointer in an SV, you can use the following three macros instead:
310
311 SvIOKp(SV*)
312 SvNOKp(SV*)
313 SvPOKp(SV*)
314
315These will tell you if you truly have an integer, double, or string pointer
d1b91892 316stored in your SV. The "p" stands for private.
a0d0e21e 317
da8c5729 318There are various ways in which the private and public flags may differ.
9090718a
FC
319For example, in perl 5.16 and earlier a tied SV may have a valid
320underlying value in the IV slot (so SvIOKp is true), but the data
321should be accessed via the FETCH routine rather than directly,
322so SvIOK is false. (In perl 5.18 onwards, tied scalars use
323the flags the same way as untied scalars.) Another is when
d7f8936a 324numeric conversion has occurred and precision has been lost: only the
10e2eb10 325private flag is set on 'lossy' values. So when an NV is converted to an
9e9796d6
JH
326IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
327
07fa94a1 328In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 329
54310121 330=head2 Working with AVs
a0d0e21e 331
07fa94a1
JO
332There are two ways to create and load an AV. The first method creates an
333empty AV:
a0d0e21e
LW
334
335 AV* newAV();
336
54310121 337The second method both creates the AV and initially populates it with SVs:
a0d0e21e 338
c70927a6 339 AV* av_make(SSize_t num, SV **ptr);
a0d0e21e 340
5f05dabc 341The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 342AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 343
da8c5729 344Once the AV has been created, the following operations are possible on it:
a0d0e21e
LW
345
346 void av_push(AV*, SV*);
347 SV* av_pop(AV*);
348 SV* av_shift(AV*);
c70927a6 349 void av_unshift(AV*, SSize_t num);
a0d0e21e
LW
350
351These should be familiar operations, with the exception of C<av_unshift>.
352This routine adds C<num> elements at the front of the array with the C<undef>
353value. You must then use C<av_store> (described below) to assign values
354to these new elements.
355
356Here are some other functions:
357
c70927a6
FC
358 SSize_t av_top_index(AV*);
359 SV** av_fetch(AV*, SSize_t key, I32 lval);
360 SV** av_store(AV*, SSize_t key, SV* val);
a0d0e21e 361
dab460cd 362The C<av_top_index> function returns the highest index value in an array (just
5f05dabc
PP
363like $#array in Perl). If the array is empty, -1 is returned. The
364C<av_fetch> function returns the value at index C<key>, but if C<lval>
365is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
366The C<av_store> function stores the value C<val> at index C<key>, and does
367not increment the reference count of C<val>. Thus the caller is responsible
368for taking care of that, and if C<av_store> returns NULL, the caller will
369have to decrement the reference count to avoid a memory leak. Note that
370C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
371return value.
d1b91892 372
da8c5729
MH
373A few more:
374
a0d0e21e 375 void av_clear(AV*);
a0d0e21e 376 void av_undef(AV*);
c70927a6 377 void av_extend(AV*, SSize_t key);
5f05dabc
PP
378
379The C<av_clear> function deletes all the elements in the AV* array, but
380does not actually delete the array itself. The C<av_undef> function will
381delete all the elements in the array plus the array itself. The
adc882cf
GS
382C<av_extend> function extends the array so that it contains at least C<key+1>
383elements. If C<key+1> is less than the currently allocated length of the array,
384then nothing is done.
a0d0e21e
LW
385
386If you know the name of an array variable, you can get a pointer to its AV
387by using the following:
388
cbfd0a87 389 AV* get_av("package::varname", 0);
a0d0e21e
LW
390
391This returns NULL if the variable does not exist.
392
04343c6d
GS
393See L<Understanding the Magic of Tied Hashes and Arrays> for more
394information on how to use the array access functions on tied arrays.
395
54310121 396=head2 Working with HVs
a0d0e21e
LW
397
398To create an HV, you use the following routine:
399
400 HV* newHV();
401
da8c5729 402Once the HV has been created, the following operations are possible on it:
a0d0e21e 403
08105a92
GS
404 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
405 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 406
5f05dabc
PP
407The C<klen> parameter is the length of the key being passed in (Note that
408you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
409length of the key). The C<val> argument contains the SV pointer to the
54310121 410scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
411you want C<hv_store> to calculate it for you). The C<lval> parameter
412indicates whether this fetch is actually a part of a store operation, in
413which case a new undefined value will be added to the HV with the supplied
414key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 415
5f05dabc
PP
416Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
417C<SV*>. To access the scalar value, you must first dereference the return
418value. However, you should check to make sure that the return value is
419not NULL before dereferencing it.
a0d0e21e 420
da8c5729
MH
421The first of these two functions checks if a hash table entry exists, and the
422second deletes it.
a0d0e21e 423
08105a92
GS
424 bool hv_exists(HV*, const char* key, U32 klen);
425 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 426
5f05dabc
PP
427If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
428create and return a mortal copy of the deleted value.
429
a0d0e21e
LW
430And more miscellaneous functions:
431
432 void hv_clear(HV*);
a0d0e21e 433 void hv_undef(HV*);
5f05dabc
PP
434
435Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
436table but does not actually delete the hash table. The C<hv_undef> deletes
437both the entries and the hash table itself.
a0d0e21e 438
a9b0660e 439Perl keeps the actual data in a linked list of structures with a typedef of HE.
d1b91892
AD
440These contain the actual key and value pointers (plus extra administrative
441overhead). The key is a string pointer; the value is an C<SV*>. However,
442once you have an C<HE*>, to get the actual key and value, use the routines
443specified below.
444
a0d0e21e
LW
445 I32 hv_iterinit(HV*);
446 /* Prepares starting point to traverse hash table */
447 HE* hv_iternext(HV*);
448 /* Get the next entry, and return a pointer to a
449 structure that has both the key and value */
450 char* hv_iterkey(HE* entry, I32* retlen);
451 /* Get the key from an HE structure and also return
452 the length of the key string */
cb1a09d0 453 SV* hv_iterval(HV*, HE* entry);
d1be9408 454 /* Return an SV pointer to the value of the HE
a0d0e21e 455 structure */
cb1a09d0 456 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
457 /* This convenience routine combines hv_iternext,
458 hv_iterkey, and hv_iterval. The key and retlen
459 arguments are return values for the key and its
460 length. The value is returned in the SV* argument */
a0d0e21e
LW
461
462If you know the name of a hash variable, you can get a pointer to its HV
463by using the following:
464
6673a63c 465 HV* get_hv("package::varname", 0);
a0d0e21e
LW
466
467This returns NULL if the variable does not exist.
468
a43e7901 469The hash algorithm is defined in the C<PERL_HASH> macro:
a0d0e21e 470
a43e7901 471 PERL_HASH(hash, key, klen)
ab192400 472
a43e7901
YO
473The exact implementation of this macro varies by architecture and version
474of perl, and the return value may change per invocation, so the value
475is only valid for the duration of a single perl process.
a0d0e21e 476
04343c6d
GS
477See L<Understanding the Magic of Tied Hashes and Arrays> for more
478information on how to use the hash access functions on tied hashes.
479
1e422769
PP
480=head2 Hash API Extensions
481
482Beginning with version 5.004, the following functions are also supported:
483
484 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
485 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 486
1e422769
PP
487 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
488 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 489
1e422769
PP
490 SV* hv_iterkeysv (HE* entry);
491
492Note that these functions take C<SV*> keys, which simplifies writing
493of extension code that deals with hash structures. These functions
494also allow passing of C<SV*> keys to C<tie> functions without forcing
495you to stringify the keys (unlike the previous set of functions).
496
497They also return and accept whole hash entries (C<HE*>), making their
498use more efficient (since the hash number for a particular string
4a4eefd0
GS
499doesn't have to be recomputed every time). See L<perlapi> for detailed
500descriptions.
1e422769
PP
501
502The following macros must always be used to access the contents of hash
503entries. Note that the arguments to these macros must be simple
504variables, since they may get evaluated more than once. See
4a4eefd0 505L<perlapi> for detailed descriptions of these macros.
1e422769
PP
506
507 HePV(HE* he, STRLEN len)
508 HeVAL(HE* he)
509 HeHASH(HE* he)
510 HeSVKEY(HE* he)
511 HeSVKEY_force(HE* he)
512 HeSVKEY_set(HE* he, SV* sv)
513
514These two lower level macros are defined, but must only be used when
515dealing with keys that are not C<SV*>s:
516
517 HeKEY(HE* he)
518 HeKLEN(HE* he)
519
04343c6d
GS
520Note that both C<hv_store> and C<hv_store_ent> do not increment the
521reference count of the stored C<val>, which is the caller's responsibility.
522If these functions return a NULL value, the caller will usually have to
523decrement the reference count of C<val> to avoid a memory leak.
1e422769 524
a9381218
MHM
525=head2 AVs, HVs and undefined values
526
10e2eb10
FC
527Sometimes you have to store undefined values in AVs or HVs. Although
528this may be a rare case, it can be tricky. That's because you're
a9381218
MHM
529used to using C<&PL_sv_undef> if you need an undefined SV.
530
531For example, intuition tells you that this XS code:
532
533 AV *av = newAV();
534 av_store( av, 0, &PL_sv_undef );
535
536is equivalent to this Perl code:
537
538 my @av;
539 $av[0] = undef;
540
f3c4ec28 541Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
a9381218
MHM
542for indicating that an array element has not yet been initialized.
543Thus, C<exists $av[0]> would be true for the above Perl code, but
f3c4ec28
FC
544false for the array generated by the XS code. In perl 5.20, storing
545&PL_sv_undef will create a read-only element, because the scalar
546&PL_sv_undef itself is stored, not a copy.
a9381218 547
f3c4ec28 548Similar problems can occur when storing C<&PL_sv_undef> in HVs:
a9381218
MHM
549
550 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
551
552This will indeed make the value C<undef>, but if you try to modify
553the value of C<key>, you'll get the following error:
554
555 Modification of non-creatable hash value attempted
556
557In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
10e2eb10 558in restricted hashes. This caused such hash entries not to appear
a9381218
MHM
559when iterating over the hash or when checking for the keys
560with the C<hv_exists> function.
561
8abccac8 562You can run into similar problems when you store C<&PL_sv_yes> or
10e2eb10 563C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
564will give you the following error:
565
566 Modification of a read-only value attempted
567
568To make a long story short, you can use the special variables
8abccac8 569C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
570HVs, but you have to make sure you know what you're doing.
571
572Generally, if you want to store an undefined value in an AV
573or HV, you should not use C<&PL_sv_undef>, but rather create a
574new undefined value using the C<newSV> function, for example:
575
576 av_store( av, 42, newSV(0) );
577 hv_store( hv, "foo", 3, newSV(0), 0 );
578
a0d0e21e
LW
579=head2 References
580
d1b91892 581References are a special type of scalar that point to other data types
a9b0660e 582(including other references).
a0d0e21e 583
07fa94a1 584To create a reference, use either of the following functions:
a0d0e21e 585
5f05dabc
PP
586 SV* newRV_inc((SV*) thing);
587 SV* newRV_noinc((SV*) thing);
a0d0e21e 588
5f05dabc 589The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
590functions are identical except that C<newRV_inc> increments the reference
591count of the C<thing>, while C<newRV_noinc> does not. For historical
592reasons, C<newRV> is a synonym for C<newRV_inc>.
593
594Once you have a reference, you can use the following macro to dereference
595the reference:
a0d0e21e
LW
596
597 SvRV(SV*)
598
599then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 600C<AV*> or C<HV*>, if required.
a0d0e21e 601
d1b91892 602To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
603
604 SvROK(SV*)
605
07fa94a1
JO
606To discover what type of value the reference refers to, use the following
607macro and then check the return value.
d1b91892
AD
608
609 SvTYPE(SvRV(SV*))
610
611The most useful types that will be returned are:
612
a5e62da0
FC
613 < SVt_PVAV Scalar
614 SVt_PVAV Array
615 SVt_PVHV Hash
616 SVt_PVCV Code
617 SVt_PVGV Glob (possibly a file handle)
618
619See L<perlapi/svtype> for more details.
d1b91892 620
cb1a09d0
AD
621=head2 Blessed References and Class Objects
622
06f6df17 623References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
624OO lexicon, an object is simply a reference that has been blessed into a
625package (or class). Once blessed, the programmer may now use the reference
626to access the various methods in the class.
