This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlguts: consistent spaces after dots
[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 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 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.
9e9796d6
JH
319For example, a tied SV may have a valid underlying value in the IV slot
320(so SvIOKp is true), but the data should be accessed via the FETCH
10e2eb10 321routine rather than directly, so SvIOK is false. Another is when
d7f8936a 322numeric conversion has occurred and precision has been lost: only the
10e2eb10 323private flag is set on 'lossy' values. So when an NV is converted to an
9e9796d6
JH
324IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
325
07fa94a1 326In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 327
54310121 328=head2 Working with AVs
a0d0e21e 329
07fa94a1
JO
330There are two ways to create and load an AV. The first method creates an
331empty AV:
a0d0e21e
LW
332
333 AV* newAV();
334
54310121 335The second method both creates the AV and initially populates it with SVs:
a0d0e21e 336
c70927a6 337 AV* av_make(SSize_t num, SV **ptr);
a0d0e21e 338
5f05dabc 339The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 340AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 341
da8c5729 342Once the AV has been created, the following operations are possible on it:
a0d0e21e
LW
343
344 void av_push(AV*, SV*);
345 SV* av_pop(AV*);
346 SV* av_shift(AV*);
c70927a6 347 void av_unshift(AV*, SSize_t num);
a0d0e21e
LW
348
349These should be familiar operations, with the exception of C<av_unshift>.
350This routine adds C<num> elements at the front of the array with the C<undef>
351value. You must then use C<av_store> (described below) to assign values
352to these new elements.
353
354Here are some other functions:
355
c70927a6
FC
356 SSize_t av_top_index(AV*);
357 SV** av_fetch(AV*, SSize_t key, I32 lval);
358 SV** av_store(AV*, SSize_t key, SV* val);
a0d0e21e 359
dab460cd 360The C<av_top_index> function returns the highest index value in an array (just
5f05dabc 361like $#array in Perl). If the array is empty, -1 is returned. The
362C<av_fetch> function returns the value at index C<key>, but if C<lval>
363is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
364The C<av_store> function stores the value C<val> at index C<key>, and does
365not increment the reference count of C<val>. Thus the caller is responsible
366for taking care of that, and if C<av_store> returns NULL, the caller will
367have to decrement the reference count to avoid a memory leak. Note that
368C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
369return value.
d1b91892 370
da8c5729
MH
371A few more:
372
a0d0e21e 373 void av_clear(AV*);
a0d0e21e 374 void av_undef(AV*);
c70927a6 375 void av_extend(AV*, SSize_t key);
5f05dabc 376
377The C<av_clear> function deletes all the elements in the AV* array, but
378does not actually delete the array itself. The C<av_undef> function will
379delete all the elements in the array plus the array itself. The
adc882cf
GS
380C<av_extend> function extends the array so that it contains at least C<key+1>
381elements. If C<key+1> is less than the currently allocated length of the array,
382then nothing is done.
a0d0e21e
LW
383
384If you know the name of an array variable, you can get a pointer to its AV
385by using the following:
386
cbfd0a87 387 AV* get_av("package::varname", 0);
a0d0e21e
LW
388
389This returns NULL if the variable does not exist.
390
04343c6d
GS
391See L<Understanding the Magic of Tied Hashes and Arrays> for more
392information on how to use the array access functions on tied arrays.
393
54310121 394=head2 Working with HVs
a0d0e21e
LW
395
396To create an HV, you use the following routine:
397
398 HV* newHV();
399
da8c5729 400Once the HV has been created, the following operations are possible on it:
a0d0e21e 401
08105a92
GS
402 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
403 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 404
5f05dabc 405The C<klen> parameter is the length of the key being passed in (Note that
406you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
407length of the key). The C<val> argument contains the SV pointer to the
54310121 408scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc 409you want C<hv_store> to calculate it for you). The C<lval> parameter
410indicates whether this fetch is actually a part of a store operation, in
411which case a new undefined value will be added to the HV with the supplied
412key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 413
5f05dabc 414Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
415C<SV*>. To access the scalar value, you must first dereference the return
416value. However, you should check to make sure that the return value is
417not NULL before dereferencing it.
a0d0e21e 418
da8c5729
MH
419The first of these two functions checks if a hash table entry exists, and the
420second deletes it.
a0d0e21e 421
08105a92
GS
422 bool hv_exists(HV*, const char* key, U32 klen);
423 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 424
5f05dabc 425If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
426create and return a mortal copy of the deleted value.
427
a0d0e21e
LW
428And more miscellaneous functions:
429
430 void hv_clear(HV*);
a0d0e21e 431 void hv_undef(HV*);
5f05dabc 432
433Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
434table but does not actually delete the hash table. The C<hv_undef> deletes
435both the entries and the hash table itself.
a0d0e21e 436
a9b0660e 437Perl keeps the actual data in a linked list of structures with a typedef of HE.
d1b91892
AD
438These contain the actual key and value pointers (plus extra administrative
439overhead). The key is a string pointer; the value is an C<SV*>. However,
440once you have an C<HE*>, to get the actual key and value, use the routines
441specified below.
442
a0d0e21e
LW
443 I32 hv_iterinit(HV*);
444 /* Prepares starting point to traverse hash table */
445 HE* hv_iternext(HV*);
446 /* Get the next entry, and return a pointer to a
447 structure that has both the key and value */
448 char* hv_iterkey(HE* entry, I32* retlen);
449 /* Get the key from an HE structure and also return
450 the length of the key string */
cb1a09d0 451 SV* hv_iterval(HV*, HE* entry);
d1be9408 452 /* Return an SV pointer to the value of the HE
a0d0e21e 453 structure */
cb1a09d0 454 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
455 /* This convenience routine combines hv_iternext,
456 hv_iterkey, and hv_iterval. The key and retlen
457 arguments are return values for the key and its
458 length. The value is returned in the SV* argument */
a0d0e21e
LW
459
460If you know the name of a hash variable, you can get a pointer to its HV
461by using the following:
462
6673a63c 463 HV* get_hv("package::varname", 0);
a0d0e21e
LW
464
465This returns NULL if the variable does not exist.
466
a43e7901 467The hash algorithm is defined in the C<PERL_HASH> macro:
a0d0e21e 468
a43e7901 469 PERL_HASH(hash, key, klen)
ab192400 470
a43e7901
YO
471The exact implementation of this macro varies by architecture and version
472of perl, and the return value may change per invocation, so the value
473is only valid for the duration of a single perl process.
a0d0e21e 474
04343c6d
GS
475See L<Understanding the Magic of Tied Hashes and Arrays> for more
476information on how to use the hash access functions on tied hashes.
477
1e422769 478=head2 Hash API Extensions
479
480Beginning with version 5.004, the following functions are also supported:
481
482 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
483 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 484
1e422769 485 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
486 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 487
1e422769 488 SV* hv_iterkeysv (HE* entry);
489
490Note that these functions take C<SV*> keys, which simplifies writing
491of extension code that deals with hash structures. These functions
492also allow passing of C<SV*> keys to C<tie> functions without forcing
493you to stringify the keys (unlike the previous set of functions).
494
495They also return and accept whole hash entries (C<HE*>), making their
496use more efficient (since the hash number for a particular string
4a4eefd0
GS
497doesn't have to be recomputed every time). See L<perlapi> for detailed
498descriptions.
1e422769 499
500The following macros must always be used to access the contents of hash
501entries. Note that the arguments to these macros must be simple
502variables, since they may get evaluated more than once. See
4a4eefd0 503L<perlapi> for detailed descriptions of these macros.
1e422769 504
505 HePV(HE* he, STRLEN len)
506 HeVAL(HE* he)
507 HeHASH(HE* he)
508 HeSVKEY(HE* he)
509 HeSVKEY_force(HE* he)
510 HeSVKEY_set(HE* he, SV* sv)
511
512These two lower level macros are defined, but must only be used when
513dealing with keys that are not C<SV*>s:
514
515 HeKEY(HE* he)
516 HeKLEN(HE* he)
517
04343c6d
GS
518Note that both C<hv_store> and C<hv_store_ent> do not increment the
519reference count of the stored C<val>, which is the caller's responsibility.
520If these functions return a NULL value, the caller will usually have to
521decrement the reference count of C<val> to avoid a memory leak.
1e422769 522
a9381218
MHM
523=head2 AVs, HVs and undefined values
524
10e2eb10
FC
525Sometimes you have to store undefined values in AVs or HVs. Although
526this may be a rare case, it can be tricky. That's because you're
a9381218
MHM
527used to using C<&PL_sv_undef> if you need an undefined SV.
528
529For example, intuition tells you that this XS code:
530
531 AV *av = newAV();
532 av_store( av, 0, &PL_sv_undef );
533
534is equivalent to this Perl code:
535
536 my @av;
537 $av[0] = undef;
538
f3c4ec28 539Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
a9381218
MHM
540for indicating that an array element has not yet been initialized.
541Thus, C<exists $av[0]> would be true for the above Perl code, but
f3c4ec28
FC
542false for the array generated by the XS code. In perl 5.20, storing
543&PL_sv_undef will create a read-only element, because the scalar
544&PL_sv_undef itself is stored, not a copy.
a9381218 545
f3c4ec28 546Similar problems can occur when storing C<&PL_sv_undef> in HVs:
a9381218
MHM
547
548 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
549
550This will indeed make the value C<undef>, but if you try to modify
551the value of C<key>, you'll get the following error:
552
553 Modification of non-creatable hash value attempted
554
555In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
10e2eb10 556in restricted hashes. This caused such hash entries not to appear
a9381218
MHM
557when iterating over the hash or when checking for the keys
558with the C<hv_exists> function.
559
8abccac8 560You can run into similar problems when you store C<&PL_sv_yes> or
10e2eb10 561C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
562will give you the following error:
563
564 Modification of a read-only value attempted
565
566To make a long story short, you can use the special variables
8abccac8 567C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
568HVs, but you have to make sure you know what you're doing.
569
570Generally, if you want to store an undefined value in an AV
571or HV, you should not use C<&PL_sv_undef>, but rather create a
572new undefined value using the C<newSV> function, for example:
573
574 av_store( av, 42, newSV(0) );
575 hv_store( hv, "foo", 3, newSV(0), 0 );
576
a0d0e21e
LW
577=head2 References
578
d1b91892 579References are a special type of scalar that point to other data types
a9b0660e 580(including other references).
a0d0e21e 581
07fa94a1 582To create a reference, use either of the following functions:
a0d0e21e 583
5f05dabc 584 SV* newRV_inc((SV*) thing);
585 SV* newRV_noinc((SV*) thing);
a0d0e21e 586
5f05dabc 587The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
588functions are identical except that C<newRV_inc> increments the reference
589count of the C<thing>, while C<newRV_noinc> does not. For historical
590reasons, C<newRV> is a synonym for C<newRV_inc>.
591
592Once you have a reference, you can use the following macro to dereference
593the reference:
a0d0e21e
LW
594
595 SvRV(SV*)
596
597then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 598C<AV*> or C<HV*>, if required.
a0d0e21e 599
d1b91892 600To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
601
602 SvROK(SV*)
603
07fa94a1
JO
604To discover what type of value the reference refers to, use the following
605macro and then check the return value.
d1b91892
AD
606
607 SvTYPE(SvRV(SV*))
608
609The most useful types that will be returned are:
610
a5e62da0
FC
611 < SVt_PVAV Scalar
612 SVt_PVAV Array
613 SVt_PVHV Hash
614 SVt_PVCV Code
615 SVt_PVGV Glob (possibly a file handle)
616
617See L<perlapi/svtype> for more details.
d1b91892 618
cb1a09d0
AD
619=head2 Blessed References and Class Objects
620
06f6df17 621References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
622OO lexicon, an object is simply a reference that has been blessed into a
623package (or class). Once blessed, the programmer may now use the reference
624to access the various methods in the class.
625
626A reference can be blessed into a package with the following function:
627
628 SV* sv_bless(SV* sv, HV* stash);
629
06f6df17
RGS
630The C<sv> argument must be a reference value. The C<stash> argument
631specifies which class the reference will belong to. See
2ae324a7 632L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
633
634/* Still under construction */
635
ddd2cc91
DM
636The following function upgrades rv to reference if not already one.