627
628A reference can be blessed into a package with the following function:
629
630 SV* sv_bless(SV* sv, HV* stash);
631
06f6df17
RGS
632The C<sv> argument must be a reference value. The C<stash> argument
633specifies which class the reference will belong to. See
2ae324a7 634L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
635
636/* Still under construction */
637
ddd2cc91
DM
638The following function upgrades rv to reference if not already one.
639Creates a new SV for rv to point to. If C<classname> is non-null, the SV
640is blessed into the specified class. SV is returned.
cb1a09d0 641
08105a92 642 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 643
ddd2cc91
DM
644The following three functions copy integer, unsigned integer or double
645into an SV whose reference is C<rv>. SV is blessed if C<classname> is
646non-null.
cb1a09d0 647
08105a92 648 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 649 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 650 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 651
ddd2cc91
DM
652The following function copies the pointer value (I<the address, not the
653string!>) into an SV whose reference is rv. SV is blessed if C<classname>
654is non-null.
cb1a09d0 655
ddd2cc91 656 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
cb1a09d0 657
a9b0660e 658The following function copies a string into an SV whose reference is C<rv>.
ddd2cc91
DM
659Set length to 0 to let Perl calculate the string length. SV is blessed if
660C<classname> is non-null.
cb1a09d0 661
a9b0660e
KW
662 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
663 STRLEN length);
cb1a09d0 664
ddd2cc91
DM
665The following function tests whether the SV is blessed into the specified
666class. It does not check inheritance relationships.
9abd00ed 667
08105a92 668 int sv_isa(SV* sv, const char* name);
9abd00ed 669
ddd2cc91 670The following function tests whether the SV is a reference to a blessed object.
9abd00ed
GS
671
672 int sv_isobject(SV* sv);
673
ddd2cc91 674The following function tests whether the SV is derived from the specified
10e2eb10
FC
675class. SV can be either a reference to a blessed object or a string
676containing a class name. This is the function implementing the
ddd2cc91 677C<UNIVERSAL::isa> functionality.
9abd00ed 678
08105a92 679 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 680
00aadd71 681To check if you've got an object derived from a specific class you have
9abd00ed
GS
682to write:
683
684 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 685
5f05dabc 686=head2 Creating New Variables
cb1a09d0 687
5f05dabc
PP
688To create a new Perl variable with an undef value which can be accessed from
689your Perl script, use the following routines, depending on the variable type.
cb1a09d0 690
64ace3f8 691 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 692 AV* get_av("package::varname", GV_ADD);
6673a63c 693 HV* get_hv("package::varname", GV_ADD);
cb1a09d0 694
058a5f6c 695Notice the use of GV_ADD as the second parameter. The new variable can now
cb1a09d0
AD
696be set, using the routines appropriate to the data type.
697
5f05dabc 698There are additional macros whose values may be bitwise OR'ed with the
058a5f6c 699C<GV_ADD> argument to enable certain extra features. Those bits are:
cb1a09d0 700
9a68f1db
SB
701=over
702
703=item GV_ADDMULTI
704
705Marks the variable as multiply defined, thus preventing the:
706
707 Name <varname> used only once: possible typo
708
709warning.
710
9a68f1db
SB
711=item GV_ADDWARN
712
713Issues the warning:
714
715 Had to create <varname> unexpectedly
716
717if the variable did not exist before the function was called.
718
719=back
cb1a09d0 720
07fa94a1
JO
721If you do not specify a package name, the variable is created in the current
722package.
cb1a09d0 723
5f05dabc 724=head2 Reference Counts and Mortality
a0d0e21e 725
10e2eb10 726Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 727AVs, or HVs (xV for short in the following) start their life with a
55497cff 728reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 729then it will be destroyed and its memory made available for reuse.
55497cff
PP
730
731This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
732undef'ed or the last variable holding a reference to it is changed or
733overwritten. At the internal level, however, reference counts can be
55497cff
PP
734manipulated with the following macros:
735
736 int SvREFCNT(SV* sv);
5f05dabc 737 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
738 void SvREFCNT_dec(SV* sv);
739
740However, there is one other function which manipulates the reference
07fa94a1
JO
741count of its argument. The C<newRV_inc> function, you will recall,
742creates a reference to the specified argument. As a side effect,
743it increments the argument's reference count. If this is not what
744you want, use C<newRV_noinc> instead.
745
746For example, imagine you want to return a reference from an XSUB function.
747Inside the XSUB routine, you create an SV which initially has a reference
748count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
749This returns the reference as a new SV, but the reference count of the
750SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
751return the reference from the XSUB routine and forget about the SV.
752But Perl hasn't! Whenever the returned reference is destroyed, the
753reference count of the original SV is decreased to one and nothing happens.
754The SV will hang around without any way to access it until Perl itself
755terminates. This is a memory leak.
5f05dabc
PP
756
757The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
758C<newRV_inc>. Then, if and when the last reference is destroyed,
759the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 760stopping any memory leak.
55497cff 761
5f05dabc 762There are some convenience functions available that can help with the
54310121 763destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
764An xV that is mortal has had its reference count marked to be decremented,
765but not actually decremented, until "a short time later". Generally the
766term "short time later" means a single Perl statement, such as a call to
54310121 767an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
768reference count decremented depends on two macros, SAVETMPS and FREETMPS.
769See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
770
771"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
772However, if you mortalize a variable twice, the reference count will
773later be decremented twice.
774
00aadd71
NIS
775"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
776For example an SV which is created just to pass a number to a called sub
06f6df17 777is made mortal to have it cleaned up automatically when it's popped off
10e2eb10 778the stack. Similarly, results returned by XSUBs (which are pushed on the
06f6df17 779stack) are often made mortal.
a0d0e21e
LW
780
781To create a mortal variable, use the functions:
782
783 SV* sv_newmortal()
784 SV* sv_2mortal(SV*)
785 SV* sv_mortalcopy(SV*)
786
00aadd71 787The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
788SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
789third creates a mortal copy of an existing SV.
da8c5729 790Because C<sv_newmortal> gives the new SV no value, it must normally be given one
9a68f1db 791via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
792
793 SV *tmp = sv_newmortal();
794 sv_setiv(tmp, an_integer);
795
796As that is multiple C statements it is quite common so see this idiom instead:
797
798 SV *tmp = sv_2mortal(newSViv(an_integer));
799
800
801You should be careful about creating mortal variables. Strange things
802can happen if you make the same value mortal within multiple contexts,
10e2eb10
FC
803or if you make a variable mortal multiple
804times. Thinking of "Mortalization"
00aadd71 805as deferred C<SvREFCNT_dec> should help to minimize such problems.
da8c5729 806For example if you are passing an SV which you I<know> has a high enough REFCNT
00aadd71
NIS
807to survive its use on the stack you need not do any mortalization.
808If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
809making a C<sv_mortalcopy> is safer.
a0d0e21e 810
ac036724 811The mortal routines are not just for SVs; AVs and HVs can be
faed5253 812made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 813C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 814
5f05dabc 815=head2 Stashes and Globs
a0d0e21e 816
06f6df17
RGS
817A B<stash> is a hash that contains all variables that are defined
818within a package. Each key of the stash is a symbol
aa689395
PP
819name (shared by all the different types of objects that have the same
820name), and each value in the hash table is a GV (Glob Value). This GV
821in turn contains references to the various objects of that name,
822including (but not limited to) the following:
cb1a09d0 823
a0d0e21e
LW
824 Scalar Value
825 Array Value
826 Hash Value
a3cb178b 827 I/O Handle
a0d0e21e
LW
828 Format
829 Subroutine
830
06f6df17
RGS
831There is a single stash called C<PL_defstash> that holds the items that exist
832in the C<main> package. To get at the items in other packages, append the
833string "::" to the package name. The items in the C<Foo> package are in
834the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
835in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 836
d1b91892 837To get the stash pointer for a particular package, use the function:
a0d0e21e 838
da51bb9b
NC
839 HV* gv_stashpv(const char* name, I32 flags)
840 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
841
842The first function takes a literal string, the second uses the string stored
d1b91892 843in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 844C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
845
846The name that C<gv_stash*v> wants is the name of the package whose symbol table
847you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
848packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
849language itself.
a0d0e21e
LW
850
851Alternately, if you have an SV that is a blessed reference, you can find
852out the stash pointer by using:
853
854 HV* SvSTASH(SvRV(SV*));
855
856then use the following to get the package name itself:
857
858 char* HvNAME(HV* stash);
859
5f05dabc
PP
860If you need to bless or re-bless an object you can use the following
861function:
a0d0e21e
LW
862
863 SV* sv_bless(SV*, HV* stash)
864
865where the first argument, an C<SV*>, must be a reference, and the second
866argument is a stash. The returned C<SV*> can now be used in the same way
867as any other SV.
868
d1b91892
AD
869For more information on references and blessings, consult L<perlref>.
870
54310121 871=head2 Double-Typed SVs
0a753a76
PP
872
873Scalar variables normally contain only one type of value, an integer,
874double, pointer, or reference. Perl will automatically convert the
875actual scalar data from the stored type into the requested type.
876
877Some scalar variables contain more than one type of scalar data. For
878example, the variable C<$!> contains either the numeric value of C<errno>
879or its string equivalent from either C<strerror> or C<sys_errlist[]>.
880
881To force multiple data values into an SV, you must do two things: use the
882C<sv_set*v> routines to add the additional scalar type, then set a flag
883so that Perl will believe it contains more than one type of data. The
884four macros to set the flags are:
885
886 SvIOK_on
887 SvNOK_on
888 SvPOK_on
889 SvROK_on
890
891The particular macro you must use depends on which C<sv_set*v> routine
892you called first. This is because every C<sv_set*v> routine turns on
893only the bit for the particular type of data being set, and turns off
894all the rest.
895
896For example, to create a new Perl variable called "dberror" that contains
897both the numeric and descriptive string error values, you could use the
898following code:
899
900 extern int dberror;
901 extern char *dberror_list;
902
64ace3f8 903 SV* sv = get_sv("dberror", GV_ADD);
0a753a76
PP
904 sv_setiv(sv, (IV) dberror);
905 sv_setpv(sv, dberror_list[dberror]);
906 SvIOK_on(sv);
907
908If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
909macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
910
911=head2 Magic Variables
a0d0e21e 912
d1b91892
AD
913[This section still under construction. Ignore everything here. Post no
914bills. Everything not permitted is forbidden.]
915
d1b91892
AD
916Any SV may be magical, that is, it has special features that a normal
917SV does not have. These features are stored in the SV structure in a
5f05dabc 918linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
919
920 struct magic {
921 MAGIC* mg_moremagic;
922 MGVTBL* mg_virtual;
923 U16 mg_private;
924 char mg_type;
925 U8 mg_flags;
b205eb13 926 I32 mg_len;
d1b91892
AD
927 SV* mg_obj;
928 char* mg_ptr;
d1b91892
AD
929 };
930
931Note this is current as of patchlevel 0, and could change at any time.
932
933=head2 Assigning Magic
934
935Perl adds magic to an SV using the sv_magic function:
936
a9b0660e 937 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
938
939The C<sv> argument is a pointer to the SV that is to acquire a new magical
940feature.
941
942If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
10e2eb10
FC
943convert C<sv> to type C<SVt_PVMG>.
944Perl then continues by adding new magic
645c22ef
DM
945to the beginning of the linked list of magical features. Any prior entry
946of the same type of magic is deleted. Note that this can be overridden,
947and multiple instances of the same type of magic can be associated with an
948SV.
d1b91892 949
54310121 950The C<name> and C<namlen> arguments are used to associate a string with
10e2eb10 951the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
952C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
953C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
954whether C<namlen> is greater than zero or equal to zero respectively. As a
955special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
956to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
957
958The sv_magic function uses C<how> to determine which, if any, predefined
959"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 960See the L<Magic Virtual Tables> section below. The C<how> argument is also
10e2eb10
FC
961stored in the C<mg_type> field. The value of
962C<how> should be chosen from the set of macros
963C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 964these macros were added, Perl internals used to directly use character
14befaf4 965literals, so you may occasionally come across old code or documentation
75d0f26d 966referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
967
968The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
969structure. If it is not the same as the C<sv> argument, the reference
970count of the C<obj> object is incremented. If it is the same, or if
645c22ef 971the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 972then C<obj> is merely stored, without the reference count being incremented.
d1b91892 973
2d8d5d5a
SH
974See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
975to an SV.
976
cb1a09d0
AD
977There is also a function to add magic to an C<HV>:
978
979 void hv_magic(HV *hv, GV *gv, int how);
980
981This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
982
983To remove the magic from an SV, call the function sv_unmagic:
984
70a53b35 985 int sv_unmagic(SV *sv, int type);
cb1a09d0
AD
986
987The C<type> argument should be equal to the C<how> value when the C<SV>
988was initially made magical.
989
f6ee7b17 990However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
10e2eb10
FC
991C<SV>. If you want to remove only certain
992magic of a C<type> based on the magic
f6ee7b17
FR
993virtual table, use C<sv_unmagicext> instead:
994
995 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
996
d1b91892
AD
997=head2 Magic Virtual Tables
998
d1be9408 999The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
1000C<MGVTBL>, which is a structure of function pointers and stands for
1001"Magic Virtual Table" to handle the various operations that might be
1002applied to that variable.
1003
301cb7e8
DM
1004The C<MGVTBL> has five (or sometimes eight) pointers to the following
1005routine types:
d1b91892
AD
1006
1007 int (*svt_get)(SV* sv, MAGIC* mg);
1008 int (*svt_set)(SV* sv, MAGIC* mg);
1009 U32 (*svt_len)(SV* sv, MAGIC* mg);
1010 int (*svt_clear)(SV* sv, MAGIC* mg);
1011 int (*svt_free)(SV* sv, MAGIC* mg);
1012
a9b0660e
KW
1013 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1014 const char *name, I32 namlen);
301cb7e8
DM
1015 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1016 int (*svt_local)(SV *nsv, MAGIC *mg);
1017
1018
06f6df17 1019This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
SM
1020currently 32 types. These different structures contain pointers to various
1021routines that perform additional actions depending on which function is
1022being called.
d1b91892 1023
a9b0660e
KW
1024 Function pointer Action taken
1025 ---------------- ------------
1026 svt_get Do something before the value of the SV is
1027 retrieved.
1028 svt_set Do something after the SV is assigned a value.
1029 svt_len Report on the SV's length.
1030 svt_clear Clear something the SV represents.
1031 svt_free Free any extra storage associated with the SV.
d1b91892 1032
a9b0660e
KW
1033 svt_copy copy tied variable magic to a tied element
1034 svt_dup duplicate a magic structure during thread cloning
1035 svt_local copy magic to local value during 'local'
301cb7e8 1036
d1b91892 1037For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1038to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1039
1040 { magic_get, magic_set, magic_len, 0, 0 }
1041
14befaf4
DM
1042Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1043if a get operation is being performed, the routine C<magic_get> is
1044called. All the various routines for the various magical types begin
1045with C<magic_>. NOTE: the magic routines are not considered part of
1046the Perl API, and may not be exported by the Perl library.
d1b91892 1047
301cb7e8
DM
1048The last three slots are a recent addition, and for source code
1049compatibility they are only checked for if one of the three flags
10e2eb10
FC
1050MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1051This means that most code can continue declaring
1052a vtable as a 5-element value. These three are
301cb7e8
DM
1053currently used exclusively by the threading code, and are highly subject
1054to change.
1055
d1b91892
AD
1056The current kinds of Magic Virtual Tables are:
1057
f1f5ddd7
FC
1058=for comment
1059This table is generated by regen/mg_vtable.pl. Any changes made here
1060will be lost.
1061
1062=for mg_vtable.pl begin
1063
a9b0660e 1064 mg_type
bd6e6c12
FC
1065 (old-style char and macro) MGVTBL Type of magic
1066 -------------------------- ------ -------------
1067 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1068 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1069 % PERL_MAGIC_rhash (none) extra data for restricted
1070 hashes
2a388207 1071 & PERL_MAGIC_proto (none) my sub prototype CV
bd6e6c12
FC
1072 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1073 : PERL_MAGIC_symtab (none) extra data for symbol
1074 tables
1075 < PERL_MAGIC_backref vtbl_backref for weak ref data
1076 @ PERL_MAGIC_arylen_p (none) to move arylen out of XPVAV
1077 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1078 (fast string search)
1079 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1080 (AMT) on stash
1081 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1082 (@+ and @- vars)
1083 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1084 element
1085 E PERL_MAGIC_env vtbl_env %ENV hash
1086 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
eccba044 1087 f PERL_MAGIC_fm vtbl_regexp Formline
bd6e6c12 1088 ('compiled' format)
bd6e6c12
FC
1089 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1090 H PERL_MAGIC_hints vtbl_hints %^H hash
1091 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1092 I PERL_MAGIC_isa vtbl_isa @ISA array
1093 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1094 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1095 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1096 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1097 element
1098 N PERL_MAGIC_shared (none) Shared between threads
1099 n PERL_MAGIC_shared_scalar (none) Shared between threads
1100 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1101 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1102 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1103 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1104 r PERL_MAGIC_qr vtbl_regexp precompiled qr// regex
1105 S PERL_MAGIC_sig (none) %SIG hash
1106 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1107 t PERL_MAGIC_taint vtbl_taint Taintedness
1108 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1109 extensions
1110 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1111 extensions
4499db73 1112 V PERL_MAGIC_vstring (none) SV was vstring literal
bd6e6c12
FC
1113 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1114 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1115 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1116 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1117 variable / smart parameter
1118 vivification
1119 ] PERL_MAGIC_checkcall vtbl_checkcall inlining/mutation of call
1120 to this CV
1121 ~ PERL_MAGIC_ext (none) Available for use by
1122 extensions
0cbee0a4 1123
f1f5ddd7 1124=for mg_vtable.pl end
d1b91892 1125
68dc0745 1126When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1127uppercase letter is typically used to represent some kind of composite type
1128(a list or a hash), and the lowercase letter is used to represent an element
10e2eb10 1129of that composite type. Some internals code makes use of this case
92f0c265 1130relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1131
1132The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1133specifically for use by extensions and will not be used by perl itself.
1134Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1135to variables (typically objects). This is especially useful because
1136there is no way for normal perl code to corrupt this private information
1137(unlike using extra elements of a hash object).
1138
1139Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1140C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1141C<mg_ptr> field points to a C<ufuncs> structure:
1142
1143 struct ufuncs {
a9402793
AB
1144 I32 (*uf_val)(pTHX_ IV, SV*);
1145 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1146 IV uf_index;
1147 };
1148
1149When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1150function will be called with C<uf_index> as the first arg and a pointer to
1151the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1152magic is shown below. Note that the ufuncs structure is copied by
1153sv_magic, so you can safely allocate it on the stack.
1154
1155 void
1156 Umagic(sv)
1157 SV *sv;
1158 PREINIT:
1159 struct ufuncs uf;
1160 CODE:
1161 uf.uf_val = &my_get_fn;
1162 uf.uf_set = &my_set_fn;
1163 uf.uf_index = 0;
14befaf4 1164 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1165
1e73acc8
AS
1166Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1167
1168For hashes there is a specialized hook that gives control over hash
1169keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1170if the "set" function in the C<ufuncs> structure is NULL. The hook
1171is activated whenever the hash is accessed with a key specified as
1172an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1173C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1174through the functions without the C<..._ent> suffix circumvents the
4509d391 1175hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1e73acc8 1176
14befaf4
DM
1177Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1178or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1179extra care to avoid conflict. Typically only using the magic on
1180objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1181For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1182C<MGVTBL>, even if all its fields will be C<0>, so that individual
1183C<MAGIC> pointers can be identified as a particular kind of magic
10e2eb10 1184using their magic virtual table. C<mg_findext> provides an easy way
f6ee7b17 1185to do that:
2f07f21a
FR
1186
1187 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1188
1189 MAGIC *mg;
f6ee7b17
FR
1190 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1191 /* this is really ours, not another module's PERL_MAGIC_ext */
1192 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1193 ...
2f07f21a 1194 }
5f05dabc 1195
ef50df4b
GS
1196Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1197earlier do B<not> invoke 'set' magic on their targets. This must
1198be done by the user either by calling the C<SvSETMAGIC()> macro after
1199calling these functions, or by using one of the C<sv_set*_mg()> or
1200C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1201C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1202obtained from external sources in functions that don't handle magic.
4a4eefd0 1203See L<perlapi> for a description of these functions.
189b2af5
GS
1204For example, calls to the C<sv_cat*()> functions typically need to be
1205followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1206since their implementation handles 'get' magic.
1207
d1b91892
AD
1208=head2 Finding Magic
1209
a9b0660e
KW
1210 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1211 * type */
f6ee7b17
FR
1212
1213This routine returns a pointer to a C<MAGIC> structure stored in the SV.
10e2eb10
FC
1214If the SV does not have that magical
1215feature, C<NULL> is returned. If the
f6ee7b17 1216SV has multiple instances of that magical feature, the first one will be
10e2eb10
FC
1217returned. C<mg_findext> can be used
1218to find a C<MAGIC> structure of an SV
da8c5729 1219based on both its magic type and its magic virtual table:
f6ee7b17
FR
1220
1221 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
d1b91892 1222
f6ee7b17
FR
1223Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1224SVt_PVMG, Perl may core dump.
d1b91892 1225
08105a92 1226 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1227
1228This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1229field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1230the mg_type field is changed to be the lowercase letter.
a0d0e21e 1231
04343c6d
GS
1232=head2 Understanding the Magic of Tied Hashes and Arrays
1233
14befaf4
DM
1234Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1235magic type.
9edb2b46
GS
1236
1237WARNING: As of the 5.004 release, proper usage of the array and hash
1238access functions requires understanding a few caveats. Some
1239of these caveats are actually considered bugs in the API, to be fixed
10e2eb10 1240in later releases, and are bracketed with [MAYCHANGE] below. If
9edb2b46
GS
1241you find yourself actually applying such information in this section, be
1242aware that the behavior may change in the future, umm, without warning.
04343c6d 1243
1526ead6 1244The perl tie function associates a variable with an object that implements
9a68f1db 1245the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1246tie function from an XSUB, you must mimic this behaviour. The code below
1247carries out the necessary steps - firstly it creates a new hash, and then
1248creates a second hash which it blesses into the class which will implement
10e2eb10 1249the tie methods. Lastly it ties the two hashes together, and returns a
1526ead6
AB
1250reference to the new tied hash. Note that the code below does NOT call the
1251TIEHASH method in the MyTie class -
1252see L<Calling Perl Routines from within C Programs> for details on how
1253to do this.
1254
1255 SV*
1256 mytie()
1257 PREINIT:
1258 HV *hash;
1259 HV *stash;
1260 SV *tie;
1261 CODE:
1262 hash = newHV();
1263 tie = newRV_noinc((SV*)newHV());
da51bb9b 1264 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1265 sv_bless(tie, stash);
899e16d0 1266 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1267 RETVAL = newRV_noinc(hash);
1268 OUTPUT:
1269 RETVAL
1270
04343c6d
GS
1271The C<av_store> function, when given a tied array argument, merely
1272copies the magic of the array onto the value to be "stored", using
1273C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1274actually need to be stored in the array. [MAYCHANGE] After a call to
1275C<av_store> on a tied array, the caller will usually need to call
1276C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1277TIEARRAY object. If C<av_store> did return NULL, a call to
1278C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1279leak. [/MAYCHANGE]
04343c6d
GS
1280
1281The previous paragraph is applicable verbatim to tied hash access using the
1282C<hv_store> and C<hv_store_ent> functions as well.
1283
1284C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1285C<hv_fetch_ent> actually return an undefined mortal value whose magic
1286has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1287need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1288need to call C<mg_get()> on the returned value in order to actually invoke
1289the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1290you may also call C<mg_set()> on the return value after possibly assigning
1291a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1292method on the TIE object. [/MAYCHANGE]
04343c6d 1293
9edb2b46 1294[MAYCHANGE]
04343c6d
GS
1295In other words, the array or hash fetch/store functions don't really
1296fetch and store actual values in the case of tied arrays and hashes. They
1297merely call C<mg_copy> to attach magic to the values that were meant to be
1298"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1299do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1300the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1301and hashes.