637Creates a new SV for rv to point to. If C<classname> is non-null, the SV
638is blessed into the specified class. SV is returned.
cb1a09d0 639
08105a92 640 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 641
ddd2cc91
DM
642The following three functions copy integer, unsigned integer or double
643into an SV whose reference is C<rv>. SV is blessed if C<classname> is
644non-null.
cb1a09d0 645
08105a92 646 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 647 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 648 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 649
ddd2cc91
DM
650The following function copies the pointer value (I<the address, not the
651string!>) into an SV whose reference is rv. SV is blessed if C<classname>
652is non-null.
cb1a09d0 653
ddd2cc91 654 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
cb1a09d0 655
a9b0660e 656The following function copies a string into an SV whose reference is C<rv>.
ddd2cc91
DM
657Set length to 0 to let Perl calculate the string length. SV is blessed if
658C<classname> is non-null.
cb1a09d0 659
a9b0660e
KW
660 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
661 STRLEN length);
cb1a09d0 662
ddd2cc91
DM
663The following function tests whether the SV is blessed into the specified
664class. It does not check inheritance relationships.
9abd00ed 665
08105a92 666 int sv_isa(SV* sv, const char* name);
9abd00ed 667
ddd2cc91 668The following function tests whether the SV is a reference to a blessed object.
9abd00ed
GS
669
670 int sv_isobject(SV* sv);
671
ddd2cc91 672The following function tests whether the SV is derived from the specified
10e2eb10
FC
673class. SV can be either a reference to a blessed object or a string
674containing a class name. This is the function implementing the
ddd2cc91 675C<UNIVERSAL::isa> functionality.
9abd00ed 676
08105a92 677 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 678
00aadd71 679To check if you've got an object derived from a specific class you have
9abd00ed
GS
680to write:
681
682 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 683
5f05dabc 684=head2 Creating New Variables
cb1a09d0 685
5f05dabc 686To create a new Perl variable with an undef value which can be accessed from
687your Perl script, use the following routines, depending on the variable type.
cb1a09d0 688
64ace3f8 689 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 690 AV* get_av("package::varname", GV_ADD);
6673a63c 691 HV* get_hv("package::varname", GV_ADD);
cb1a09d0 692
058a5f6c 693Notice the use of GV_ADD as the second parameter. The new variable can now
cb1a09d0
AD
694be set, using the routines appropriate to the data type.
695
5f05dabc 696There are additional macros whose values may be bitwise OR'ed with the
058a5f6c 697C<GV_ADD> argument to enable certain extra features. Those bits are:
cb1a09d0 698
9a68f1db
SB
699=over
700
701=item GV_ADDMULTI
702
703Marks the variable as multiply defined, thus preventing the:
704
705 Name <varname> used only once: possible typo
706
707warning.
708
9a68f1db
SB
709=item GV_ADDWARN
710
711Issues the warning:
712
713 Had to create <varname> unexpectedly
714
715if the variable did not exist before the function was called.
716
717=back
cb1a09d0 718
07fa94a1
JO
719If you do not specify a package name, the variable is created in the current
720package.
cb1a09d0 721
5f05dabc 722=head2 Reference Counts and Mortality
a0d0e21e 723
10e2eb10 724Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 725AVs, or HVs (xV for short in the following) start their life with a
55497cff 726reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 727then it will be destroyed and its memory made available for reuse.
55497cff 728
729This normally doesn't happen at the Perl level unless a variable is
5f05dabc 730undef'ed or the last variable holding a reference to it is changed or
731overwritten. At the internal level, however, reference counts can be
55497cff 732manipulated with the following macros:
733
734 int SvREFCNT(SV* sv);
5f05dabc 735 SV* SvREFCNT_inc(SV* sv);
55497cff 736 void SvREFCNT_dec(SV* sv);
737
738However, there is one other function which manipulates the reference
07fa94a1
JO
739count of its argument. The C<newRV_inc> function, you will recall,
740creates a reference to the specified argument. As a side effect,
741it increments the argument's reference count. If this is not what
742you want, use C<newRV_noinc> instead.
743
744For example, imagine you want to return a reference from an XSUB function.
745Inside the XSUB routine, you create an SV which initially has a reference
746count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc 747This returns the reference as a new SV, but the reference count of the
748SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
749return the reference from the XSUB routine and forget about the SV.
750But Perl hasn't! Whenever the returned reference is destroyed, the
751reference count of the original SV is decreased to one and nothing happens.
752The SV will hang around without any way to access it until Perl itself
753terminates. This is a memory leak.
5f05dabc 754
755The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
756C<newRV_inc>. Then, if and when the last reference is destroyed,
757the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 758stopping any memory leak.
55497cff 759
5f05dabc 760There are some convenience functions available that can help with the
54310121 761destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
762An xV that is mortal has had its reference count marked to be decremented,
763but not actually decremented, until "a short time later". Generally the
764term "short time later" means a single Perl statement, such as a call to
54310121 765an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
766reference count decremented depends on two macros, SAVETMPS and FREETMPS.
767See L<perlcall> and L<perlxs> for more details on these macros.
55497cff 768
769"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
770However, if you mortalize a variable twice, the reference count will
771later be decremented twice.
772
00aadd71
NIS
773"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
774For example an SV which is created just to pass a number to a called sub
06f6df17 775is made mortal to have it cleaned up automatically when it's popped off
10e2eb10 776the stack. Similarly, results returned by XSUBs (which are pushed on the
06f6df17 777stack) are often made mortal.
a0d0e21e
LW
778
779To create a mortal variable, use the functions:
780
781 SV* sv_newmortal()
782 SV* sv_2mortal(SV*)
783 SV* sv_mortalcopy(SV*)
784
00aadd71 785The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc 786SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
787third creates a mortal copy of an existing SV.
da8c5729 788Because C<sv_newmortal> gives the new SV no value, it must normally be given one
9a68f1db 789via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
790
791 SV *tmp = sv_newmortal();
792 sv_setiv(tmp, an_integer);
793
794As that is multiple C statements it is quite common so see this idiom instead:
795
796 SV *tmp = sv_2mortal(newSViv(an_integer));
797
798
799You should be careful about creating mortal variables. Strange things
800can happen if you make the same value mortal within multiple contexts,
10e2eb10
FC
801or if you make a variable mortal multiple
802times. Thinking of "Mortalization"
00aadd71 803as deferred C<SvREFCNT_dec> should help to minimize such problems.
da8c5729 804For example if you are passing an SV which you I<know> has a high enough REFCNT
00aadd71
NIS
805to survive its use on the stack you need not do any mortalization.
806If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
807making a C<sv_mortalcopy> is safer.
a0d0e21e 808
ac036724 809The mortal routines are not just for SVs; AVs and HVs can be
faed5253 810made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 811C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 812
5f05dabc 813=head2 Stashes and Globs
a0d0e21e 814
06f6df17
RGS
815A B<stash> is a hash that contains all variables that are defined
816within a package. Each key of the stash is a symbol
aa689395 817name (shared by all the different types of objects that have the same
818name), and each value in the hash table is a GV (Glob Value). This GV
819in turn contains references to the various objects of that name,
820including (but not limited to) the following:
cb1a09d0 821
a0d0e21e
LW
822 Scalar Value
823 Array Value
824 Hash Value
a3cb178b 825 I/O Handle
a0d0e21e
LW
826 Format
827 Subroutine
828
06f6df17
RGS
829There is a single stash called C<PL_defstash> that holds the items that exist
830in the C<main> package. To get at the items in other packages, append the
831string "::" to the package name. The items in the C<Foo> package are in
832the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
833in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 834
d1b91892 835To get the stash pointer for a particular package, use the function:
a0d0e21e 836
da51bb9b
NC
837 HV* gv_stashpv(const char* name, I32 flags)
838 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
839
840The first function takes a literal string, the second uses the string stored
d1b91892 841in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 842C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
843
844The name that C<gv_stash*v> wants is the name of the package whose symbol table
845you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
846packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
847language itself.
a0d0e21e
LW
848
849Alternately, if you have an SV that is a blessed reference, you can find
850out the stash pointer by using:
851
852 HV* SvSTASH(SvRV(SV*));
853
854then use the following to get the package name itself:
855
856 char* HvNAME(HV* stash);
857
5f05dabc 858If you need to bless or re-bless an object you can use the following
859function:
a0d0e21e
LW
860
861 SV* sv_bless(SV*, HV* stash)
862
863where the first argument, an C<SV*>, must be a reference, and the second
864argument is a stash. The returned C<SV*> can now be used in the same way
865as any other SV.
866
d1b91892
AD
867For more information on references and blessings, consult L<perlref>.
868
54310121 869=head2 Double-Typed SVs
0a753a76 870
871Scalar variables normally contain only one type of value, an integer,
872double, pointer, or reference. Perl will automatically convert the
873actual scalar data from the stored type into the requested type.
874
875Some scalar variables contain more than one type of scalar data. For
876example, the variable C<$!> contains either the numeric value of C<errno>
877or its string equivalent from either C<strerror> or C<sys_errlist[]>.
878
879To force multiple data values into an SV, you must do two things: use the
880C<sv_set*v> routines to add the additional scalar type, then set a flag
881so that Perl will believe it contains more than one type of data. The
882four macros to set the flags are:
883
884 SvIOK_on
885 SvNOK_on
886 SvPOK_on
887 SvROK_on
888
889The particular macro you must use depends on which C<sv_set*v> routine
890you called first. This is because every C<sv_set*v> routine turns on
891only the bit for the particular type of data being set, and turns off
892all the rest.
893
894For example, to create a new Perl variable called "dberror" that contains
895both the numeric and descriptive string error values, you could use the
896following code:
897
898 extern int dberror;
899 extern char *dberror_list;
900
64ace3f8 901 SV* sv = get_sv("dberror", GV_ADD);
0a753a76 902 sv_setiv(sv, (IV) dberror);
903 sv_setpv(sv, dberror_list[dberror]);
904 SvIOK_on(sv);
905
906If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
907macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
908
909=head2 Magic Variables
a0d0e21e 910
d1b91892
AD
911[This section still under construction. Ignore everything here. Post no
912bills. Everything not permitted is forbidden.]
913
d1b91892
AD
914Any SV may be magical, that is, it has special features that a normal
915SV does not have. These features are stored in the SV structure in a
5f05dabc 916linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
917
918 struct magic {
919 MAGIC* mg_moremagic;
920 MGVTBL* mg_virtual;
921 U16 mg_private;
922 char mg_type;
923 U8 mg_flags;
b205eb13 924 I32 mg_len;
d1b91892
AD
925 SV* mg_obj;
926 char* mg_ptr;
d1b91892
AD
927 };
928
929Note this is current as of patchlevel 0, and could change at any time.
930
931=head2 Assigning Magic
932
933Perl adds magic to an SV using the sv_magic function:
934
a9b0660e 935 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
936
937The C<sv> argument is a pointer to the SV that is to acquire a new magical
938feature.
939
940If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
10e2eb10
FC
941convert C<sv> to type C<SVt_PVMG>.
942Perl then continues by adding new magic
645c22ef
DM
943to the beginning of the linked list of magical features. Any prior entry
944of the same type of magic is deleted. Note that this can be overridden,
945and multiple instances of the same type of magic can be associated with an
946SV.
d1b91892 947
54310121 948The C<name> and C<namlen> arguments are used to associate a string with
10e2eb10 949the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
950C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
951C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
952whether C<namlen> is greater than zero or equal to zero respectively. As a
953special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
954to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
955
956The sv_magic function uses C<how> to determine which, if any, predefined
957"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 958See the L<Magic Virtual Tables> section below. The C<how> argument is also
10e2eb10
FC
959stored in the C<mg_type> field. The value of
960C<how> should be chosen from the set of macros
961C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 962these macros were added, Perl internals used to directly use character
14befaf4 963literals, so you may occasionally come across old code or documentation
75d0f26d 964referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
965
966The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
967structure. If it is not the same as the C<sv> argument, the reference
968count of the C<obj> object is incremented. If it is the same, or if
645c22ef 969the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 970then C<obj> is merely stored, without the reference count being incremented.
d1b91892 971
2d8d5d5a
SH
972See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
973to an SV.
974
cb1a09d0
AD
975There is also a function to add magic to an C<HV>:
976
977 void hv_magic(HV *hv, GV *gv, int how);
978
979This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
980
981To remove the magic from an SV, call the function sv_unmagic:
982
70a53b35 983 int sv_unmagic(SV *sv, int type);
cb1a09d0
AD
984
985The C<type> argument should be equal to the C<how> value when the C<SV>
986was initially made magical.
987
f6ee7b17 988However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
10e2eb10
FC
989C<SV>. If you want to remove only certain
990magic of a C<type> based on the magic
f6ee7b17
FR
991virtual table, use C<sv_unmagicext> instead:
992
993 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
994
d1b91892
AD
995=head2 Magic Virtual Tables
996
d1be9408 997The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
998C<MGVTBL>, which is a structure of function pointers and stands for
999"Magic Virtual Table" to handle the various operations that might be
1000applied to that variable.
1001
301cb7e8
DM
1002The C<MGVTBL> has five (or sometimes eight) pointers to the following
1003routine types:
d1b91892
AD
1004
1005 int (*svt_get)(SV* sv, MAGIC* mg);
1006 int (*svt_set)(SV* sv, MAGIC* mg);
1007 U32 (*svt_len)(SV* sv, MAGIC* mg);
1008 int (*svt_clear)(SV* sv, MAGIC* mg);
1009 int (*svt_free)(SV* sv, MAGIC* mg);
1010
a9b0660e
KW
1011 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1012 const char *name, I32 namlen);
301cb7e8
DM
1013 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1014 int (*svt_local)(SV *nsv, MAGIC *mg);
1015
1016
06f6df17 1017This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
S
1018currently 32 types. These different structures contain pointers to various
1019routines that perform additional actions depending on which function is
1020being called.
d1b91892 1021
a9b0660e
KW
1022 Function pointer Action taken
1023 ---------------- ------------
1024 svt_get Do something before the value of the SV is
1025 retrieved.
1026 svt_set Do something after the SV is assigned a value.
1027 svt_len Report on the SV's length.
1028 svt_clear Clear something the SV represents.
1029 svt_free Free any extra storage associated with the SV.
d1b91892 1030
a9b0660e
KW
1031 svt_copy copy tied variable magic to a tied element
1032 svt_dup duplicate a magic structure during thread cloning
1033 svt_local copy magic to local value during 'local'
301cb7e8 1034
d1b91892 1035For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1036to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1037
1038 { magic_get, magic_set, magic_len, 0, 0 }
1039
14befaf4
DM
1040Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1041if a get operation is being performed, the routine C<magic_get> is
1042called. All the various routines for the various magical types begin
1043with C<magic_>. NOTE: the magic routines are not considered part of
1044the Perl API, and may not be exported by the Perl library.
d1b91892 1045
301cb7e8
DM
1046The last three slots are a recent addition, and for source code
1047compatibility they are only checked for if one of the three flags
10e2eb10
FC
1048MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1049This means that most code can continue declaring
1050a vtable as a 5-element value. These three are
301cb7e8
DM
1051currently used exclusively by the threading code, and are highly subject
1052to change.
1053
d1b91892
AD
1054The current kinds of Magic Virtual Tables are:
1055
f1f5ddd7
FC
1056=for comment
1057This table is generated by regen/mg_vtable.pl. Any changes made here
1058will be lost.
1059
1060=for mg_vtable.pl begin
1061
a9b0660e 1062 mg_type
bd6e6c12
FC
1063 (old-style char and macro) MGVTBL Type of magic
1064 -------------------------- ------ -------------
1065 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1066 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1067 % PERL_MAGIC_rhash (none) extra data for restricted
1068 hashes
2a388207 1069 & PERL_MAGIC_proto (none) my sub prototype CV
bd6e6c12
FC
1070 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1071 : PERL_MAGIC_symtab (none) extra data for symbol
1072 tables
1073 < PERL_MAGIC_backref vtbl_backref for weak ref data
1074 @ PERL_MAGIC_arylen_p (none) to move arylen out of XPVAV
1075 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1076 (fast string search)
1077 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1078 (AMT) on stash
1079 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1080 (@+ and @- vars)
1081 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1082 element
1083 E PERL_MAGIC_env vtbl_env %ENV hash
1084 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
eccba044 1085 f PERL_MAGIC_fm vtbl_regexp Formline
bd6e6c12 1086 ('compiled' format)
bd6e6c12
FC
1087 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1088 H PERL_MAGIC_hints vtbl_hints %^H hash
1089 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1090 I PERL_MAGIC_isa vtbl_isa @ISA array
1091 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1092 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1093 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1094 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1095 element
1096 N PERL_MAGIC_shared (none) Shared between threads
1097 n PERL_MAGIC_shared_scalar (none) Shared between threads
1098 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1099 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1100 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1101 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1102 r PERL_MAGIC_qr vtbl_regexp precompiled qr// regex
1103 S PERL_MAGIC_sig (none) %SIG hash
1104 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1105 t PERL_MAGIC_taint vtbl_taint Taintedness
1106 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1107 extensions
1108 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1109 extensions
4499db73 1110 V PERL_MAGIC_vstring (none) SV was vstring literal
bd6e6c12
FC
1111 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1112 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1113 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1114 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1115 variable / smart parameter
1116 vivification
1117 ] PERL_MAGIC_checkcall vtbl_checkcall inlining/mutation of call
1118 to this CV
1119 ~ PERL_MAGIC_ext (none) Available for use by
1120 extensions
0cbee0a4 1121
f1f5ddd7 1122=for mg_vtable.pl end
d1b91892 1123
68dc0745 1124When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1125uppercase letter is typically used to represent some kind of composite type
1126(a list or a hash), and the lowercase letter is used to represent an element
10e2eb10 1127of that composite type. Some internals code makes use of this case
92f0c265 1128relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1129
1130The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1131specifically for use by extensions and will not be used by perl itself.
1132Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1133to variables (typically objects). This is especially useful because
1134there is no way for normal perl code to corrupt this private information
1135(unlike using extra elements of a hash object).
1136
1137Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1138C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1139C<mg_ptr> field points to a C<ufuncs> structure:
1140
1141 struct ufuncs {
a9402793
AB
1142 I32 (*uf_val)(pTHX_ IV, SV*);
1143 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1144 IV uf_index;
1145 };
1146
1147When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1148function will be called with C<uf_index> as the first arg and a pointer to
1149the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1150magic is shown below. Note that the ufuncs structure is copied by
1151sv_magic, so you can safely allocate it on the stack.
1152
1153 void
1154 Umagic(sv)
1155 SV *sv;
1156 PREINIT:
1157 struct ufuncs uf;
1158 CODE:
1159 uf.uf_val = &my_get_fn;
1160 uf.uf_set = &my_set_fn;
1161 uf.uf_index = 0;
14befaf4 1162 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1163
1e73acc8
AS
1164Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1165
1166For hashes there is a specialized hook that gives control over hash
1167keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1168if the "set" function in the C<ufuncs> structure is NULL. The hook
1169is activated whenever the hash is accessed with a key specified as
1170an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1171C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1172through the functions without the C<..._ent> suffix circumvents the
4509d391 1173hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1e73acc8 1174
14befaf4
DM
1175Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1176or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1177extra care to avoid conflict. Typically only using the magic on
1178objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1179For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1180C<MGVTBL>, even if all its fields will be C<0>, so that individual
1181C<MAGIC> pointers can be identified as a particular kind of magic
10e2eb10 1182using their magic virtual table. C<mg_findext> provides an easy way
f6ee7b17 1183to do that:
2f07f21a
FR
1184
1185 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1186
1187 MAGIC *mg;
f6ee7b17
FR
1188 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1189 /* this is really ours, not another module's PERL_MAGIC_ext */
1190 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1191 ...
2f07f21a 1192 }
5f05dabc 1193
ef50df4b
GS
1194Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1195earlier do B<not> invoke 'set' magic on their targets. This must
1196be done by the user either by calling the C<SvSETMAGIC()> macro after
1197calling these functions, or by using one of the C<sv_set*_mg()> or
1198C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1199C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1200obtained from external sources in functions that don't handle magic.
4a4eefd0 1201See L<perlapi> for a description of these functions.
189b2af5
GS
1202For example, calls to the C<sv_cat*()> functions typically need to be
1203followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1204since their implementation handles 'get' magic.
1205
d1b91892
AD
1206=head2 Finding Magic
1207
a9b0660e
KW
1208 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1209 * type */
f6ee7b17
FR
1210
1211This routine returns a pointer to a C<MAGIC> structure stored in the SV.
10e2eb10
FC
1212If the SV does not have that magical
1213feature, C<NULL> is returned. If the
f6ee7b17 1214SV has multiple instances of that magical feature, the first one will be
10e2eb10
FC
1215returned. C<mg_findext> can be used
1216to find a C<MAGIC> structure of an SV
da8c5729 1217based on both its magic type and its magic virtual table:
f6ee7b17
FR
1218
1219 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
d1b91892 1220
f6ee7b17
FR
1221Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1222SVt_PVMG, Perl may core dump.
d1b91892 1223
08105a92 1224 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1225
1226This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745 1227field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1228the mg_type field is changed to be the lowercase letter.
a0d0e21e 1229
04343c6d
GS
1230=head2 Understanding the Magic of Tied Hashes and Arrays
1231
14befaf4
DM
1232Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1233magic type.
9edb2b46
GS
1234
1235WARNING: As of the 5.004 release, proper usage of the array and hash
1236access functions requires understanding a few caveats. Some
1237of these caveats are actually considered bugs in the API, to be fixed
10e2eb10 1238in later releases, and are bracketed with [MAYCHANGE] below. If
9edb2b46
GS
1239you find yourself actually applying such information in this section, be
1240aware that the behavior may change in the future, umm, without warning.
04343c6d 1241
1526ead6 1242The perl tie function associates a variable with an object that implements
9a68f1db 1243the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1244tie function from an XSUB, you must mimic this behaviour. The code below
1245carries out the necessary steps - firstly it creates a new hash, and then
1246creates a second hash which it blesses into the class which will implement
10e2eb10 1247the tie methods. Lastly it ties the two hashes together, and returns a
1526ead6
AB
1248reference to the new tied hash. Note that the code below does NOT call the
1249TIEHASH method in the MyTie class -
1250see L<Calling Perl Routines from within C Programs> for details on how
1251to do this.
1252
1253 SV*
1254 mytie()
1255 PREINIT:
1256 HV *hash;
1257 HV *stash;
1258 SV *tie;
1259 CODE:
1260 hash = newHV();
1261 tie = newRV_noinc((SV*)newHV());
da51bb9b 1262 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1263 sv_bless(tie, stash);
899e16d0 1264 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1265 RETVAL = newRV_noinc(hash);
1266 OUTPUT:
1267 RETVAL
1268
04343c6d
GS
1269The C<av_store> function, when given a tied array argument, merely
1270copies the magic of the array onto the value to be "stored", using
1271C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1272actually need to be stored in the array. [MAYCHANGE] After a call to
1273C<av_store> on a tied array, the caller will usually need to call
1274C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1275TIEARRAY object. If C<av_store> did return NULL, a call to
1276C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1277leak. [/MAYCHANGE]
04343c6d
GS
1278
1279The previous paragraph is applicable verbatim to tied hash access using the
1280C<hv_store> and C<hv_store_ent> functions as well.
1281
1282C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1283C<hv_fetch_ent> actually return an undefined mortal value whose magic
1284has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1285need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1286need to call C<mg_get()> on the returned value in order to actually invoke
1287the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1288you may also call C<mg_set()> on the return value after possibly assigning
1289a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1290method on the TIE object. [/MAYCHANGE]
04343c6d 1291
9edb2b46 1292[MAYCHANGE]
04343c6d
GS
1293In other words, the array or hash fetch/store functions don't really
1294fetch and store actual values in the case of tied arrays and hashes. They
1295merely call C<mg_copy> to attach magic to the values that were meant to be
1296"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1297do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1298the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1299and hashes.
1300
1301Currently (as of perl version 5.004), use of the hash and array access
1302functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1303"normal" hashes and arrays, or on their tied variants. The API may be
1304changed to provide more transparent access to both tied and normal data
1305types in future versions.