1302
1303Currently (as of perl version 5.004), use of the hash and array access
1304functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1305"normal" hashes and arrays, or on their tied variants. The API may be
1306changed to provide more transparent access to both tied and normal data
1307types in future versions.
1308[/MAYCHANGE]
04343c6d
GS
1309
1310You would do well to understand that the TIEARRAY and TIEHASH interfaces
1311are mere sugar to invoke some perl method calls while using the uniform hash
1312and array syntax. The use of this sugar imposes some overhead (typically
1313about two to four extra opcodes per FETCH/STORE operation, in addition to
1314the creation of all the mortal variables required to invoke the methods).
1315This overhead will be comparatively small if the TIE methods are themselves
1316substantial, but if they are only a few statements long, the overhead
1317will not be insignificant.
1318
d1c897a1
IZ
1319=head2 Localizing changes
1320
1321Perl has a very handy construction
1322
1323 {
1324 local $var = 2;
1325 ...
1326 }
1327
1328This construction is I<approximately> equivalent to
1329
1330 {
1331 my $oldvar = $var;
1332 $var = 2;
1333 ...
1334 $var = $oldvar;
1335 }
1336
1337The biggest difference is that the first construction would
1338reinstate the initial value of $var, irrespective of how control exits
10e2eb10 1339the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1340more efficient as well.
1341
1342There is a way to achieve a similar task from C via Perl API: create a
1343I<pseudo-block>, and arrange for some changes to be automatically
1344undone at the end of it, either explicit, or via a non-local exit (via
10e2eb10 1345die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1346C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1347Such a construct may be created specially for some important localized
1348task, or an existing one (like boundaries of enclosing Perl
1349subroutine/block, or an existing pair for freeing TMPs) may be
10e2eb10
FC
1350used. (In the second case the overhead of additional localization must
1351be almost negligible.) Note that any XSUB is automatically enclosed in
b687b08b 1352an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1353
1354Inside such a I<pseudo-block> the following service is available:
1355
13a2d996 1356=over 4
d1c897a1
IZ
1357
1358=item C<SAVEINT(int i)>
1359
1360=item C<SAVEIV(IV i)>
1361
1362=item C<SAVEI32(I32 i)>
1363
1364=item C<SAVELONG(long i)>
1365
1366These macros arrange things to restore the value of integer variable
1367C<i> at the end of enclosing I<pseudo-block>.
1368
1369=item C<SAVESPTR(s)>
1370
1371=item C<SAVEPPTR(p)>
1372
1373These macros arrange things to restore the value of pointers C<s> and
10e2eb10 1374C<p>. C<s> must be a pointer of a type which survives conversion to
d1c897a1
IZ
1375C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1376and back.
1377
1378=item C<SAVEFREESV(SV *sv)>
1379
1380The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1381I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1382mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1383extends the lifetime of C<sv> until the beginning of the next statement,
1384C<SAVEFREESV> extends it until the end of the enclosing scope. These
1385lifetimes can be wildly different.
1386
1387Also compare C<SAVEMORTALIZESV>.
1388
1389=item C<SAVEMORTALIZESV(SV *sv)>
1390
1391Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1392scope instead of decrementing its reference count. This usually has the
1393effect of keeping C<sv> alive until the statement that called the currently
1394live scope has finished executing.
d1c897a1
IZ
1395
1396=item C<SAVEFREEOP(OP *op)>
1397
1398The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1399
1400=item C<SAVEFREEPV(p)>
1401
1402The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1403end of I<pseudo-block>.
1404
1405=item C<SAVECLEARSV(SV *sv)>
1406
1407Clears a slot in the current scratchpad which corresponds to C<sv> at
1408the end of I<pseudo-block>.
1409
1410=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1411
10e2eb10 1412The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
d1c897a1
IZ
1413string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1414short-lived storage, the corresponding string may be reallocated like
1415this:
1416
9cde0e7f 1417 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1418
c76ac1ee 1419=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1420
1421At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1422only argument C<p>.
1423
1424=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1425
1426At the end of I<pseudo-block> the function C<f> is called with the
1427implicit context argument (if any), and C<p>.
d1c897a1
IZ
1428
1429=item C<SAVESTACK_POS()>
1430
1431The current offset on the Perl internal stack (cf. C<SP>) is restored
1432at the end of I<pseudo-block>.
1433
1434=back
1435
1436The following API list contains functions, thus one needs to
1437provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1438or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1439function takes C<int *>.
1440
13a2d996 1441=over 4
d1c897a1
IZ
1442
1443=item C<SV* save_scalar(GV *gv)>
1444
1445Equivalent to Perl code C<local $gv>.
1446
1447=item C<AV* save_ary(GV *gv)>
1448
1449=item C<HV* save_hash(GV *gv)>
1450
1451Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1452
1453=item C<void save_item(SV *item)>
1454
1455Duplicates the current value of C<SV>, on the exit from the current
1456C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
10e2eb10 1457using the stored value. It doesn't handle magic. Use C<save_scalar> if
038fcae3 1458magic is affected.
d1c897a1
IZ
1459
1460=item C<void save_list(SV **sarg, I32 maxsarg)>
1461
1462A variant of C<save_item> which takes multiple arguments via an array
1463C<sarg> of C<SV*> of length C<maxsarg>.
1464
1465=item C<SV* save_svref(SV **sptr)>
1466
d1be9408 1467Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1468
1469=item C<void save_aptr(AV **aptr)>
1470
1471=item C<void save_hptr(HV **hptr)>
1472
1473Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1474
1475=back
1476
1477The C<Alias> module implements localization of the basic types within the
1478I<caller's scope>. People who are interested in how to localize things in
1479the containing scope should take a look there too.
1480
0a753a76 1481=head1 Subroutines
a0d0e21e 1482
68dc0745 1483=head2 XSUBs and the Argument Stack
5f05dabc
PP
1484
1485The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1486An XSUB routine will have a stack that contains the arguments from the Perl
1487program, and a way to map from the Perl data structures to a C equivalent.
1488
1489The stack arguments are accessible through the C<ST(n)> macro, which returns
1490the C<n>'th stack argument. Argument 0 is the first argument passed in the
1491Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1492an C<SV*> is used.
1493
1494Most of the time, output from the C routine can be handled through use of
1495the RETVAL and OUTPUT directives. However, there are some cases where the
1496argument stack is not already long enough to handle all the return values.
1497An example is the POSIX tzname() call, which takes no arguments, but returns
1498two, the local time zone's standard and summer time abbreviations.
1499
1500To handle this situation, the PPCODE directive is used and the stack is
1501extended using the macro:
1502
924508f0 1503 EXTEND(SP, num);
5f05dabc 1504
924508f0
GS
1505where C<SP> is the macro that represents the local copy of the stack pointer,
1506and C<num> is the number of elements the stack should be extended by.
5f05dabc 1507
00aadd71 1508Now that there is room on the stack, values can be pushed on it using C<PUSHs>
10e2eb10 1509macro. The pushed values will often need to be "mortal" (See
d82b684c 1510L</Reference Counts and Mortality>):
5f05dabc 1511
00aadd71 1512 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1513 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1514 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1515 PUSHs(sv_2mortal(newSVpv("Some String",0)))
a9b0660e
KW
1516 /* Although the last example is better written as the more
1517 * efficient: */
a3179684 1518 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
5f05dabc
PP
1519
1520And now the Perl program calling C<tzname>, the two values will be assigned
1521as in:
1522
1523 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1524
1525An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1526to use the macro:
5f05dabc 1527
5f05dabc
PP
1528 XPUSHs(SV*)
1529
da8c5729 1530This macro automatically adjusts the stack for you, if needed. Thus, you
5f05dabc 1531do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1532
1533Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1534C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1535For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1536C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc
PP
1537
1538For more information, consult L<perlxs> and L<perlxstut>.
1539
5b36e945
FC
1540=head2 Autoloading with XSUBs
1541
1542If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1543fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1544of the XSUB's package.
1545
1546But it also puts the same information in certain fields of the XSUB itself:
1547
1548 HV *stash = CvSTASH(cv);
1549 const char *subname = SvPVX(cv);
1550 STRLEN name_length = SvCUR(cv); /* in bytes */
1551 U32 is_utf8 = SvUTF8(cv);
f703fc96 1552
5b36e945 1553C<SvPVX(cv)> contains just the sub name itself, not including the package.
d8893903
FC
1554For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1555C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
5b36e945
FC
1556
1557B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1558XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1559XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1560to support 5.8-5.14, use the XSUB's fields.
1561
5f05dabc 1562=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1563
1564There are four routines that can be used to call a Perl subroutine from
1565within a C program. These four are:
1566
954c1994
GS
1567 I32 call_sv(SV*, I32);
1568 I32 call_pv(const char*, I32);
1569 I32 call_method(const char*, I32);
5aaab254 1570 I32 call_argv(const char*, I32, char**);
a0d0e21e 1571
954c1994 1572The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1573contains either the name of the Perl subroutine to be called, or a
1574reference to the subroutine. The second argument consists of flags
1575that control the context in which the subroutine is called, whether
1576or not the subroutine is being passed arguments, how errors should be
1577trapped, and how to treat return values.
a0d0e21e
LW
1578
1579All four routines return the number of arguments that the subroutine returned
1580on the Perl stack.
1581
9a68f1db 1582These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1583but those names are now deprecated; macros of the same name are provided for
1584compatibility.
1585
1586When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1587must manipulate the Perl stack. These include the following macros and
1588functions:
a0d0e21e
LW
1589
1590 dSP
924508f0 1591 SP
a0d0e21e
LW
1592 PUSHMARK()
1593 PUTBACK
1594 SPAGAIN
1595 ENTER
1596 SAVETMPS
1597 FREETMPS
1598 LEAVE
1599 XPUSH*()
cb1a09d0 1600 POP*()
a0d0e21e 1601
5f05dabc
PP
1602For a detailed description of calling conventions from C to Perl,
1603consult L<perlcall>.
a0d0e21e 1604
5f05dabc 1605=head2 Memory Allocation
a0d0e21e 1606
06f6df17
RGS
1607=head3 Allocation
1608
86058a2d
GS
1609All memory meant to be used with the Perl API functions should be manipulated
1610using the macros described in this section. The macros provide the necessary
1611transparency between differences in the actual malloc implementation that is
1612used within perl.
1613
1614It is suggested that you enable the version of malloc that is distributed
5f05dabc 1615with Perl. It keeps pools of various sizes of unallocated memory in
07fa94a1
JO
1616order to satisfy allocation requests more quickly. However, on some
1617platforms, it may cause spurious malloc or free errors.
d1b91892 1618
06f6df17
RGS
1619The following three macros are used to initially allocate memory :
1620
9f653bb5
SH
1621 Newx(pointer, number, type);
1622 Newxc(pointer, number, type, cast);
1623 Newxz(pointer, number, type);
d1b91892 1624
9f653bb5 1625The first argument C<pointer> should be the name of a variable that will
5f05dabc 1626point to the newly allocated memory.
d1b91892 1627
9f653bb5 1628The second and third arguments C<number> and C<type> specify how many of
d1b91892 1629the specified type of data structure should be allocated. The argument
9f653bb5 1630C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
d1b91892
AD
1631should be used if the C<pointer> argument is different from the C<type>
1632argument.
1633
9f653bb5 1634Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
d1b91892
AD
1635to zero out all the newly allocated memory.
1636
06f6df17
RGS
1637=head3 Reallocation
1638
d1b91892
AD
1639 Renew(pointer, number, type);
1640 Renewc(pointer, number, type, cast);
1641 Safefree(pointer)
1642
1643These three macros are used to change a memory buffer size or to free a
1644piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1645match those of C<New> and C<Newc> with the exception of not needing the
1646"magic cookie" argument.
1647
06f6df17
RGS
1648=head3 Moving
1649
d1b91892
AD
1650 Move(source, dest, number, type);
1651 Copy(source, dest, number, type);
1652 Zero(dest, number, type);
1653
1654These three macros are used to move, copy, or zero out previously allocated
1655memory. The C<source> and C<dest> arguments point to the source and
1656destination starting points. Perl will move, copy, or zero out C<number>
1657instances of the size of the C<type> data structure (using the C<sizeof>
1658function).
a0d0e21e 1659
5f05dabc 1660=head2 PerlIO
ce3d39e2 1661
da8c5729 1662The most recent development releases of Perl have been experimenting with
5f05dabc
PP
1663removing Perl's dependency on the "normal" standard I/O suite and allowing
1664other stdio implementations to be used. This involves creating a new
1665abstraction layer that then calls whichever implementation of stdio Perl
68dc0745 1666was compiled with. All XSUBs should now use the functions in the PerlIO
5f05dabc
PP
1667abstraction layer and not make any assumptions about what kind of stdio
1668is being used.