1306[/MAYCHANGE]
04343c6d
GS
1307
1308You would do well to understand that the TIEARRAY and TIEHASH interfaces
1309are mere sugar to invoke some perl method calls while using the uniform hash
1310and array syntax. The use of this sugar imposes some overhead (typically
1311about two to four extra opcodes per FETCH/STORE operation, in addition to
1312the creation of all the mortal variables required to invoke the methods).
1313This overhead will be comparatively small if the TIE methods are themselves
1314substantial, but if they are only a few statements long, the overhead
1315will not be insignificant.
1316
d1c897a1
IZ
1317=head2 Localizing changes
1318
1319Perl has a very handy construction
1320
1321 {
1322 local $var = 2;
1323 ...
1324 }
1325
1326This construction is I<approximately> equivalent to
1327
1328 {
1329 my $oldvar = $var;
1330 $var = 2;
1331 ...
1332 $var = $oldvar;
1333 }
1334
1335The biggest difference is that the first construction would
1336reinstate the initial value of $var, irrespective of how control exits
10e2eb10 1337the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1338more efficient as well.
1339
1340There is a way to achieve a similar task from C via Perl API: create a
1341I<pseudo-block>, and arrange for some changes to be automatically
1342undone at the end of it, either explicit, or via a non-local exit (via
10e2eb10 1343die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1344C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1345Such a construct may be created specially for some important localized
1346task, or an existing one (like boundaries of enclosing Perl
1347subroutine/block, or an existing pair for freeing TMPs) may be
10e2eb10
FC
1348used. (In the second case the overhead of additional localization must
1349be almost negligible.) Note that any XSUB is automatically enclosed in
b687b08b 1350an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1351
1352Inside such a I<pseudo-block> the following service is available:
1353
13a2d996 1354=over 4
d1c897a1
IZ
1355
1356=item C<SAVEINT(int i)>
1357
1358=item C<SAVEIV(IV i)>
1359
1360=item C<SAVEI32(I32 i)>
1361
1362=item C<SAVELONG(long i)>
1363
1364These macros arrange things to restore the value of integer variable
1365C<i> at the end of enclosing I<pseudo-block>.
1366
1367=item C<SAVESPTR(s)>
1368
1369=item C<SAVEPPTR(p)>
1370
1371These macros arrange things to restore the value of pointers C<s> and
10e2eb10 1372C<p>. C<s> must be a pointer of a type which survives conversion to
d1c897a1
IZ
1373C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1374and back.
1375
1376=item C<SAVEFREESV(SV *sv)>
1377
1378The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1379I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1380mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1381extends the lifetime of C<sv> until the beginning of the next statement,
1382C<SAVEFREESV> extends it until the end of the enclosing scope. These
1383lifetimes can be wildly different.
1384
1385Also compare C<SAVEMORTALIZESV>.
1386
1387=item C<SAVEMORTALIZESV(SV *sv)>
1388
1389Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1390scope instead of decrementing its reference count. This usually has the
1391effect of keeping C<sv> alive until the statement that called the currently
1392live scope has finished executing.
d1c897a1
IZ
1393
1394=item C<SAVEFREEOP(OP *op)>
1395
1396The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1397
1398=item C<SAVEFREEPV(p)>
1399
1400The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1401end of I<pseudo-block>.
1402
1403=item C<SAVECLEARSV(SV *sv)>
1404
1405Clears a slot in the current scratchpad which corresponds to C<sv> at
1406the end of I<pseudo-block>.
1407
1408=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1409
10e2eb10 1410The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
d1c897a1
IZ
1411string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1412short-lived storage, the corresponding string may be reallocated like
1413this:
1414
9cde0e7f 1415 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1416
c76ac1ee 1417=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1418
1419At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1420only argument C<p>.
1421
1422=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1423
1424At the end of I<pseudo-block> the function C<f> is called with the
1425implicit context argument (if any), and C<p>.
d1c897a1
IZ
1426
1427=item C<SAVESTACK_POS()>
1428
1429The current offset on the Perl internal stack (cf. C<SP>) is restored
1430at the end of I<pseudo-block>.
1431
1432=back
1433
1434The following API list contains functions, thus one needs to
1435provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1436or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1437function takes C<int *>.
1438
13a2d996 1439=over 4
d1c897a1
IZ
1440
1441=item C<SV* save_scalar(GV *gv)>
1442
1443Equivalent to Perl code C<local $gv>.
1444
1445=item C<AV* save_ary(GV *gv)>
1446
1447=item C<HV* save_hash(GV *gv)>
1448
1449Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1450
1451=item C<void save_item(SV *item)>
1452
1453Duplicates the current value of C<SV>, on the exit from the current
1454C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
10e2eb10 1455using the stored value. It doesn't handle magic. Use C<save_scalar> if
038fcae3 1456magic is affected.
d1c897a1
IZ
1457
1458=item C<void save_list(SV **sarg, I32 maxsarg)>
1459
1460A variant of C<save_item> which takes multiple arguments via an array
1461C<sarg> of C<SV*> of length C<maxsarg>.
1462
1463=item C<SV* save_svref(SV **sptr)>
1464
d1be9408 1465Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1466
1467=item C<void save_aptr(AV **aptr)>
1468
1469=item C<void save_hptr(HV **hptr)>
1470
1471Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1472
1473=back
1474
1475The C<Alias> module implements localization of the basic types within the
1476I<caller's scope>. People who are interested in how to localize things in
1477the containing scope should take a look there too.
1478
0a753a76 1479=head1 Subroutines
a0d0e21e 1480
68dc0745 1481=head2 XSUBs and the Argument Stack
5f05dabc 1482
1483The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1484An XSUB routine will have a stack that contains the arguments from the Perl
1485program, and a way to map from the Perl data structures to a C equivalent.
1486
1487The stack arguments are accessible through the C<ST(n)> macro, which returns
1488the C<n>'th stack argument. Argument 0 is the first argument passed in the
1489Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1490an C<SV*> is used.
1491
1492Most of the time, output from the C routine can be handled through use of
1493the RETVAL and OUTPUT directives. However, there are some cases where the
1494argument stack is not already long enough to handle all the return values.
1495An example is the POSIX tzname() call, which takes no arguments, but returns
1496two, the local time zone's standard and summer time abbreviations.
1497
1498To handle this situation, the PPCODE directive is used and the stack is
1499extended using the macro:
1500
924508f0 1501 EXTEND(SP, num);
5f05dabc 1502
924508f0
GS
1503where C<SP> is the macro that represents the local copy of the stack pointer,
1504and C<num> is the number of elements the stack should be extended by.
5f05dabc 1505
00aadd71 1506Now that there is room on the stack, values can be pushed on it using C<PUSHs>
10e2eb10 1507macro. The pushed values will often need to be "mortal" (See
d82b684c 1508L</Reference Counts and Mortality>):
5f05dabc 1509
00aadd71 1510 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1511 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1512 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1513 PUSHs(sv_2mortal(newSVpv("Some String",0)))
a9b0660e
KW
1514 /* Although the last example is better written as the more
1515 * efficient: */
a3179684 1516 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
5f05dabc 1517
1518And now the Perl program calling C<tzname>, the two values will be assigned
1519as in:
1520
1521 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1522
1523An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1524to use the macro:
5f05dabc 1525
5f05dabc 1526 XPUSHs(SV*)
1527
da8c5729 1528This macro automatically adjusts the stack for you, if needed. Thus, you
5f05dabc 1529do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1530
1531Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1532C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1533For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1534C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc 1535
1536For more information, consult L<perlxs> and L<perlxstut>.
1537
5b36e945
FC
1538=head2 Autoloading with XSUBs
1539
1540If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1541fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1542of the XSUB's package.
1543
1544But it also puts the same information in certain fields of the XSUB itself:
1545
1546 HV *stash = CvSTASH(cv);
1547 const char *subname = SvPVX(cv);
1548 STRLEN name_length = SvCUR(cv); /* in bytes */
1549 U32 is_utf8 = SvUTF8(cv);
f703fc96 1550
5b36e945 1551C<SvPVX(cv)> contains just the sub name itself, not including the package.
d8893903
FC
1552For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1553C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
5b36e945
FC
1554
1555B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1556XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1557XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1558to support 5.8-5.14, use the XSUB's fields.
1559
5f05dabc 1560=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1561
1562There are four routines that can be used to call a Perl subroutine from
1563within a C program. These four are:
1564
954c1994
GS
1565 I32 call_sv(SV*, I32);
1566 I32 call_pv(const char*, I32);
1567 I32 call_method(const char*, I32);
5aaab254 1568 I32 call_argv(const char*, I32, char**);
a0d0e21e 1569
954c1994 1570The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1571contains either the name of the Perl subroutine to be called, or a
1572reference to the subroutine. The second argument consists of flags
1573that control the context in which the subroutine is called, whether
1574or not the subroutine is being passed arguments, how errors should be
1575trapped, and how to treat return values.
a0d0e21e
LW
1576
1577All four routines return the number of arguments that the subroutine returned
1578on the Perl stack.
1579
9a68f1db 1580These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1581but those names are now deprecated; macros of the same name are provided for
1582compatibility.
1583
1584When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1585must manipulate the Perl stack. These include the following macros and
1586functions:
a0d0e21e
LW
1587
1588 dSP
924508f0 1589 SP
a0d0e21e
LW
1590 PUSHMARK()
1591 PUTBACK
1592 SPAGAIN
1593 ENTER
1594 SAVETMPS
1595 FREETMPS
1596 LEAVE
1597 XPUSH*()
cb1a09d0 1598 POP*()
a0d0e21e 1599
5f05dabc 1600For a detailed description of calling conventions from C to Perl,
1601consult L<perlcall>.
a0d0e21e 1602
5f05dabc 1603=head2 Memory Allocation
a0d0e21e 1604
06f6df17
RGS
1605=head3 Allocation
1606
86058a2d
GS
1607All memory meant to be used with the Perl API functions should be manipulated
1608using the macros described in this section. The macros provide the necessary
1609transparency between differences in the actual malloc implementation that is
1610used within perl.
1611
1612It is suggested that you enable the version of malloc that is distributed
5f05dabc 1613with Perl. It keeps pools of various sizes of unallocated memory in
07fa94a1
JO
1614order to satisfy allocation requests more quickly. However, on some
1615platforms, it may cause spurious malloc or free errors.
d1b91892 1616
06f6df17
RGS
1617The following three macros are used to initially allocate memory :
1618
9f653bb5
SH
1619 Newx(pointer, number, type);
1620 Newxc(pointer, number, type, cast);
1621 Newxz(pointer, number, type);
d1b91892 1622
9f653bb5 1623The first argument C<pointer> should be the name of a variable that will
5f05dabc 1624point to the newly allocated memory.
d1b91892 1625
9f653bb5 1626The second and third arguments C<number> and C<type> specify how many of
d1b91892 1627the specified type of data structure should be allocated. The argument
9f653bb5 1628C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
d1b91892
AD
1629should be used if the C<pointer> argument is different from the C<type>
1630argument.
1631
9f653bb5 1632Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
d1b91892
AD
1633to zero out all the newly allocated memory.
1634
06f6df17
RGS
1635=head3 Reallocation
1636
d1b91892
AD
1637 Renew(pointer, number, type);
1638 Renewc(pointer, number, type, cast);
1639 Safefree(pointer)
1640
1641These three macros are used to change a memory buffer size or to free a
1642piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1643match those of C<New> and C<Newc> with the exception of not needing the
1644"magic cookie" argument.
1645
06f6df17
RGS
1646=head3 Moving
1647
d1b91892
AD
1648 Move(source, dest, number, type);
1649 Copy(source, dest, number, type);
1650 Zero(dest, number, type);
1651
1652These three macros are used to move, copy, or zero out previously allocated
1653memory. The C<source> and C<dest> arguments point to the source and
1654destination starting points. Perl will move, copy, or zero out C<number>
1655instances of the size of the C<type> data structure (using the C<sizeof>
1656function).
a0d0e21e 1657
5f05dabc 1658=head2 PerlIO
ce3d39e2 1659
da8c5729 1660The most recent development releases of Perl have been experimenting with
5f05dabc 1661removing Perl's dependency on the "normal" standard I/O suite and allowing
1662other stdio implementations to be used. This involves creating a new
1663abstraction layer that then calls whichever implementation of stdio Perl
68dc0745 1664was compiled with. All XSUBs should now use the functions in the PerlIO
5f05dabc 1665abstraction layer and not make any assumptions about what kind of stdio
1666is being used.