1669
1670For a complete description of the PerlIO abstraction, consult L<perlapio>.
1671
8ebc5c01 1672=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1673
1674A lot of opcodes (this is an elementary operation in the internal perl
10e2eb10
FC
1675stack machine) put an SV* on the stack. However, as an optimization
1676the corresponding SV is (usually) not recreated each time. The opcodes
ce3d39e2
IZ
1677reuse specially assigned SVs (I<target>s) which are (as a corollary)
1678not constantly freed/created.
1679
0a753a76 1680Each of the targets is created only once (but see
ce3d39e2
IZ
1681L<Scratchpads and recursion> below), and when an opcode needs to put
1682an integer, a double, or a string on stack, it just sets the
1683corresponding parts of its I<target> and puts the I<target> on stack.
1684
1685The macro to put this target on stack is C<PUSHTARG>, and it is
1686directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1687others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1688
1bd1c0d5 1689Because the target is reused, you must be careful when pushing multiple
10e2eb10 1690values on the stack. The following code will not do what you think:
1bd1c0d5
SC
1691
1692 XPUSHi(10);
1693 XPUSHi(20);
1694
1695This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1696the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1697At the end of the operation, the stack does not contain the values 10
1698and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1699to 20.
1bd1c0d5 1700
d82b684c
SH
1701If you need to push multiple different values then you should either use
1702the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1703none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1704SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1705will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1706this a little easier to achieve by creating a new mortal for you (via
1707C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1708in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1709Thus, instead of writing this to "fix" the example above:
1710
1711 XPUSHs(sv_2mortal(newSViv(10)))
1712 XPUSHs(sv_2mortal(newSViv(20)))
1713
1714you can simply write:
1715
1716 mXPUSHi(10)
1717 mXPUSHi(20)
1718
1719On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1720need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1721macros can make use of the local variable C<TARG>. See also C<dTARGET>
1722and C<dXSTARG>.
1bd1c0d5 1723
8ebc5c01 1724=head2 Scratchpads
ce3d39e2 1725
54310121 1726The question remains on when the SVs which are I<target>s for opcodes
10e2eb10 1727are created. The answer is that they are created when the current
ac036724 1728unit--a subroutine or a file (for opcodes for statements outside of
10e2eb10 1729subroutines)--is compiled. During this time a special anonymous Perl
ac036724 1730array is created, which is called a scratchpad for the current unit.
ce3d39e2 1731
54310121 1732A scratchpad keeps SVs which are lexicals for the current unit and are
d777b41a
FC
1733targets for opcodes. A previous version of this document
1734stated that one can deduce that an SV lives on a scratchpad
ce3d39e2 1735by looking on its flags: lexicals have C<SVs_PADMY> set, and
d777b41a
FC
1736I<target>s have C<SVs_PADTMP> set. But this have never been fully true.
1737C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
1738While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
1739that have never resided in a pad, but nonetheless act like I<target>s.
ce3d39e2 1740
10e2eb10 1741The correspondence between OPs and I<target>s is not 1-to-1. Different
54310121 1742OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1743would not conflict with the expected life of the temporary.
1744
2ae324a7 1745=head2 Scratchpads and recursion
ce3d39e2
IZ
1746
1747In fact it is not 100% true that a compiled unit contains a pointer to
10e2eb10
FC
1748the scratchpad AV. In fact it contains a pointer to an AV of
1749(initially) one element, and this element is the scratchpad AV. Why do
ce3d39e2
IZ
1750we need an extra level of indirection?
1751
10e2eb10 1752The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2 1753these can create several execution pointers going into the same
10e2eb10 1754subroutine. For the subroutine-child not write over the temporaries
ce3d39e2
IZ
1755for the subroutine-parent (lifespan of which covers the call to the
1756child), the parent and the child should have different
10e2eb10 1757scratchpads. (I<And> the lexicals should be separate anyway!)
ce3d39e2 1758
5f05dabc
PP
1759So each subroutine is born with an array of scratchpads (of length 1).
1760On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1761depth of the recursion is not more than the length of this array, and
1762if it is, new scratchpad is created and pushed into the array.
1763
1764The I<target>s on this scratchpad are C<undef>s, but they are already
1765marked with correct flags.
1766
0a753a76
PP
1767=head1 Compiled code
1768
1769=head2 Code tree
1770
1771Here we describe the internal form your code is converted to by
10e2eb10 1772Perl. Start with a simple example:
0a753a76
PP
1773
1774 $a = $b + $c;
1775
1776This is converted to a tree similar to this one:
1777
1778 assign-to
1779 / \
1780 + $a
1781 / \
1782 $b $c
1783
7b8d334a 1784(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1785parsed your code, but has nothing to do with the execution order.
1786There is an additional "thread" going through the nodes of the tree
1787which shows the order of execution of the nodes. In our simplified
1788example above it looks like:
1789
1790 $b ---> $c ---> + ---> $a ---> assign-to
1791
1792But with the actual compile tree for C<$a = $b + $c> it is different:
1793some nodes I<optimized away>. As a corollary, though the actual tree
1794contains more nodes than our simplified example, the execution order
1795is the same as in our example.
1796
1797=head2 Examining the tree
1798
06f6df17
RGS
1799If you have your perl compiled for debugging (usually done with
1800C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1801compiled tree by specifying C<-Dx> on the Perl command line. The
1802output takes several lines per node, and for C<$b+$c> it looks like
1803this:
1804
1805 5 TYPE = add ===> 6
1806 TARG = 1
1807 FLAGS = (SCALAR,KIDS)
1808 {
1809 TYPE = null ===> (4)
1810 (was rv2sv)
1811 FLAGS = (SCALAR,KIDS)
1812 {
1813 3 TYPE = gvsv ===> 4
1814 FLAGS = (SCALAR)
1815 GV = main::b
1816 }
1817 }
1818 {
1819 TYPE = null ===> (5)
1820 (was rv2sv)
1821 FLAGS = (SCALAR,KIDS)
1822 {
1823 4 TYPE = gvsv ===> 5
1824 FLAGS = (SCALAR)
1825 GV = main::c
1826 }
1827 }
1828
1829This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1830not optimized away (one per number in the left column). The immediate
1831children of the given node correspond to C<{}> pairs on the same level
1832of indentation, thus this listing corresponds to the tree:
1833
1834 add
1835 / \
1836 null null
1837 | |
1838 gvsv gvsv
1839
1840The execution order is indicated by C<===E<gt>> marks, thus it is C<3
18414 5 6> (node C<6> is not included into above listing), i.e.,
1842C<gvsv gvsv add whatever>.
1843
9afa14e3 1844Each of these nodes represents an op, a fundamental operation inside the
10e2eb10 1845Perl core. The code which implements each operation can be found in the
9afa14e3 1846F<pp*.c> files; the function which implements the op with type C<gvsv>
10e2eb10 1847is C<pp_gvsv>, and so on. As the tree above shows, different ops have
9afa14e3 1848different numbers of children: C<add> is a binary operator, as one would
10e2eb10 1849expect, and so has two children. To accommodate the various different
9afa14e3
SC
1850numbers of children, there are various types of op data structure, and
1851they link together in different ways.
1852
10e2eb10 1853The simplest type of op structure is C<OP>: this has no children. Unary
9afa14e3 1854operators, C<UNOP>s, have one child, and this is pointed to by the
10e2eb10
FC
1855C<op_first> field. Binary operators (C<BINOP>s) have not only an
1856C<op_first> field but also an C<op_last> field. The most complex type of
1857op is a C<LISTOP>, which has any number of children. In this case, the
9afa14e3 1858first child is pointed to by C<op_first> and the last child by
10e2eb10 1859C<op_last>. The children in between can be found by iteratively
9afa14e3
SC
1860following the C<op_sibling> pointer from the first child to the last.
1861
1862There are also two other op types: a C<PMOP> holds a regular expression,
10e2eb10
FC
1863and has no children, and a C<LOOP> may or may not have children. If the
1864C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
9afa14e3
SC
1865complicate matters, if a C<UNOP> is actually a C<null> op after
1866optimization (see L</Compile pass 2: context propagation>) it will still
1867have children in accordance with its former type.
1868
06f6df17
RGS
1869Another way to examine the tree is to use a compiler back-end module, such
1870as L<B::Concise>.
1871
0a753a76
PP
1872=head2 Compile pass 1: check routines
1873
8870b5c7 1874The tree is created by the compiler while I<yacc> code feeds it
10e2eb10 1875the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
1876the first pass of perl compilation.
1877
1878What makes this pass interesting for perl developers is that some
1879optimization may be performed on this pass. This is optimization by
8870b5c7 1880so-called "check routines". The correspondence between node names
0a753a76
PP
1881and corresponding check routines is described in F<opcode.pl> (do not
1882forget to run C<make regen_headers> if you modify this file).
1883
1884A check routine is called when the node is fully constructed except
7b8d334a 1885for the execution-order thread. Since at this time there are no
0a753a76
PP
1886back-links to the currently constructed node, one can do most any
1887operation to the top-level node, including freeing it and/or creating
1888new nodes above/below it.
1889
1890The check routine returns the node which should be inserted into the
1891tree (if the top-level node was not modified, check routine returns
1892its argument).
1893
10e2eb10 1894By convention, check routines have names C<ck_*>. They are usually
0a753a76
PP
1895called from C<new*OP> subroutines (or C<convert>) (which in turn are
1896called from F<perly.y>).
1897
1898=head2 Compile pass 1a: constant folding
1899
1900Immediately after the check routine is called the returned node is
1901checked for being compile-time executable. If it is (the value is
1902judged to be constant) it is immediately executed, and a I<constant>
1903node with the "return value" of the corresponding subtree is
1904substituted instead. The subtree is deleted.
1905
1906If constant folding was not performed, the execution-order thread is
1907created.
1908
1909=head2 Compile pass 2: context propagation
1910
1911When a context for a part of compile tree is known, it is propagated
a3cb178b 1912down through the tree. At this time the context can have 5 values
0a753a76
PP
1913(instead of 2 for runtime context): void, boolean, scalar, list, and
1914lvalue. In contrast with the pass 1 this pass is processed from top
1915to bottom: a node's context determines the context for its children.
1916
1917Additional context-dependent optimizations are performed at this time.
1918Since at this moment the compile tree contains back-references (via
1919"thread" pointers), nodes cannot be free()d now. To allow
1920optimized-away nodes at this stage, such nodes are null()ified instead
1921of free()ing (i.e. their type is changed to OP_NULL).
1922
1923=head2 Compile pass 3: peephole optimization
1924
1925After the compile tree for a subroutine (or for an C<eval> or a file)
10e2eb10 1926is created, an additional pass over the code is performed. This pass
0a753a76 1927is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
1928additional complications for conditionals). Optimizations performed
1929at this stage are subject to the same restrictions as in the pass 2.
1930
1931Peephole optimizations are done by calling the function pointed to
1932by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1933calls the function pointed to by the global variable C<PL_rpeepp>.
1934By default, that performs some basic op fixups and optimisations along
1935the execution-order op chain, and recursively calls C<PL_rpeepp> for
1936each side chain of ops (resulting from conditionals). Extensions may
1937provide additional optimisations or fixups, hooking into either the
1938per-subroutine or recursive stage, like this:
1939
1940 static peep_t prev_peepp;
1941 static void my_peep(pTHX_ OP *o)
1942 {
1943 /* custom per-subroutine optimisation goes here */
f0358462 1944 prev_peepp(aTHX_ o);
9ea12537
Z
1945 /* custom per-subroutine optimisation may also go here */
1946 }
1947 BOOT:
1948 prev_peepp = PL_peepp;
1949 PL_peepp = my_peep;
1950
1951 static peep_t prev_rpeepp;
1952 static void my_rpeep(pTHX_ OP *o)
1953 {
1954 OP *orig_o = o;
1955 for(; o; o = o->op_next) {
1956 /* custom per-op optimisation goes here */
1957 }
f0358462 1958 prev_rpeepp(aTHX_ orig_o);
9ea12537
Z
1959 }
1960 BOOT:
1961 prev_rpeepp = PL_rpeepp;
1962 PL_rpeepp = my_rpeep;
0a753a76 1963
1ba7f851
PJ
1964=head2 Pluggable runops
1965
1966The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
1967functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1968with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1969control over the execution of the compile tree it is possible to provide
1970your own runops function.