1667
1668For a complete description of the PerlIO abstraction, consult L<perlapio>.
1669
8ebc5c01 1670=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1671
1672A lot of opcodes (this is an elementary operation in the internal perl
10e2eb10
FC
1673stack machine) put an SV* on the stack. However, as an optimization
1674the corresponding SV is (usually) not recreated each time. The opcodes
ce3d39e2
IZ
1675reuse specially assigned SVs (I<target>s) which are (as a corollary)
1676not constantly freed/created.
1677
0a753a76 1678Each of the targets is created only once (but see
ce3d39e2
IZ
1679L<Scratchpads and recursion> below), and when an opcode needs to put
1680an integer, a double, or a string on stack, it just sets the
1681corresponding parts of its I<target> and puts the I<target> on stack.
1682
1683The macro to put this target on stack is C<PUSHTARG>, and it is
1684directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1685others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1686
1bd1c0d5 1687Because the target is reused, you must be careful when pushing multiple
10e2eb10 1688values on the stack. The following code will not do what you think:
1bd1c0d5
SC
1689
1690 XPUSHi(10);
1691 XPUSHi(20);
1692
1693This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1694the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1695At the end of the operation, the stack does not contain the values 10
1696and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1697to 20.
1bd1c0d5 1698
d82b684c
SH
1699If you need to push multiple different values then you should either use
1700the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1701none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1702SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1703will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1704this a little easier to achieve by creating a new mortal for you (via
1705C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1706in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1707Thus, instead of writing this to "fix" the example above:
1708
1709 XPUSHs(sv_2mortal(newSViv(10)))
1710 XPUSHs(sv_2mortal(newSViv(20)))
1711
1712you can simply write:
1713
1714 mXPUSHi(10)
1715 mXPUSHi(20)
1716
1717On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1718need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1719macros can make use of the local variable C<TARG>. See also C<dTARGET>
1720and C<dXSTARG>.
1bd1c0d5 1721
8ebc5c01 1722=head2 Scratchpads
ce3d39e2 1723
54310121 1724The question remains on when the SVs which are I<target>s for opcodes
10e2eb10 1725are created. The answer is that they are created when the current
ac036724 1726unit--a subroutine or a file (for opcodes for statements outside of
10e2eb10 1727subroutines)--is compiled. During this time a special anonymous Perl
ac036724 1728array is created, which is called a scratchpad for the current unit.
ce3d39e2 1729
54310121 1730A scratchpad keeps SVs which are lexicals for the current unit and are
d777b41a
FC
1731targets for opcodes. A previous version of this document
1732stated that one can deduce that an SV lives on a scratchpad
ce3d39e2 1733by looking on its flags: lexicals have C<SVs_PADMY> set, and
d777b41a
FC
1734I<target>s have C<SVs_PADTMP> set. But this have never been fully true.
1735C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
1736While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
1737that have never resided in a pad, but nonetheless act like I<target>s.
ce3d39e2 1738
10e2eb10 1739The correspondence between OPs and I<target>s is not 1-to-1. Different
54310121 1740OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1741would not conflict with the expected life of the temporary.
1742
2ae324a7 1743=head2 Scratchpads and recursion
ce3d39e2
IZ
1744
1745In fact it is not 100% true that a compiled unit contains a pointer to
10e2eb10
FC
1746the scratchpad AV. In fact it contains a pointer to an AV of
1747(initially) one element, and this element is the scratchpad AV. Why do
ce3d39e2
IZ
1748we need an extra level of indirection?
1749
10e2eb10 1750The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2 1751these can create several execution pointers going into the same
10e2eb10 1752subroutine. For the subroutine-child not write over the temporaries
ce3d39e2
IZ
1753for the subroutine-parent (lifespan of which covers the call to the
1754child), the parent and the child should have different
10e2eb10 1755scratchpads. (I<And> the lexicals should be separate anyway!)
ce3d39e2 1756
5f05dabc 1757So each subroutine is born with an array of scratchpads (of length 1).
1758On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1759depth of the recursion is not more than the length of this array, and
1760if it is, new scratchpad is created and pushed into the array.
1761
1762The I<target>s on this scratchpad are C<undef>s, but they are already
1763marked with correct flags.
1764
0a753a76 1765=head1 Compiled code
1766
1767=head2 Code tree
1768
1769Here we describe the internal form your code is converted to by
10e2eb10 1770Perl. Start with a simple example:
0a753a76 1771
1772 $a = $b + $c;
1773
1774This is converted to a tree similar to this one:
1775
1776 assign-to
1777 / \
1778 + $a
1779 / \
1780 $b $c
1781
7b8d334a 1782(but slightly more complicated). This tree reflects the way Perl
0a753a76 1783parsed your code, but has nothing to do with the execution order.
1784There is an additional "thread" going through the nodes of the tree
1785which shows the order of execution of the nodes. In our simplified
1786example above it looks like:
1787
1788 $b ---> $c ---> + ---> $a ---> assign-to
1789
1790But with the actual compile tree for C<$a = $b + $c> it is different:
1791some nodes I<optimized away>. As a corollary, though the actual tree
1792contains more nodes than our simplified example, the execution order
1793is the same as in our example.
1794
1795=head2 Examining the tree
1796
06f6df17
RGS
1797If you have your perl compiled for debugging (usually done with
1798C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76 1799compiled tree by specifying C<-Dx> on the Perl command line. The
1800output takes several lines per node, and for C<$b+$c> it looks like
1801this:
1802
1803 5 TYPE = add ===> 6
1804 TARG = 1
1805 FLAGS = (SCALAR,KIDS)
1806 {
1807 TYPE = null ===> (4)
1808 (was rv2sv)
1809 FLAGS = (SCALAR,KIDS)
1810 {
1811 3 TYPE = gvsv ===> 4
1812 FLAGS = (SCALAR)
1813 GV = main::b
1814 }
1815 }
1816 {
1817 TYPE = null ===> (5)
1818 (was rv2sv)
1819 FLAGS = (SCALAR,KIDS)
1820 {
1821 4 TYPE = gvsv ===> 5
1822 FLAGS = (SCALAR)
1823 GV = main::c
1824 }
1825 }
1826
1827This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1828not optimized away (one per number in the left column). The immediate
1829children of the given node correspond to C<{}> pairs on the same level
1830of indentation, thus this listing corresponds to the tree:
1831
1832 add
1833 / \
1834 null null
1835 | |
1836 gvsv gvsv
1837
1838The execution order is indicated by C<===E<gt>> marks, thus it is C<3
18394 5 6> (node C<6> is not included into above listing), i.e.,
1840C<gvsv gvsv add whatever>.
1841
9afa14e3 1842Each of these nodes represents an op, a fundamental operation inside the
10e2eb10 1843Perl core. The code which implements each operation can be found in the
9afa14e3 1844F<pp*.c> files; the function which implements the op with type C<gvsv>
10e2eb10 1845is C<pp_gvsv>, and so on. As the tree above shows, different ops have
9afa14e3 1846different numbers of children: C<add> is a binary operator, as one would
10e2eb10 1847expect, and so has two children. To accommodate the various different
9afa14e3
SC
1848numbers of children, there are various types of op data structure, and
1849they link together in different ways.
1850
10e2eb10 1851The simplest type of op structure is C<OP>: this has no children. Unary
9afa14e3 1852operators, C<UNOP>s, have one child, and this is pointed to by the
10e2eb10
FC
1853C<op_first> field. Binary operators (C<BINOP>s) have not only an
1854C<op_first> field but also an C<op_last> field. The most complex type of
1855op is a C<LISTOP>, which has any number of children. In this case, the
9afa14e3 1856first child is pointed to by C<op_first> and the last child by
10e2eb10 1857C<op_last>. The children in between can be found by iteratively
9afa14e3
SC
1858following the C<op_sibling> pointer from the first child to the last.
1859
1860There are also two other op types: a C<PMOP> holds a regular expression,
10e2eb10
FC
1861and has no children, and a C<LOOP> may or may not have children. If the
1862C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
9afa14e3
SC
1863complicate matters, if a C<UNOP> is actually a C<null> op after
1864optimization (see L</Compile pass 2: context propagation>) it will still
1865have children in accordance with its former type.
1866
06f6df17
RGS
1867Another way to examine the tree is to use a compiler back-end module, such
1868as L<B::Concise>.
1869
0a753a76 1870=head2 Compile pass 1: check routines
1871
8870b5c7 1872The tree is created by the compiler while I<yacc> code feeds it
10e2eb10 1873the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76 1874the first pass of perl compilation.
1875
1876What makes this pass interesting for perl developers is that some
1877optimization may be performed on this pass. This is optimization by
8870b5c7 1878so-called "check routines". The correspondence between node names
0a753a76 1879and corresponding check routines is described in F<opcode.pl> (do not
1880forget to run C<make regen_headers> if you modify this file).
1881
1882A check routine is called when the node is fully constructed except
7b8d334a 1883for the execution-order thread. Since at this time there are no
0a753a76 1884back-links to the currently constructed node, one can do most any
1885operation to the top-level node, including freeing it and/or creating
1886new nodes above/below it.
1887
1888The check routine returns the node which should be inserted into the
1889tree (if the top-level node was not modified, check routine returns
1890its argument).
1891
10e2eb10 1892By convention, check routines have names C<ck_*>. They are usually
0a753a76 1893called from C<new*OP> subroutines (or C<convert>) (which in turn are
1894called from F<perly.y>).
1895
1896=head2 Compile pass 1a: constant folding
1897
1898Immediately after the check routine is called the returned node is
1899checked for being compile-time executable. If it is (the value is
1900judged to be constant) it is immediately executed, and a I<constant>
1901node with the "return value" of the corresponding subtree is
1902substituted instead. The subtree is deleted.
1903
1904If constant folding was not performed, the execution-order thread is
1905created.
1906
1907=head2 Compile pass 2: context propagation
1908
1909When a context for a part of compile tree is known, it is propagated
a3cb178b 1910down through the tree. At this time the context can have 5 values
0a753a76 1911(instead of 2 for runtime context): void, boolean, scalar, list, and
1912lvalue. In contrast with the pass 1 this pass is processed from top
1913to bottom: a node's context determines the context for its children.
1914
1915Additional context-dependent optimizations are performed at this time.
1916Since at this moment the compile tree contains back-references (via
1917"thread" pointers), nodes cannot be free()d now. To allow
1918optimized-away nodes at this stage, such nodes are null()ified instead
1919of free()ing (i.e. their type is changed to OP_NULL).
1920
1921=head2 Compile pass 3: peephole optimization
1922
1923After the compile tree for a subroutine (or for an C<eval> or a file)
10e2eb10 1924is created, an additional pass over the code is performed. This pass
0a753a76 1925is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
1926additional complications for conditionals). Optimizations performed
1927at this stage are subject to the same restrictions as in the pass 2.
1928
1929Peephole optimizations are done by calling the function pointed to
1930by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1931calls the function pointed to by the global variable C<PL_rpeepp>.
1932By default, that performs some basic op fixups and optimisations along
1933the execution-order op chain, and recursively calls C<PL_rpeepp> for
1934each side chain of ops (resulting from conditionals). Extensions may
1935provide additional optimisations or fixups, hooking into either the
1936per-subroutine or recursive stage, like this:
1937
1938 static peep_t prev_peepp;
1939 static void my_peep(pTHX_ OP *o)
1940 {
1941 /* custom per-subroutine optimisation goes here */
f0358462 1942 prev_peepp(aTHX_ o);
9ea12537
Z
1943 /* custom per-subroutine optimisation may also go here */
1944 }
1945 BOOT:
1946 prev_peepp = PL_peepp;
1947 PL_peepp = my_peep;
1948
1949 static peep_t prev_rpeepp;
1950 static void my_rpeep(pTHX_ OP *o)
1951 {
1952 OP *orig_o = o;
1953 for(; o; o = o->op_next) {
1954 /* custom per-op optimisation goes here */
1955 }
f0358462 1956 prev_rpeepp(aTHX_ orig_o);
9ea12537
Z
1957 }
1958 BOOT:
1959 prev_rpeepp = PL_rpeepp;
1960 PL_rpeepp = my_rpeep;
0a753a76 1961
1ba7f851
PJ
1962=head2 Pluggable runops
1963
1964The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
1965functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1966with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1967control over the execution of the compile tree it is possible to provide
1968your own runops function.
1ba7f851
PJ
1969
1970It's probably best to copy one of the existing runops functions and
1971change it to suit your needs. Then, in the BOOT section of your XS
1972file, add the line:
1973
1974 PL_runops = my_runops;
1975
1976This function should be as efficient as possible to keep your programs
1977running as fast as possible.
1978
fd85fad2
BM
1979=head2 Compile-time scope hooks
1980
1981As of perl 5.14 it is possible to hook into the compile-time lexical
10e2eb10 1982scope mechanism using C<Perl_blockhook_register>. This is used like
fd85fad2
BM
1983this:
1984
1985 STATIC void my_start_hook(pTHX_ int full);
1986 STATIC BHK my_hooks;
1987
1988 BOOT:
a88d97bf 1989 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
1990 Perl_blockhook_register(aTHX_ &my_hooks);
1991
1992This will arrange to have C<my_start_hook> called at the start of
10e2eb10 1993compiling every lexical scope. The available hooks are:
fd85fad2
BM
1994
1995=over 4
1996
a88d97bf 1997=item C<void bhk_start(pTHX_ int full)>
fd85fad2 1998
10e2eb10 1999This is called just after starting a new lexical scope. Note that Perl
fd85fad2
BM
2000code like
2001
2002 if ($x) { ... }
2003
2004creates two scopes: the first starts at the C<(> and has C<full == 1>,
10e2eb10
FC
2005the second starts at the C<{> and has C<full == 0>. Both end at the
2006C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
fd85fad2
BM
2007pushed onto the save stack by this hook will be popped just before the
2008scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2009
a88d97bf 2010=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
2011
2012This is called at the end of a lexical scope, just before unwinding the
10e2eb10 2013stack. I<o> is the root of the optree representing the scope; it is a
fd85fad2
BM
2014double pointer so you can replace the OP if you need to.
2015
a88d97bf 2016=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
2017
2018This is called at the end of a lexical scope, just after unwinding the
10e2eb10 2019stack. I<o> is as above. Note that it is possible for calls to C<pre_>
fd85fad2
BM
2020and C<post_end> to nest, if there is something on the save stack that
2021calls string eval.
2022
a88d97bf 2023=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
2024
2025This is called just before starting to compile an C<eval STRING>, C<do
10e2eb10 2026FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
fd85fad2
BM
2027OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2028C<OP_DOFILE> or C<OP_REQUIRE>.
2029
2030=back
2031
2032Once you have your hook functions, you need a C<BHK> structure to put
10e2eb10
FC
2033them in. It's best to allocate it statically, since there is no way to
2034free it once it's registered. The function pointers should be inserted
fd85fad2 2035into this structure using the C<BhkENTRY_set> macro, which will also set
10e2eb10 2036flags indicating which entries are valid. If you do need to allocate
fd85fad2
BM
2037your C<BHK> dynamically for some reason, be sure to zero it before you
2038start.
2039
2040Once registered, there is no mechanism to switch these hooks off, so if
10e2eb10 2041that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
2042is probably the best way, so the effect is lexically scoped; however it
2043is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
10e2eb10 2044temporarily switch entries on and off. You should also be aware that
a3e07c87
BM
2045generally speaking at least one scope will have opened before your
2046extension is loaded, so you will see some C<pre/post_end> pairs that
2047didn't have a matching C<start>.
fd85fad2 2048
9afa14e3
SC
2049=head1 Examining internal data structures with the C<dump> functions
2050
2051To aid debugging, the source file F<dump.c> contains a number of
2052functions which produce formatted output of internal data structures.
2053
2054The most commonly used of these functions is C<Perl_sv_dump>; it's used
10e2eb10 2055for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
9afa14e3 2056C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 2057module should already be familiar with its format.
9afa14e3
SC
2058
2059C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 2060derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
2061C<Perl_dump_eval> will dump the main root of the code being evaluated,
2062exactly like C<-Dx>.
2063
2064Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2065op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2066subroutines in a package like so: (Thankfully, these are all xsubs, so
2067there is no op tree)
2068
2069 (gdb) print Perl_dump_packsubs(PL_defstash)
2070
2071 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2072
2073 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2074
2075 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2076
2077 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2078
2079 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2080
2081and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2082the op tree of the main root.
2083
954c1994 2084=head1 How multiple interpreters and concurrency are supported
ee072b34 2085
ee072b34
GS
2086=head2 Background and PERL_IMPLICIT_CONTEXT
2087
2088The Perl interpreter can be regarded as a closed box: it has an API
2089for feeding it code or otherwise making it do things, but it also has
2090functions for its own use. This smells a lot like an object, and
2091there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2092interpreters, with one interpreter represented either as a C structure,
2093or inside a thread-specific structure. These structures contain all
2094the context, the state of that interpreter.
2095
10e2eb10 2096One macro controls the major Perl build flavor: MULTIPLICITY. The
7b52221d 2097MULTIPLICITY build has a C structure that packages all the interpreter
10e2eb10 2098state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
7b52221d 2099normally defined, and enables the support for passing in a "hidden" first
10e2eb10 2100argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2101multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2102to the macro USE_ITHREADS.)
54aff467 2103
27da23d5
JH
2104Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2105PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2106former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2107internal variables of Perl to be wrapped inside a single global struct,
2108struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2109the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2110one step further, there is still a single struct (allocated in main()
2111either from heap or from stack) but there are no global data symbols
3bf17896 2112pointing to it. In either case the global struct should be initialized
27da23d5
JH
2113as the very first thing in main() using Perl_init_global_struct() and
2114correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2115please see F<miniperlmain.c> for usage details. You may also need
2116to use C<dVAR> in your coding to "declare the global variables"
2117when you are using them. dTHX does this for you automatically.
2118
bc028b6b
JH
2119To see whether you have non-const data you can use a BSD-compatible C<nm>:
2120
2121 nm libperl.a | grep -v ' [TURtr] '
2122
2123If this displays any C<D> or C<d> symbols, you have non-const data.
2124
27da23d5
JH
2125For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2126doesn't actually hide all symbols inside a big global struct: some
2127PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2128then hides everything (see how the PERLIO_FUNCS_DECL is used).
2129
54aff467 2130All this obviously requires a way for the Perl internal functions to be
acfe0abc 2131either subroutines taking some kind of structure as the first
ee072b34 2132argument, or subroutines taking nothing as the first argument. To
acfe0abc 2133enable these two very different ways of building the interpreter,
ee072b34
GS
2134the Perl source (as it does in so many other situations) makes heavy
2135use of macros and subroutine naming conventions.
2136
54aff467 2137First problem: deciding which functions will be public API functions and
00aadd71 2138which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2139(think "S" for "secret" or "static"). All other functions begin with
2140"Perl_", but just because a function begins with "Perl_" does not mean it is
10e2eb10
FC
2141part of the API. (See L</Internal
2142Functions>.) The easiest way to be B<sure> a
00aadd71
NIS
2143function is part of the API is to find its entry in L<perlapi>.
2144If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2145think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2146L<perlbug> explaining why you think it should be.
ee072b34
GS
2147
2148Second problem: there must be a syntax so that the same subroutine
2149declarations and calls can pass a structure as their first argument,
2150or pass nothing. To solve this, the subroutines are named and
2151declared in a particular way. Here's a typical start of a static
2152function used within the Perl guts:
2153
2154 STATIC void
2155 S_incline(pTHX_ char *s)
2156
acfe0abc 2157STATIC becomes "static" in C, and may be #define'd to nothing in some
da8c5729 2158configurations in the future.
ee072b34 2159
651a3225
GS
2160A public function (i.e. part of the internal API, but not necessarily
2161sanctioned for use in extensions) begins like this:
ee072b34
GS
2162
2163 void
2307c6d0 2164 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2165
0147cd53 2166C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2167details of the interpreter's context. THX stands for "thread", "this",
2168or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2169The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2170or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2171their variants.
ee072b34 2172
a7486cbb
JH
2173When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2174first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2175in the pTHX_ macro indicates that the macro expansion needs a comma
2176after the context argument because other arguments follow it. If
2177PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2178subroutine is not prototyped to take the extra argument. The form of the
2179macro without the trailing underscore is used when there are no additional
ee072b34
GS
2180explicit arguments.
2181
54aff467 2182When a core function calls another, it must pass the context. This
2307c6d0 2183is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2184something like this:
2185
2307c6d0
SB
2186 #ifdef PERL_IMPLICIT_CONTEXT
2187 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2188 /* can't do this for vararg functions, see below */
2307c6d0
SB
2189 #else
2190 #define sv_setiv Perl_sv_setiv
2191 #endif
ee072b34
GS
2192
2193This works well, and means that XS authors can gleefully write:
2194
2307c6d0 2195 sv_setiv(foo, bar);
ee072b34
GS
2196
2197and still have it work under all the modes Perl could have been
2198compiled with.
2199
ee072b34
GS
2200This doesn't work so cleanly for varargs functions, though, as macros
2201imply that the number of arguments is known in advance. Instead we
2202either need to spell them out fully, passing C<aTHX_> as the first
2203argument (the Perl core tends to do this with functions like
2204Perl_warner), or use a context-free version.
2205
2206The context-free version of Perl_warner is called
2207Perl_warner_nocontext, and does not take the extra argument. Instead
2208it does dTHX; to get the context from thread-local storage. We
2209C<#define warner Perl_warner_nocontext> so that extensions get source
2210compatibility at the expense of performance. (Passing an arg is
2211cheaper than grabbing it from thread-local storage.)
2212
acfe0abc 2213You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2214Those are strictly for use within the core. Extensions and embedders
2215need only be aware of [pad]THX.
2216
a7486cbb
JH
2217=head2 So what happened to dTHR?
2218
2219C<dTHR> was introduced in perl 5.005 to support the older thread model.
2220The older thread model now uses the C<THX> mechanism to pass context
2221pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2222later still have it for backward source compatibility, but it is defined
2223to be a no-op.
2224
ee072b34
GS
2225=head2 How do I use all this in extensions?
2226
2227When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2228any functions in the Perl API will need to pass the initial context
2229argument somehow. The kicker is that you will need to write it in
2230such a way that the extension still compiles when Perl hasn't been
2231built with PERL_IMPLICIT_CONTEXT enabled.
2232
2233There are three ways to do this. First, the easy but inefficient way,
2234which is also the default, in order to maintain source compatibility
0147cd53 2235with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2236and aTHX_ macros to call a function that will return the context.
2237Thus, something like:
2238
2307c6d0 2239 sv_setiv(sv, num);
ee072b34 2240
4375e838 2241in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2242in effect:
ee072b34 2243
2307c6d0 2244 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2245
54aff467 2246or to this otherwise:
ee072b34 2247
2307c6d0 2248 Perl_sv_setiv(sv, num);
ee072b34 2249
da8c5729 2250You don't have to do anything new in your extension to get this; since
2fa86c13 2251the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2252work.
2253
2254The second, more efficient way is to use the following template for
2255your Foo.xs:
2256
c52f9dcd
JH
2257 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2258 #include "EXTERN.h"
2259 #include "perl.h"
2260 #include "XSUB.h"
ee072b34 2261
fd061412 2262 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2263
fd061412 2264 STATIC void
c52f9dcd
JH
2265 my_private_function(int arg1, int arg2)
2266 {
2267 dTHX; /* fetch context */
2268 ... call many Perl API functions ...
2269 }
ee072b34
GS
2270
2271 [... etc ...]