1ba7f851
PJ
1971
1972It's probably best to copy one of the existing runops functions and
1973change it to suit your needs. Then, in the BOOT section of your XS
1974file, add the line:
1975
1976 PL_runops = my_runops;
1977
1978This function should be as efficient as possible to keep your programs
1979running as fast as possible.
1980
fd85fad2
BM
1981=head2 Compile-time scope hooks
1982
1983As of perl 5.14 it is possible to hook into the compile-time lexical
10e2eb10 1984scope mechanism using C<Perl_blockhook_register>. This is used like
fd85fad2
BM
1985this:
1986
1987 STATIC void my_start_hook(pTHX_ int full);
1988 STATIC BHK my_hooks;
1989
1990 BOOT:
a88d97bf 1991 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
1992 Perl_blockhook_register(aTHX_ &my_hooks);
1993
1994This will arrange to have C<my_start_hook> called at the start of
10e2eb10 1995compiling every lexical scope. The available hooks are:
fd85fad2
BM
1996
1997=over 4
1998
a88d97bf 1999=item C<void bhk_start(pTHX_ int full)>
fd85fad2 2000
10e2eb10 2001This is called just after starting a new lexical scope. Note that Perl
fd85fad2
BM
2002code like
2003
2004 if ($x) { ... }
2005
2006creates two scopes: the first starts at the C<(> and has C<full == 1>,
10e2eb10
FC
2007the second starts at the C<{> and has C<full == 0>. Both end at the
2008C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
fd85fad2
BM
2009pushed onto the save stack by this hook will be popped just before the
2010scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2011
a88d97bf 2012=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
2013
2014This is called at the end of a lexical scope, just before unwinding the
10e2eb10 2015stack. I<o> is the root of the optree representing the scope; it is a
fd85fad2
BM
2016double pointer so you can replace the OP if you need to.
2017
a88d97bf 2018=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
2019
2020This is called at the end of a lexical scope, just after unwinding the
10e2eb10 2021stack. I<o> is as above. Note that it is possible for calls to C<pre_>
fd85fad2
BM
2022and C<post_end> to nest, if there is something on the save stack that
2023calls string eval.
2024
a88d97bf 2025=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
2026
2027This is called just before starting to compile an C<eval STRING>, C<do
10e2eb10 2028FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
fd85fad2
BM
2029OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2030C<OP_DOFILE> or C<OP_REQUIRE>.
2031
2032=back
2033
2034Once you have your hook functions, you need a C<BHK> structure to put
10e2eb10
FC
2035them in. It's best to allocate it statically, since there is no way to
2036free it once it's registered. The function pointers should be inserted
fd85fad2 2037into this structure using the C<BhkENTRY_set> macro, which will also set
10e2eb10 2038flags indicating which entries are valid. If you do need to allocate
fd85fad2
BM
2039your C<BHK> dynamically for some reason, be sure to zero it before you
2040start.
2041
2042Once registered, there is no mechanism to switch these hooks off, so if
10e2eb10 2043that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
2044is probably the best way, so the effect is lexically scoped; however it
2045is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
10e2eb10 2046temporarily switch entries on and off. You should also be aware that
a3e07c87
BM
2047generally speaking at least one scope will have opened before your
2048extension is loaded, so you will see some C<pre/post_end> pairs that
2049didn't have a matching C<start>.
fd85fad2 2050
9afa14e3
SC
2051=head1 Examining internal data structures with the C<dump> functions
2052
2053To aid debugging, the source file F<dump.c> contains a number of
2054functions which produce formatted output of internal data structures.
2055
2056The most commonly used of these functions is C<Perl_sv_dump>; it's used
10e2eb10 2057for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
9afa14e3 2058C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 2059module should already be familiar with its format.
9afa14e3
SC
2060
2061C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 2062derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
2063C<Perl_dump_eval> will dump the main root of the code being evaluated,
2064exactly like C<-Dx>.
2065
2066Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2067op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2068subroutines in a package like so: (Thankfully, these are all xsubs, so
2069there is no op tree)
2070
2071 (gdb) print Perl_dump_packsubs(PL_defstash)
2072
2073 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2074
2075 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2076
2077 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2078
2079 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2080
2081 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2082
2083and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2084the op tree of the main root.
2085
954c1994 2086=head1 How multiple interpreters and concurrency are supported
ee072b34 2087
ee072b34
GS
2088=head2 Background and PERL_IMPLICIT_CONTEXT
2089
2090The Perl interpreter can be regarded as a closed box: it has an API
2091for feeding it code or otherwise making it do things, but it also has
2092functions for its own use. This smells a lot like an object, and
2093there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2094interpreters, with one interpreter represented either as a C structure,
2095or inside a thread-specific structure. These structures contain all
2096the context, the state of that interpreter.
2097
10e2eb10 2098One macro controls the major Perl build flavor: MULTIPLICITY. The
7b52221d 2099MULTIPLICITY build has a C structure that packages all the interpreter
10e2eb10 2100state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
7b52221d 2101normally defined, and enables the support for passing in a "hidden" first
10e2eb10 2102argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2103multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2104to the macro USE_ITHREADS.)
54aff467 2105
27da23d5
JH
2106Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2107PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2108former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2109internal variables of Perl to be wrapped inside a single global struct,
2110struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2111the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2112one step further, there is still a single struct (allocated in main()
2113either from heap or from stack) but there are no global data symbols
3bf17896 2114pointing to it. In either case the global struct should be initialized
27da23d5
JH
2115as the very first thing in main() using Perl_init_global_struct() and
2116correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2117please see F<miniperlmain.c> for usage details. You may also need
2118to use C<dVAR> in your coding to "declare the global variables"
2119when you are using them. dTHX does this for you automatically.
2120
bc028b6b
JH
2121To see whether you have non-const data you can use a BSD-compatible C<nm>:
2122
2123 nm libperl.a | grep -v ' [TURtr] '
2124
2125If this displays any C<D> or C<d> symbols, you have non-const data.
2126
27da23d5
JH
2127For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2128doesn't actually hide all symbols inside a big global struct: some
2129PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2130then hides everything (see how the PERLIO_FUNCS_DECL is used).
2131
54aff467 2132All this obviously requires a way for the Perl internal functions to be
acfe0abc 2133either subroutines taking some kind of structure as the first
ee072b34 2134argument, or subroutines taking nothing as the first argument. To
acfe0abc 2135enable these two very different ways of building the interpreter,
ee072b34
GS
2136the Perl source (as it does in so many other situations) makes heavy
2137use of macros and subroutine naming conventions.
2138
54aff467 2139First problem: deciding which functions will be public API functions and
00aadd71 2140which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2141(think "S" for "secret" or "static"). All other functions begin with
2142"Perl_", but just because a function begins with "Perl_" does not mean it is
10e2eb10
FC
2143part of the API. (See L</Internal
2144Functions>.) The easiest way to be B<sure> a
00aadd71
NIS
2145function is part of the API is to find its entry in L<perlapi>.
2146If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2147think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2148L<perlbug> explaining why you think it should be.
ee072b34
GS
2149
2150Second problem: there must be a syntax so that the same subroutine
2151declarations and calls can pass a structure as their first argument,
2152or pass nothing. To solve this, the subroutines are named and
2153declared in a particular way. Here's a typical start of a static
2154function used within the Perl guts:
2155
2156 STATIC void
2157 S_incline(pTHX_ char *s)
2158
acfe0abc 2159STATIC becomes "static" in C, and may be #define'd to nothing in some
da8c5729 2160configurations in the future.
ee072b34 2161
651a3225
GS
2162A public function (i.e. part of the internal API, but not necessarily
2163sanctioned for use in extensions) begins like this:
ee072b34
GS
2164
2165 void
2307c6d0 2166 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2167
0147cd53 2168C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2169details of the interpreter's context. THX stands for "thread", "this",
2170or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2171The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2172or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2173their variants.
ee072b34 2174
a7486cbb
JH
2175When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2176first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2177in the pTHX_ macro indicates that the macro expansion needs a comma
2178after the context argument because other arguments follow it. If
2179PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2180subroutine is not prototyped to take the extra argument. The form of the
2181macro without the trailing underscore is used when there are no additional
ee072b34
GS
2182explicit arguments.
2183
54aff467 2184When a core function calls another, it must pass the context. This
2307c6d0 2185is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2186something like this:
2187
2307c6d0
SB
2188 #ifdef PERL_IMPLICIT_CONTEXT
2189 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2190 /* can't do this for vararg functions, see below */
2307c6d0
SB
2191 #else
2192 #define sv_setiv Perl_sv_setiv
2193 #endif
ee072b34
GS
2194
2195This works well, and means that XS authors can gleefully write:
2196
2307c6d0 2197 sv_setiv(foo, bar);
ee072b34
GS
2198
2199and still have it work under all the modes Perl could have been
2200compiled with.
2201
ee072b34
GS
2202This doesn't work so cleanly for varargs functions, though, as macros
2203imply that the number of arguments is known in advance. Instead we
2204either need to spell them out fully, passing C<aTHX_> as the first
2205argument (the Perl core tends to do this with functions like
2206Perl_warner), or use a context-free version.
2207
2208The context-free version of Perl_warner is called
2209Perl_warner_nocontext, and does not take the extra argument. Instead
2210it does dTHX; to get the context from thread-local storage. We
2211C<#define warner Perl_warner_nocontext> so that extensions get source
2212compatibility at the expense of performance. (Passing an arg is
2213cheaper than grabbing it from thread-local storage.)
2214
acfe0abc 2215You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2216Those are strictly for use within the core. Extensions and embedders
2217need only be aware of [pad]THX.
2218
a7486cbb
JH
2219=head2 So what happened to dTHR?
2220
2221C<dTHR> was introduced in perl 5.005 to support the older thread model.
2222The older thread model now uses the C<THX> mechanism to pass context
2223pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2224later still have it for backward source compatibility, but it is defined
2225to be a no-op.
2226
ee072b34
GS
2227=head2 How do I use all this in extensions?
2228
2229When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2230any functions in the Perl API will need to pass the initial context
2231argument somehow. The kicker is that you will need to write it in
2232such a way that the extension still compiles when Perl hasn't been
2233built with PERL_IMPLICIT_CONTEXT enabled.
2234
2235There are three ways to do this. First, the easy but inefficient way,
2236which is also the default, in order to maintain source compatibility
0147cd53 2237with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2238and aTHX_ macros to call a function that will return the context.
2239Thus, something like:
2240
2307c6d0 2241 sv_setiv(sv, num);
ee072b34 2242
4375e838 2243in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2244in effect:
ee072b34 2245
2307c6d0 2246 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2247
54aff467 2248or to this otherwise:
ee072b34 2249
2307c6d0 2250 Perl_sv_setiv(sv, num);
ee072b34 2251
da8c5729 2252You don't have to do anything new in your extension to get this; since
2fa86c13 2253the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2254work.
2255
2256The second, more efficient way is to use the following template for
2257your Foo.xs:
2258
c52f9dcd
JH
2259 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2260 #include "EXTERN.h"
2261 #include "perl.h"
2262 #include "XSUB.h"
ee072b34 2263
fd061412 2264 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2265
fd061412 2266 STATIC void
c52f9dcd
JH
2267 my_private_function(int arg1, int arg2)
2268 {
2269 dTHX; /* fetch context */
2270 ... call many Perl API functions ...
2271 }
ee072b34
GS
2272
2273 [... etc ...]