2272
c52f9dcd 2273 MODULE = Foo PACKAGE = Foo
ee072b34 2274
c52f9dcd 2275 /* typical XSUB */
ee072b34 2276
c52f9dcd
JH
2277 void
2278 my_xsub(arg)
2279 int arg
2280 CODE:
2281 my_private_function(arg, 10);
ee072b34
GS
2282
2283Note that the only two changes from the normal way of writing an
2284extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2285including the Perl headers, followed by a C<dTHX;> declaration at
2286the start of every function that will call the Perl API. (You'll
2287know which functions need this, because the C compiler will complain
2288that there's an undeclared identifier in those functions.) No changes
2289are needed for the XSUBs themselves, because the XS() macro is
2290correctly defined to pass in the implicit context if needed.
2291
2292The third, even more efficient way is to ape how it is done within
2293the Perl guts:
2294
2295
c52f9dcd
JH
2296 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2297 #include "EXTERN.h"
2298 #include "perl.h"
2299 #include "XSUB.h"
ee072b34
GS
2300
2301 /* pTHX_ only needed for functions that call Perl API */
fd061412 2302 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2303
fd061412 2304 STATIC void
c52f9dcd
JH
2305 my_private_function(pTHX_ int arg1, int arg2)
2306 {
2307 /* dTHX; not needed here, because THX is an argument */
2308 ... call Perl API functions ...
2309 }
ee072b34
GS
2310
2311 [... etc ...]
2312
c52f9dcd 2313 MODULE = Foo PACKAGE = Foo
ee072b34 2314
c52f9dcd 2315 /* typical XSUB */
ee072b34 2316
c52f9dcd
JH
2317 void
2318 my_xsub(arg)
2319 int arg
2320 CODE:
2321 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2322
2323This implementation never has to fetch the context using a function
2324call, since it is always passed as an extra argument. Depending on
2325your needs for simplicity or efficiency, you may mix the previous
2326two approaches freely.
2327
651a3225
GS
2328Never add a comma after C<pTHX> yourself--always use the form of the
2329macro with the underscore for functions that take explicit arguments,
2330or the form without the argument for functions with no explicit arguments.
ee072b34 2331
27da23d5
JH
2332If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2333definition is needed if the Perl global variables (see F<perlvars.h>
2334or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2335used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2336the need for C<dVAR> only with the said compile-time define, because
2337otherwise the Perl global variables are visible as-is.
2338
a7486cbb
JH
2339=head2 Should I do anything special if I call perl from multiple threads?
2340
2341If you create interpreters in one thread and then proceed to call them in
2342another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2343initialized correctly in each of those threads.
2344
2345The C<perl_alloc> and C<perl_clone> API functions will automatically set
2346the TLS slot to the interpreter they created, so that there is no need to do
2347anything special if the interpreter is always accessed in the same thread that
2348created it, and that thread did not create or call any other interpreters
2349afterwards. If that is not the case, you have to set the TLS slot of the
2350thread before calling any functions in the Perl API on that particular
2351interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2352thread as the first thing you do:
2353
2354 /* do this before doing anything else with some_perl */
2355 PERL_SET_CONTEXT(some_perl);
2356
2357 ... other Perl API calls on some_perl go here ...
2358
ee072b34
GS
2359=head2 Future Plans and PERL_IMPLICIT_SYS
2360
2361Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2362that the interpreter knows about itself and pass it around, so too are
2363there plans to allow the interpreter to bundle up everything it knows
2364about the environment it's running on. This is enabled with the
7b52221d
RGS
2365PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2366Windows.
ee072b34
GS
2367
2368This allows the ability to provide an extra pointer (called the "host"
2369environment) for all the system calls. This makes it possible for
2370all the system stuff to maintain their own state, broken down into
2371seven C structures. These are thin wrappers around the usual system
0147cd53 2372calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2373more ambitious host (like the one that would do fork() emulation) all
2374the extra work needed to pretend that different interpreters are
2375actually different "processes", would be done here.
2376
2377The Perl engine/interpreter and the host are orthogonal entities.
2378There could be one or more interpreters in a process, and one or
2379more "hosts", with free association between them.
2380
a422fd2d
SC
2381=head1 Internal Functions
2382
2383All of Perl's internal functions which will be exposed to the outside
06f6df17 2384world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d 2385functions or functions used in a program in which Perl is embedded.
10e2eb10 2386Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2387static functions start with C<S_>.)
a422fd2d 2388
0972ecdf
DM
2389Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2390either with or without the C<Perl_> prefix, thanks to a bunch of defines
10e2eb10 2391that live in F<embed.h>. Note that extension code should I<not> set
0972ecdf
DM
2392C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2393breakage of the XS in each new perl release.
2394
2395The file F<embed.h> is generated automatically from
10e2eb10 2396F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
dc9b1d22 2397header files for the internal functions, generates the documentation
10e2eb10 2398and a lot of other bits and pieces. It's important that when you add
dc9b1d22 2399a new function to the core or change an existing one, you change the
10e2eb10 2400data in the table in F<embed.fnc> as well. Here's a sample entry from
dc9b1d22 2401that table:
a422fd2d
SC
2402
2403 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2404
10e2eb10
FC
2405The second column is the return type, the third column the name. Columns
2406after that are the arguments. The first column is a set of flags:
a422fd2d
SC
2407
2408=over 3
2409
2410=item A
2411
10e2eb10
FC
2412This function is a part of the public
2413API. All such functions should also
1aa6ea50 2414have 'd', very few do not.
a422fd2d
SC
2415
2416=item p
2417
1aa6ea50
JC
2418This function has a C<Perl_> prefix; i.e. it is defined as
2419C<Perl_av_fetch>.
a422fd2d
SC
2420
2421=item d
2422
2423This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2424look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2425
2426=back
2427
2428Other available flags are:
2429
2430=over 3
2431
2432=item s
2433
1aa6ea50
JC
2434This is a static function and is defined as C<STATIC S_whatever>, and
2435usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2436
2437=item n
2438
da8c5729 2439This does not need an interpreter context, so the definition has no
1aa6ea50 2440C<pTHX>, and it follows that callers don't use C<aTHX>. (See
d3a43cd8 2441L</Background and PERL_IMPLICIT_CONTEXT>.)
a422fd2d
SC
2442
2443=item r
2444
2445This function never returns; C<croak>, C<exit> and friends.
2446
2447=item f
2448
2449This function takes a variable number of arguments, C<printf> style.
2450The argument list should end with C<...>, like this:
2451
2452 Afprd |void |croak |const char* pat|...
2453
a7486cbb 2454=item M
a422fd2d 2455
00aadd71 2456This function is part of the experimental development API, and may change
a422fd2d
SC
2457or disappear without notice.
2458
2459=item o
2460
2461This function should not have a compatibility macro to define, say,
10e2eb10 2462C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
a422fd2d 2463
a422fd2d
SC
2464=item x
2465
2466This function isn't exported out of the Perl core.
2467
dc9b1d22
MHM
2468=item m
2469
2470This is implemented as a macro.
2471
2472=item X
2473
2474This function is explicitly exported.
2475
2476=item E
2477
2478This function is visible to extensions included in the Perl core.
2479
2480=item b
2481
2482Binary backward compatibility; this function is a macro but also has
2483a C<Perl_> implementation (which is exported).
2484
1aa6ea50
JC
2485=item others
2486
2487See the comments at the top of C<embed.fnc> for others.
2488
a422fd2d
SC
2489=back
2490
dc9b1d22
MHM
2491If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2492C<make regen_headers> to force a rebuild of F<embed.h> and other
2493auto-generated files.
a422fd2d 2494
6b4667fc 2495=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2496
6b4667fc
A
2497If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2498formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2499following macros for portability
9dd9db0b 2500
c52f9dcd
JH
2501 IVdf IV in decimal
2502 UVuf UV in decimal
2503 UVof UV in octal
2504 UVxf UV in hexadecimal
2505 NVef NV %e-like
2506 NVff NV %f-like
2507 NVgf NV %g-like
9dd9db0b 2508
6b4667fc
A
2509These will take care of 64-bit integers and long doubles.
2510For example:
2511
c52f9dcd 2512 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2513
2514The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2515
8908e76d
JH
2516If you are printing addresses of pointers, use UVxf combined
2517with PTR2UV(), do not use %lx or %p.
2518
2519=head2 Pointer-To-Integer and Integer-To-Pointer
2520
2521Because pointer size does not necessarily equal integer size,
2522use the follow macros to do it right.
2523
c52f9dcd
JH
2524 PTR2UV(pointer)
2525 PTR2IV(pointer)
2526 PTR2NV(pointer)
2527 INT2PTR(pointertotype, integer)
8908e76d
JH
2528
2529For example:
2530
c52f9dcd
JH
2531 IV iv = ...;
2532 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2533
2534and
2535
c52f9dcd
JH
2536 AV *av = ...;
2537 UV uv = PTR2UV(av);
8908e76d 2538
0ca3a874
MHM
2539=head2 Exception Handling
2540
9b5c3821 2541There are a couple of macros to do very basic exception handling in XS
10e2eb10 2542modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
9b5c3821
MHM
2543be able to use these macros:
2544
2545 #define NO_XSLOCKS
2546 #include "XSUB.h"
2547
2548You can use these macros if you call code that may croak, but you need
10e2eb10 2549to do some cleanup before giving control back to Perl. For example:
0ca3a874 2550
d7f8936a 2551 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2552
2553 XCPT_TRY_START {
2554 code_that_may_croak();
2555 } XCPT_TRY_END
2556
2557 XCPT_CATCH
2558 {
2559 /* do cleanup here */
2560 XCPT_RETHROW;
2561 }
2562
2563Note that you always have to rethrow an exception that has been
10e2eb10
FC
2564caught. Using these macros, it is not possible to just catch the
2565exception and ignore it. If you have to ignore the exception, you
0ca3a874
MHM
2566have to use the C<call_*> function.
2567
2568The advantage of using the above macros is that you don't have
2569to setup an extra function for C<call_*>, and that using these
2570macros is faster than using C<call_*>.
2571
a422fd2d
SC
2572=head2 Source Documentation
2573
2574There's an effort going on to document the internal functions and
2575automatically produce reference manuals from them - L<perlapi> is one
2576such manual which details all the functions which are available to XS
10e2eb10 2577writers. L<perlintern> is the autogenerated manual for the functions
a422fd2d
SC
2578which are not part of the API and are supposedly for internal use only.
2579
2580Source documentation is created by putting POD comments into the C
2581source, like this:
2582
2583 /*
2584 =for apidoc sv_setiv
2585
2586 Copies an integer into the given SV. Does not handle 'set' magic. See
2587 C<sv_setiv_mg>.
2588
2589 =cut
2590 */
2591
2592Please try and supply some documentation if you add functions to the
2593Perl core.
2594
0d098d33
MHM
2595=head2 Backwards compatibility
2596
10e2eb10
FC
2597The Perl API changes over time. New functions are
2598added or the interfaces of existing functions are
2599changed. The C<Devel::PPPort> module tries to
0d098d33
MHM
2600provide compatibility code for some of these changes, so XS writers don't
2601have to code it themselves when supporting multiple versions of Perl.
2602
2603C<Devel::PPPort> generates a C header file F<ppport.h> that can also
10e2eb10 2604be run as a Perl script. To generate F<ppport.h>, run:
0d098d33
MHM
2605
2606 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2607
2608Besides checking existing XS code, the script can also be used to retrieve
2609compatibility information for various API calls using the C<--api-info>
10e2eb10 2610command line switch. For example:
0d098d33
MHM
2611
2612 % perl ppport.h --api-info=sv_magicext
2613
2614For details, see C<perldoc ppport.h>.
2615
a422fd2d
SC
2616=head1 Unicode Support
2617
10e2eb10 2618Perl 5.6.0 introduced Unicode support. It's important for porters and XS
a422fd2d
SC
2619writers to understand this support and make sure that the code they
2620write does not corrupt Unicode data.
2621
2622=head2 What B<is> Unicode, anyway?
2623
10e2eb10
FC
2624In the olden, less enlightened times, we all used to use ASCII. Most of
2625us did, anyway. The big problem with ASCII is that it's American. Well,
a422fd2d 2626no, that's not actually the problem; the problem is that it's not
10e2eb10 2627particularly useful for people who don't use the Roman alphabet. What
a422fd2d 2628used to happen was that particular languages would stick their own
10e2eb10 2629alphabet in the upper range of the sequence, between 128 and 255. Of
a422fd2d
SC
2630course, we then ended up with plenty of variants that weren't quite
2631ASCII, and the whole point of it being a standard was lost.