2274
c52f9dcd 2275 MODULE = Foo PACKAGE = Foo
ee072b34 2276
c52f9dcd 2277 /* typical XSUB */
ee072b34 2278
c52f9dcd
JH
2279 void
2280 my_xsub(arg)
2281 int arg
2282 CODE:
2283 my_private_function(arg, 10);
ee072b34
GS
2284
2285Note that the only two changes from the normal way of writing an
2286extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2287including the Perl headers, followed by a C<dTHX;> declaration at
2288the start of every function that will call the Perl API. (You'll
2289know which functions need this, because the C compiler will complain
2290that there's an undeclared identifier in those functions.) No changes
2291are needed for the XSUBs themselves, because the XS() macro is
2292correctly defined to pass in the implicit context if needed.
2293
2294The third, even more efficient way is to ape how it is done within
2295the Perl guts:
2296
2297
c52f9dcd
JH
2298 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2299 #include "EXTERN.h"
2300 #include "perl.h"
2301 #include "XSUB.h"
ee072b34
GS
2302
2303 /* pTHX_ only needed for functions that call Perl API */
fd061412 2304 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2305
fd061412 2306 STATIC void
c52f9dcd
JH
2307 my_private_function(pTHX_ int arg1, int arg2)
2308 {
2309 /* dTHX; not needed here, because THX is an argument */
2310 ... call Perl API functions ...
2311 }
ee072b34
GS
2312
2313 [... etc ...]
2314
c52f9dcd 2315 MODULE = Foo PACKAGE = Foo
ee072b34 2316
c52f9dcd 2317 /* typical XSUB */
ee072b34 2318
c52f9dcd
JH
2319 void
2320 my_xsub(arg)
2321 int arg
2322 CODE:
2323 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2324
2325This implementation never has to fetch the context using a function
2326call, since it is always passed as an extra argument. Depending on
2327your needs for simplicity or efficiency, you may mix the previous
2328two approaches freely.
2329
651a3225
GS
2330Never add a comma after C<pTHX> yourself--always use the form of the
2331macro with the underscore for functions that take explicit arguments,
2332or the form without the argument for functions with no explicit arguments.
ee072b34 2333
27da23d5
JH
2334If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2335definition is needed if the Perl global variables (see F<perlvars.h>
2336or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2337used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2338the need for C<dVAR> only with the said compile-time define, because
2339otherwise the Perl global variables are visible as-is.
2340
a7486cbb
JH
2341=head2 Should I do anything special if I call perl from multiple threads?
2342
2343If you create interpreters in one thread and then proceed to call them in
2344another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2345initialized correctly in each of those threads.
2346
2347The C<perl_alloc> and C<perl_clone> API functions will automatically set
2348the TLS slot to the interpreter they created, so that there is no need to do
2349anything special if the interpreter is always accessed in the same thread that
2350created it, and that thread did not create or call any other interpreters
2351afterwards. If that is not the case, you have to set the TLS slot of the
2352thread before calling any functions in the Perl API on that particular
2353interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2354thread as the first thing you do:
2355
2356 /* do this before doing anything else with some_perl */
2357 PERL_SET_CONTEXT(some_perl);
2358
2359 ... other Perl API calls on some_perl go here ...
2360
ee072b34
GS
2361=head2 Future Plans and PERL_IMPLICIT_SYS
2362
2363Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2364that the interpreter knows about itself and pass it around, so too are
2365there plans to allow the interpreter to bundle up everything it knows
2366about the environment it's running on. This is enabled with the
7b52221d
RGS
2367PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2368Windows.
ee072b34
GS
2369
2370This allows the ability to provide an extra pointer (called the "host"
2371environment) for all the system calls. This makes it possible for
2372all the system stuff to maintain their own state, broken down into
2373seven C structures. These are thin wrappers around the usual system
0147cd53 2374calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2375more ambitious host (like the one that would do fork() emulation) all
2376the extra work needed to pretend that different interpreters are
2377actually different "processes", would be done here.
2378
2379The Perl engine/interpreter and the host are orthogonal entities.
2380There could be one or more interpreters in a process, and one or
2381more "hosts", with free association between them.
2382
a422fd2d
SC
2383=head1 Internal Functions
2384
2385All of Perl's internal functions which will be exposed to the outside
06f6df17 2386world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d 2387functions or functions used in a program in which Perl is embedded.
10e2eb10 2388Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2389static functions start with C<S_>.)
a422fd2d 2390
0972ecdf
DM
2391Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2392either with or without the C<Perl_> prefix, thanks to a bunch of defines
10e2eb10 2393that live in F<embed.h>. Note that extension code should I<not> set
0972ecdf
DM
2394C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2395breakage of the XS in each new perl release.
2396
2397The file F<embed.h> is generated automatically from
10e2eb10 2398F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
dc9b1d22 2399header files for the internal functions, generates the documentation
10e2eb10 2400and a lot of other bits and pieces. It's important that when you add
dc9b1d22 2401a new function to the core or change an existing one, you change the
10e2eb10 2402data in the table in F<embed.fnc> as well. Here's a sample entry from
dc9b1d22 2403that table:
a422fd2d
SC
2404
2405 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2406
10e2eb10
FC
2407The second column is the return type, the third column the name. Columns
2408after that are the arguments. The first column is a set of flags:
a422fd2d
SC
2409
2410=over 3
2411
2412=item A
2413
10e2eb10
FC
2414This function is a part of the public
2415API. All such functions should also
1aa6ea50 2416have 'd', very few do not.
a422fd2d
SC
2417
2418=item p
2419
1aa6ea50
JC
2420This function has a C<Perl_> prefix; i.e. it is defined as
2421C<Perl_av_fetch>.
a422fd2d
SC
2422
2423=item d
2424
2425This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2426look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2427
2428=back
2429
2430Other available flags are:
2431
2432=over 3
2433
2434=item s
2435
1aa6ea50
JC
2436This is a static function and is defined as C<STATIC S_whatever>, and
2437usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2438
2439=item n
2440
da8c5729 2441This does not need an interpreter context, so the definition has no
1aa6ea50 2442C<pTHX>, and it follows that callers don't use C<aTHX>. (See
d3a43cd8 2443L</Background and PERL_IMPLICIT_CONTEXT>.)
a422fd2d
SC
2444
2445=item r
2446
2447This function never returns; C<croak>, C<exit> and friends.
2448
2449=item f
2450
2451This function takes a variable number of arguments, C<printf> style.
2452The argument list should end with C<...>, like this:
2453
2454 Afprd |void |croak |const char* pat|...
2455
a7486cbb 2456=item M
a422fd2d 2457
00aadd71 2458This function is part of the experimental development API, and may change
a422fd2d
SC
2459or disappear without notice.
2460
2461=item o
2462
2463This function should not have a compatibility macro to define, say,
10e2eb10 2464C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
a422fd2d 2465
a422fd2d
SC
2466=item x
2467
2468This function isn't exported out of the Perl core.
2469
dc9b1d22
MHM
2470=item m
2471
2472This is implemented as a macro.
2473
2474=item X
2475
2476This function is explicitly exported.
2477
2478=item E
2479
2480This function is visible to extensions included in the Perl core.
2481
2482=item b
2483
2484Binary backward compatibility; this function is a macro but also has
2485a C<Perl_> implementation (which is exported).
2486
1aa6ea50
JC
2487=item others
2488
2489See the comments at the top of C<embed.fnc> for others.
2490
a422fd2d
SC
2491=back
2492
dc9b1d22
MHM
2493If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2494C<make regen_headers> to force a rebuild of F<embed.h> and other
2495auto-generated files.
a422fd2d 2496
6b4667fc 2497=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2498
6b4667fc
A
2499If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2500formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2501following macros for portability
9dd9db0b 2502
c52f9dcd
JH
2503 IVdf IV in decimal
2504 UVuf UV in decimal
2505 UVof UV in octal
2506 UVxf UV in hexadecimal
2507 NVef NV %e-like
2508 NVff NV %f-like
2509 NVgf NV %g-like
9dd9db0b 2510
6b4667fc
A
2511These will take care of 64-bit integers and long doubles.
2512For example:
2513
c52f9dcd 2514 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2515
2516The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2517
8908e76d
JH
2518If you are printing addresses of pointers, use UVxf combined
2519with PTR2UV(), do not use %lx or %p.
2520
2521=head2 Pointer-To-Integer and Integer-To-Pointer
2522
2523Because pointer size does not necessarily equal integer size,
2524use the follow macros to do it right.
2525
c52f9dcd
JH
2526 PTR2UV(pointer)
2527 PTR2IV(pointer)
2528 PTR2NV(pointer)
2529 INT2PTR(pointertotype, integer)
8908e76d
JH
2530
2531For example:
2532
c52f9dcd
JH
2533 IV iv = ...;
2534 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2535
2536and
2537
c52f9dcd
JH
2538 AV *av = ...;
2539 UV uv = PTR2UV(av);
8908e76d 2540
0ca3a874
MHM
2541=head2 Exception Handling
2542
9b5c3821 2543There are a couple of macros to do very basic exception handling in XS
10e2eb10 2544modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
9b5c3821
MHM
2545be able to use these macros:
2546
2547 #define NO_XSLOCKS
2548 #include "XSUB.h"
2549
2550You can use these macros if you call code that may croak, but you need
10e2eb10 2551to do some cleanup before giving control back to Perl. For example:
0ca3a874 2552
d7f8936a 2553 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2554
2555 XCPT_TRY_START {
2556 code_that_may_croak();
2557 } XCPT_TRY_END
2558
2559 XCPT_CATCH
2560 {
2561 /* do cleanup here */
2562 XCPT_RETHROW;
2563 }
2564
2565Note that you always have to rethrow an exception that has been
10e2eb10
FC
2566caught. Using these macros, it is not possible to just catch the
2567exception and ignore it. If you have to ignore the exception, you
0ca3a874
MHM
2568have to use the C<call_*> function.
2569
2570The advantage of using the above macros is that you don't have
2571to setup an extra function for C<call_*>, and that using these
2572macros is faster than using C<call_*>.
2573
a422fd2d
SC
2574=head2 Source Documentation
2575
2576There's an effort going on to document the internal functions and
2577automatically produce reference manuals from them - L<perlapi> is one
2578such manual which details all the functions which are available to XS
10e2eb10 2579writers. L<perlintern> is the autogenerated manual for the functions
a422fd2d
SC
2580which are not part of the API and are supposedly for internal use only.
2581
2582Source documentation is created by putting POD comments into the C
2583source, like this:
2584
2585 /*
2586 =for apidoc sv_setiv
2587
2588 Copies an integer into the given SV. Does not handle 'set' magic. See
2589 C<sv_setiv_mg>.
2590
2591 =cut
2592 */
2593
2594Please try and supply some documentation if you add functions to the
2595Perl core.
2596
0d098d33
MHM
2597=head2 Backwards compatibility
2598
10e2eb10
FC
2599The Perl API changes over time. New functions are
2600added or the interfaces of existing functions are
2601changed. The C<Devel::PPPort> module tries to
0d098d33
MHM
2602provide compatibility code for some of these changes, so XS writers don't
2603have to code it themselves when supporting multiple versions of Perl.
2604
2605C<Devel::PPPort> generates a C header file F<ppport.h> that can also
10e2eb10 2606be run as a Perl script. To generate F<ppport.h>, run:
0d098d33
MHM
2607
2608 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2609
2610Besides checking existing XS code, the script can also be used to retrieve
2611compatibility information for various API calls using the C<--api-info>
10e2eb10 2612command line switch. For example:
0d098d33
MHM
2613
2614 % perl ppport.h --api-info=sv_magicext
2615
2616For details, see C<perldoc ppport.h>.
2617
a422fd2d
SC
2618=head1 Unicode Support
2619
10e2eb10 2620Perl 5.6.0 introduced Unicode support. It's important for porters and XS
a422fd2d
SC
2621writers to understand this support and make sure that the code they
2622write does not corrupt Unicode data.
2623
2624=head2 What B<is> Unicode, anyway?
2625
10e2eb10
FC
2626In the olden, less enlightened times, we all used to use ASCII. Most of
2627us did, anyway. The big problem with ASCII is that it's American. Well,
a422fd2d 2628no, that's not actually the problem; the problem is that it's not
10e2eb10 2629particularly useful for people who don't use the Roman alphabet. What
a422fd2d 2630used to happen was that particular languages would stick their own
10e2eb10 2631alphabet in the upper range of the sequence, between 128 and 255. Of
a422fd2d
SC
2632course, we then ended up with plenty of variants that weren't quite
2633ASCII, and the whole point of it being a standard was lost.