2632
2633Worse still, if you've got a language like Chinese or
2634Japanese that has hundreds or thousands of characters, then you really
2635can't fit them into a mere 256, so they had to forget about ASCII
2636altogether, and build their own systems using pairs of numbers to refer
2637to one character.
2638
2639To fix this, some people formed Unicode, Inc. and
2640produced a new character set containing all the characters you can
10e2eb10
FC
2641possibly think of and more. There are several ways of representing these
2642characters, and the one Perl uses is called UTF-8. UTF-8 uses
2643a variable number of bytes to represent a character. You can learn more
2575c402 2644about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2645
1e54db1a 2646=head2 How can I recognise a UTF-8 string?
a422fd2d 2647
10e2eb10
FC
2648You can't. This is because UTF-8 data is stored in bytes just like
2649non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d 2650capital E with a grave accent, is represented by the two bytes
10e2eb10
FC
2651C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2652has that byte sequence as well. So you can't tell just by looking - this
a422fd2d
SC
2653is what makes Unicode input an interesting problem.
2654
2575c402
JW
2655In general, you either have to know what you're dealing with, or you
2656have to guess. The API function C<is_utf8_string> can help; it'll tell
10e2eb10
FC
2657you if a string contains only valid UTF-8 characters. However, it can't
2658do the work for you. On a character-by-character basis,
49f4c4e4 2659C<is_utf8_char_buf>
2575c402 2660will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2661
1e54db1a 2662=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2663
1e54db1a 2664As mentioned above, UTF-8 uses a variable number of bytes to store a
10e2eb10
FC
2665character. Characters with values 0...127 are stored in one
2666byte, just like good ol' ASCII. Character 128 is stored as
2667C<v194.128>; this continues up to character 191, which is
2668C<v194.191>. Now we've run out of bits (191 is binary
2669C<10111111>) so we move on; 192 is C<v195.128>. And
a422fd2d
SC
2670so it goes on, moving to three bytes at character 2048.
2671
1e54db1a 2672Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2673how long the first character in it is with the C<UTF8SKIP> macro:
2674
2675 char *utf = "\305\233\340\240\201";
2676 I32 len;
2677
2678 len = UTF8SKIP(utf); /* len is 2 here */
2679 utf += len;
2680 len = UTF8SKIP(utf); /* len is 3 here */
2681
1e54db1a 2682Another way to skip over characters in a UTF-8 string is to use
a422fd2d 2683C<utf8_hop>, which takes a string and a number of characters to skip
10e2eb10 2684over. You're on your own about bounds checking, though, so don't use it
a422fd2d
SC
2685lightly.
2686
1e54db1a 2687All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2688so you can test if you need to do something special with this
2689character like this (the UTF8_IS_INVARIANT() is a macro that tests
9f98c7fe 2690whether the byte is encoded as a single byte even in UTF-8):
a422fd2d 2691
3a2263fe 2692 U8 *utf;
4b88fb76 2693 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
3a2263fe 2694 UV uv; /* Note: a UV, not a U8, not a char */
95701e00 2695 STRLEN len; /* length of character in bytes */
a422fd2d 2696
3a2263fe 2697 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2698 /* Must treat this as UTF-8 */
4b88fb76 2699 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
a422fd2d
SC
2700 else
2701 /* OK to treat this character as a byte */
2702 uv = *utf;
2703
4b88fb76 2704You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
95701e00 2705value of the character; the inverse function C<uvchr_to_utf8> is available
1e54db1a 2706for putting a UV into UTF-8:
a422fd2d 2707
3a2263fe 2708 if (!UTF8_IS_INVARIANT(uv))
a422fd2d 2709 /* Must treat this as UTF8 */
95701e00 2710 utf8 = uvchr_to_utf8(utf8, uv);
a422fd2d
SC
2711 else
2712 /* OK to treat this character as a byte */
2713 *utf8++ = uv;
2714
2715You B<must> convert characters to UVs using the above functions if
1e54db1a 2716you're ever in a situation where you have to match UTF-8 and non-UTF-8
10e2eb10 2717characters. You may not skip over UTF-8 characters in this case. If you
1e54db1a
JH
2718do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2719for instance, if your UTF-8 string contains C<v196.172>, and you skip
2720that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2721So don't do that!
2722
1e54db1a 2723=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2724
2725Currently, Perl deals with Unicode strings and non-Unicode strings
10e2eb10
FC
2726slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2727string is internally encoded as UTF-8. Without it, the byte value is the
2575c402 2728codepoint number and vice versa (in other words, the string is encoded
e1b711da 2729as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
10e2eb10 2730semantics). You can check and manipulate this flag with the
2575c402 2731following macros:
a422fd2d
SC
2732
2733 SvUTF8(sv)
2734 SvUTF8_on(sv)
2735 SvUTF8_off(sv)
2736
2737This flag has an important effect on Perl's treatment of the string: if
2738Unicode data is not properly distinguished, regular expressions,
2739C<length>, C<substr> and other string handling operations will have
2740undesirable results.
2741
2742The problem comes when you have, for instance, a string that isn't
2575c402 2743flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
1e54db1a 2744especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2745
2746Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2747need be sure you don't accidentally knock it off while you're
10e2eb10 2748manipulating SVs. More specifically, you cannot expect to do this:
a422fd2d
SC
2749
2750 SV *sv;
2751 SV *nsv;
2752 STRLEN len;
2753 char *p;
2754
2755 p = SvPV(sv, len);
2756 frobnicate(p);
2757 nsv = newSVpvn(p, len);
2758
2759The C<char*> string does not tell you the whole story, and you can't
10e2eb10 2760copy or reconstruct an SV just by copying the string value. Check if the
2575c402 2761old SV has the UTF8 flag set, and act accordingly:
a422fd2d
SC
2762
2763 p = SvPV(sv, len);
2764 frobnicate(p);
2765 nsv = newSVpvn(p, len);
2766 if (SvUTF8(sv))
2767 SvUTF8_on(nsv);
2768
2769In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2770not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2771appropriately.
2772
3a2263fe 2773Since just passing an SV to an XS function and copying the data of
2575c402 2774the SV is not enough to copy the UTF8 flags, even less right is just
3a2263fe
RGS
2775passing a C<char *> to an XS function.
2776
1e54db1a 2777=head2 How do I convert a string to UTF-8?
a422fd2d 2778
2575c402 2779If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
10e2eb10 2780one of the strings to UTF-8. If you've got an SV, the easiest way to do
2575c402 2781this is:
a422fd2d
SC
2782
2783 sv_utf8_upgrade(sv);
2784
2785However, you must not do this, for example:
2786
2787 if (!SvUTF8(left))
2788 sv_utf8_upgrade(left);
2789
2790If you do this in a binary operator, you will actually change one of the
b1866b2d 2791strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2792by the end user, it can cause problems in deficient code.
a422fd2d 2793
1e54db1a 2794Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
10e2eb10
FC
2795string argument. This is useful for having the data available for
2796comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2797C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2798the string contains any characters above 255 that can't be represented
2799in a single byte.
2800
2801=head2 Is there anything else I need to know?
2802
10e2eb10 2803Not really. Just remember these things:
a422fd2d
SC
2804
2805=over 3
2806
2807=item *
2808
10e2eb10
FC
2809There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2810is UTF-8 by looking at its C<SvUTF8> flag. Don't forget to set the flag if
2811something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2812it's not - if you pass on the PV to somewhere, pass on the flag too.
2813
2814=item *
2815
4b88fb76 2816If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3a2263fe 2817unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2818
2819=item *
2820
1e54db1a 2821When writing a character C<uv> to a UTF-8 string, B<always> use
95701e00 2822C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
3a2263fe 2823you can use C<*s = uv>.
a422fd2d
SC
2824
2825=item *
2826
10e2eb10
FC
2827Mixing UTF-8 and non-UTF-8 strings is
2828tricky. Use C<bytes_to_utf8> to get
2bbc8d55 2829a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
2830
2831=back
2832
53e06cf0
SC
2833=head1 Custom Operators
2834
2a0fd0f1 2835Custom operator support is an experimental feature that allows you to
10e2eb10 2836define your own ops. This is primarily to allow the building of
53e06cf0
SC
2837interpreters for other languages in the Perl core, but it also allows
2838optimizations through the creation of "macro-ops" (ops which perform the
2839functions of multiple ops which are usually executed together, such as
1aa6ea50 2840C<gvsv, gvsv, add>.)
53e06cf0 2841
10e2eb10 2842This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0 2843core does not "know" anything special about this op type, and so it will
10e2eb10 2844not be involved in any optimizations. This also means that you can
53e06cf0
SC
2845define your custom ops to be any op structure - unary, binary, list and
2846so on - you like.
2847
10e2eb10
FC
2848It's important to know what custom operators won't do for you. They
2849won't let you add new syntax to Perl, directly. They won't even let you
2850add new keywords, directly. In fact, they won't change the way Perl
2851compiles a program at all. You have to do those changes yourself, after
2852Perl has compiled the program. You do this either by manipulating the op
53e06cf0
SC
2853tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2854a custom peephole optimizer with the C<optimize> module.
2855
2856When you do this, you replace ordinary Perl ops with custom ops by
407f86e1 2857creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
10e2eb10
FC
2858PP function. This should be defined in XS code, and should look like
2859the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
53e06cf0
SC
2860takes the appropriate number of values from the stack, and you are
2861responsible for adding stack marks if necessary.
2862
2863You should also "register" your op with the Perl interpreter so that it
10e2eb10 2864can produce sensible error and warning messages. Since it is possible to
53e06cf0 2865have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
9733086d 2866Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
10e2eb10 2867it is dealing with. You should create an C<XOP> structure for each
9733086d
BM
2868ppaddr you use, set the properties of the custom op with
2869C<XopENTRY_set>, and register the structure against the ppaddr using
10e2eb10 2870C<Perl_custom_op_register>. A trivial example might look like:
9733086d
BM
2871
2872 static XOP my_xop;
2873 static OP *my_pp(pTHX);
2874
2875 BOOT:
2876 XopENTRY_set(&my_xop, xop_name, "myxop");
2877 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2878 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2879
2880The available fields in the structure are:
2881
2882=over 4
2883
2884=item xop_name
2885
10e2eb10 2886A short name for your op. This will be included in some error messages,
9733086d
BM
2887and will also be returned as C<< $op->name >> by the L<B|B> module, so
2888it will appear in the output of module like L<B::Concise|B::Concise>.
2889
2890=item xop_desc
2891
2892A short description of the function of the op.
2893
2894=item xop_class
2895
10e2eb10 2896Which of the various C<*OP> structures this op uses. This should be one of
9733086d
BM
2897the C<OA_*> constants from F<op.h>, namely
2898
2899=over 4
2900
2901=item OA_BASEOP
2902
2903=item OA_UNOP
2904
2905=item OA_BINOP
2906
2907=item OA_LOGOP
2908
2909=item OA_LISTOP
2910
2911=item OA_PMOP
2912
2913=item OA_SVOP
2914
2915=item OA_PADOP
2916
2917=item OA_PVOP_OR_SVOP
2918
10e2eb10 2919This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
9733086d
BM
2920the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
2921
2922=item OA_LOOP
2923
2924=item OA_COP
2925
2926=back
2927
2928The other C<OA_*> constants should not be used.
2929
2930=item xop_peep
2931
2932This member is of type C<Perl_cpeep_t>, which expands to C<void
10e2eb10 2933(*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
9733086d 2934will be called from C<Perl_rpeep> when ops of this type are encountered
10e2eb10 2935by the peephole optimizer. I<o> is the OP that needs optimizing;
9733086d
BM
2936I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
2937
2938=back
53e06cf0 2939
e7d4c058 2940C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 2941
954c1994 2942=head1 AUTHORS
e89caa19 2943
954c1994 2944Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
2945E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2946itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 2947
954c1994
GS
2948With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2949Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2950Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2951Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 2952
954c1994 2953=head1 SEE ALSO
cb1a09d0 2954
ba555bf5 2955L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>