2634
2635Worse still, if you've got a language like Chinese or
2636Japanese that has hundreds or thousands of characters, then you really
2637can't fit them into a mere 256, so they had to forget about ASCII
2638altogether, and build their own systems using pairs of numbers to refer
2639to one character.
2640
2641To fix this, some people formed Unicode, Inc. and
2642produced a new character set containing all the characters you can
10e2eb10
FC
2643possibly think of and more. There are several ways of representing these
2644characters, and the one Perl uses is called UTF-8. UTF-8 uses
2645a variable number of bytes to represent a character. You can learn more
2575c402 2646about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2647
1e54db1a 2648=head2 How can I recognise a UTF-8 string?
a422fd2d 2649
10e2eb10
FC
2650You can't. This is because UTF-8 data is stored in bytes just like
2651non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d 2652capital E with a grave accent, is represented by the two bytes
10e2eb10
FC
2653C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2654has that byte sequence as well. So you can't tell just by looking - this
a422fd2d
SC
2655is what makes Unicode input an interesting problem.
2656
2575c402
JW
2657In general, you either have to know what you're dealing with, or you
2658have to guess. The API function C<is_utf8_string> can help; it'll tell
10e2eb10
FC
2659you if a string contains only valid UTF-8 characters. However, it can't
2660do the work for you. On a character-by-character basis,
49f4c4e4 2661C<is_utf8_char_buf>
2575c402 2662will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2663
1e54db1a 2664=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2665
1e54db1a 2666As mentioned above, UTF-8 uses a variable number of bytes to store a
10e2eb10
FC
2667character. Characters with values 0...127 are stored in one
2668byte, just like good ol' ASCII. Character 128 is stored as
2669C<v194.128>; this continues up to character 191, which is
2670C<v194.191>. Now we've run out of bits (191 is binary
2671C<10111111>) so we move on; 192 is C<v195.128>. And
a422fd2d
SC
2672so it goes on, moving to three bytes at character 2048.
2673
1e54db1a 2674Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2675how long the first character in it is with the C<UTF8SKIP> macro:
2676
2677 char *utf = "\305\233\340\240\201";
2678 I32 len;
2679
2680 len = UTF8SKIP(utf); /* len is 2 here */
2681 utf += len;
2682 len = UTF8SKIP(utf); /* len is 3 here */
2683
1e54db1a 2684Another way to skip over characters in a UTF-8 string is to use
a422fd2d 2685C<utf8_hop>, which takes a string and a number of characters to skip
10e2eb10 2686over. You're on your own about bounds checking, though, so don't use it
a422fd2d
SC
2687lightly.
2688
1e54db1a 2689All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2690so you can test if you need to do something special with this
2691character like this (the UTF8_IS_INVARIANT() is a macro that tests
9f98c7fe 2692whether the byte is encoded as a single byte even in UTF-8):
a422fd2d 2693
3a2263fe 2694 U8 *utf;
4b88fb76 2695 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
3a2263fe 2696 UV uv; /* Note: a UV, not a U8, not a char */
95701e00 2697 STRLEN len; /* length of character in bytes */
a422fd2d 2698
3a2263fe 2699 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2700 /* Must treat this as UTF-8 */
4b88fb76 2701 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
a422fd2d
SC
2702 else
2703 /* OK to treat this character as a byte */
2704 uv = *utf;
2705
4b88fb76 2706You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
95701e00 2707value of the character; the inverse function C<uvchr_to_utf8> is available
1e54db1a 2708for putting a UV into UTF-8:
a422fd2d 2709
3a2263fe 2710 if (!UTF8_IS_INVARIANT(uv))
a422fd2d 2711 /* Must treat this as UTF8 */
95701e00 2712 utf8 = uvchr_to_utf8(utf8, uv);
a422fd2d
SC
2713 else
2714 /* OK to treat this character as a byte */
2715 *utf8++ = uv;
2716
2717You B<must> convert characters to UVs using the above functions if
1e54db1a 2718you're ever in a situation where you have to match UTF-8 and non-UTF-8
10e2eb10 2719characters. You may not skip over UTF-8 characters in this case. If you
1e54db1a
JH
2720do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2721for instance, if your UTF-8 string contains C<v196.172>, and you skip
2722that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2723So don't do that!
2724
1e54db1a 2725=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2726
2727Currently, Perl deals with Unicode strings and non-Unicode strings
10e2eb10
FC
2728slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2729string is internally encoded as UTF-8. Without it, the byte value is the
2575c402 2730codepoint number and vice versa (in other words, the string is encoded
e1b711da 2731as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
c31cc9fc
FC
2732semantics). This flag is only meaningful if the SV is C<SvPOK>
2733or immediately after stringification via C<SvPV> or a similar
2734macro. You can check and manipulate this flag with the
2575c402 2735following macros:
a422fd2d
SC
2736
2737 SvUTF8(sv)
2738 SvUTF8_on(sv)
2739 SvUTF8_off(sv)
2740
2741This flag has an important effect on Perl's treatment of the string: if
2742Unicode data is not properly distinguished, regular expressions,
2743C<length>, C<substr> and other string handling operations will have
2744undesirable results.
2745
2746The problem comes when you have, for instance, a string that isn't
2575c402 2747flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
1e54db1a 2748especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2749
2750Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2751need be sure you don't accidentally knock it off while you're
10e2eb10 2752manipulating SVs. More specifically, you cannot expect to do this:
a422fd2d
SC
2753
2754 SV *sv;
2755 SV *nsv;
2756 STRLEN len;
2757 char *p;
2758
2759 p = SvPV(sv, len);
2760 frobnicate(p);
2761 nsv = newSVpvn(p, len);
2762
2763The C<char*> string does not tell you the whole story, and you can't
10e2eb10 2764copy or reconstruct an SV just by copying the string value. Check if the
c31cc9fc
FC
2765old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
2766accordingly:
a422fd2d
SC
2767
2768 p = SvPV(sv, len);
2769 frobnicate(p);
2770 nsv = newSVpvn(p, len);
2771 if (SvUTF8(sv))
2772 SvUTF8_on(nsv);
2773
2774In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2775not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2776appropriately.
2777
3a2263fe 2778Since just passing an SV to an XS function and copying the data of
2575c402 2779the SV is not enough to copy the UTF8 flags, even less right is just
3a2263fe
RGS
2780passing a C<char *> to an XS function.
2781
1e54db1a 2782=head2 How do I convert a string to UTF-8?
a422fd2d 2783
2575c402 2784If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
10e2eb10 2785one of the strings to UTF-8. If you've got an SV, the easiest way to do
2575c402 2786this is:
a422fd2d
SC
2787
2788 sv_utf8_upgrade(sv);
2789
2790However, you must not do this, for example:
2791
2792 if (!SvUTF8(left))
2793 sv_utf8_upgrade(left);
2794
2795If you do this in a binary operator, you will actually change one of the
b1866b2d 2796strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2797by the end user, it can cause problems in deficient code.
a422fd2d 2798
1e54db1a 2799Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
10e2eb10
FC
2800string argument. This is useful for having the data available for
2801comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2802C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2803the string contains any characters above 255 that can't be represented
2804in a single byte.
2805
2806=head2 Is there anything else I need to know?
2807
10e2eb10 2808Not really. Just remember these things:
a422fd2d
SC
2809
2810=over 3
2811
2812=item *
2813
10e2eb10 2814There's no way to tell if a string is UTF-8 or not. You can tell if an SV
c31cc9fc
FC
2815is UTF-8 by looking at its C<SvUTF8> flag after stringifying it
2816with C<SvPV> or a similar macro. Don't forget to set the flag if
10e2eb10 2817something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2818it's not - if you pass on the PV to somewhere, pass on the flag too.
2819
2820=item *
2821
4b88fb76 2822If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3a2263fe 2823unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2824
2825=item *
2826
1e54db1a 2827When writing a character C<uv> to a UTF-8 string, B<always> use
95701e00 2828C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
3a2263fe 2829you can use C<*s = uv>.
a422fd2d
SC
2830
2831=item *
2832
10e2eb10
FC
2833Mixing UTF-8 and non-UTF-8 strings is
2834tricky. Use C<bytes_to_utf8> to get
2bbc8d55 2835a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
2836
2837=back
2838
53e06cf0
SC
2839=head1 Custom Operators
2840
2a0fd0f1 2841Custom operator support is an experimental feature that allows you to
10e2eb10 2842define your own ops. This is primarily to allow the building of
53e06cf0
SC
2843interpreters for other languages in the Perl core, but it also allows
2844optimizations through the creation of "macro-ops" (ops which perform the
2845functions of multiple ops which are usually executed together, such as
1aa6ea50 2846C<gvsv, gvsv, add>.)
53e06cf0 2847
10e2eb10 2848This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0 2849core does not "know" anything special about this op type, and so it will
10e2eb10 2850not be involved in any optimizations. This also means that you can
53e06cf0
SC
2851define your custom ops to be any op structure - unary, binary, list and
2852so on - you like.
2853
10e2eb10
FC
2854It's important to know what custom operators won't do for you. They
2855won't let you add new syntax to Perl, directly. They won't even let you
2856add new keywords, directly. In fact, they won't change the way Perl
2857compiles a program at all. You have to do those changes yourself, after
2858Perl has compiled the program. You do this either by manipulating the op
53e06cf0
SC
2859tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2860a custom peephole optimizer with the C<optimize> module.
2861
2862When you do this, you replace ordinary Perl ops with custom ops by
407f86e1 2863creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
10e2eb10
FC
2864PP function. This should be defined in XS code, and should look like
2865the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
53e06cf0
SC
2866takes the appropriate number of values from the stack, and you are
2867responsible for adding stack marks if necessary.
2868
2869You should also "register" your op with the Perl interpreter so that it
10e2eb10 2870can produce sensible error and warning messages. Since it is possible to
53e06cf0 2871have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
9733086d 2872Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
10e2eb10 2873it is dealing with. You should create an C<XOP> structure for each
9733086d
BM
2874ppaddr you use, set the properties of the custom op with
2875C<XopENTRY_set>, and register the structure against the ppaddr using
10e2eb10 2876C<Perl_custom_op_register>. A trivial example might look like:
9733086d
BM
2877
2878 static XOP my_xop;
2879 static OP *my_pp(pTHX);
2880
2881 BOOT:
2882 XopENTRY_set(&my_xop, xop_name, "myxop");
2883 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2884 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2885
2886The available fields in the structure are:
2887
2888=over 4
2889
2890=item xop_name
2891
10e2eb10 2892A short name for your op. This will be included in some error messages,
9733086d
BM
2893and will also be returned as C<< $op->name >> by the L<B|B> module, so
2894it will appear in the output of module like L<B::Concise|B::Concise>.
2895
2896=item xop_desc
2897
2898A short description of the function of the op.
2899
2900=item xop_class
2901
10e2eb10 2902Which of the various C<*OP> structures this op uses. This should be one of
9733086d
BM
2903the C<OA_*> constants from F<op.h>, namely
2904
2905=over 4
2906
2907=item OA_BASEOP
2908
2909=item OA_UNOP
2910
2911=item OA_BINOP
2912
2913=item OA_LOGOP
2914
2915=item OA_LISTOP
2916
2917=item OA_PMOP
2918
2919=item OA_SVOP
2920
2921=item OA_PADOP
2922
2923=item OA_PVOP_OR_SVOP
2924
10e2eb10 2925This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
9733086d
BM
2926the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
2927
2928=item OA_LOOP
2929
2930=item OA_COP
2931
2932=back
2933
2934The other C<OA_*> constants should not be used.
2935
2936=item xop_peep
2937
2938This member is of type C<Perl_cpeep_t>, which expands to C<void
10e2eb10 2939(*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
9733086d 2940will be called from C<Perl_rpeep> when ops of this type are encountered
10e2eb10 2941by the peephole optimizer. I<o> is the OP that needs optimizing;
9733086d
BM
2942I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
2943
2944=back
53e06cf0 2945
e7d4c058 2946C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 2947
954c1994 2948=head1 AUTHORS
e89caa19 2949
954c1994 2950Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
2951E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2952itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 2953
954c1994
GS
2954With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2955Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2956Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2957Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 2958
954c1994 2959=head1 SEE ALSO
cb1a09d0 2960
ba555bf5 2961L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>