This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Tests for the new custom op registrations.
[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
06f6df17 8to provide some info on the basic workings of the Perl core. It is far
b3b6085d
PP
9from complete and probably contains many errors. Please refer any
10questions or comments to the author below.
a0d0e21e 11
0a753a76
PP
12=head1 Variables
13
5f05dabc 14=head2 Datatypes
a0d0e21e
LW
15
16Perl has three typedefs that handle Perl's three main data types:
17
18 SV Scalar Value
19 AV Array Value
20 HV Hash Value
21
d1b91892 22Each typedef has specific routines that manipulate the various data types.
a0d0e21e
LW
23
24=head2 What is an "IV"?
25
954c1994 26Perl uses a special typedef IV which is a simple signed integer type that is
5f05dabc 27guaranteed to be large enough to hold a pointer (as well as an integer).
954c1994 28Additionally, there is the UV, which is simply an unsigned IV.
a0d0e21e 29
d1b91892 30Perl also uses two special typedefs, I32 and I16, which will always be at
954c1994 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).
a0d0e21e 40
20dbd849 41The seven routines are:
a0d0e21e
LW
42
43 SV* newSViv(IV);
20dbd849 44 SV* newSVuv(UV);
a0d0e21e 45 SV* newSVnv(double);
06f6df17
RGS
46 SV* newSVpv(const char*, STRLEN);
47 SV* newSVpvn(const char*, STRLEN);
46fc3d4c 48 SV* newSVpvf(const char*, ...);
a0d0e21e
LW
49 SV* newSVsv(SV*);
50
06f6df17
RGS
51C<STRLEN> is an integer type (Size_t, usually defined as size_t in
52F<config.h>) guaranteed to be large enough to represent the size of
53any string that perl can handle.
54
55In the unlikely case of a SV requiring more complex initialisation, you
56can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
57type NULL is returned, else an SV of type PV is returned with len + 1 (for
58the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
59the SV has value undef.
20dbd849 60
06f6df17
RGS
61 SV *sv = newSV(0); /* no storage allocated */
62 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage allocated */
20dbd849 63
06f6df17 64To change the value of an I<already-existing> SV, there are eight routines:
a0d0e21e
LW
65
66 void sv_setiv(SV*, IV);
deb3007b 67 void sv_setuv(SV*, UV);
a0d0e21e 68 void sv_setnv(SV*, double);
08105a92 69 void sv_setpv(SV*, const char*);
06f6df17 70 void sv_setpvn(SV*, const char*, STRLEN)
46fc3d4c 71 void sv_setpvf(SV*, const char*, ...);
5b3218b9 72 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool *);
a0d0e21e
LW
73 void sv_setsv(SV*, SV*);
74
75Notice that you can choose to specify the length of the string to be
9da1e3b5
MUN
76assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
77allow Perl to calculate the length by using C<sv_setpv> or by specifying
780 as the second argument to C<newSVpv>. Be warned, though, that Perl will
79determine the string's length by using C<strlen>, which depends on the
9abd00ed
GS
80string terminating with a NUL character.
81
82The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
83formatted output becomes the value.
84
328bf373 85C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
9abd00ed
GS
86either a pointer to a variable argument list or the address and length of
87an array of SVs. The last argument points to a boolean; on return, if that
88boolean is true, then locale-specific information has been used to format
c2611fb3 89the string, and the string's contents are therefore untrustworthy (see
9abd00ed
GS
90L<perlsec>). This pointer may be NULL if that information is not
91important. Note that this function requires you to specify the length of
92the format.
93
9da1e3b5
MUN
94The C<sv_set*()> functions are not generic enough to operate on values
95that have "magic". See L<Magic Virtual Tables> later in this document.
a0d0e21e 96
a3cb178b
GS
97All SVs that contain strings should be terminated with a NUL character.
98If it is not NUL-terminated there is a risk of
5f05dabc
PP
99core dumps and corruptions from code which passes the string to C
100functions or system calls which expect a NUL-terminated string.
101Perl's own functions typically add a trailing NUL for this reason.
102Nevertheless, you should be very careful when you pass a string stored
103in an SV to a C function or system call.
104
a0d0e21e
LW
105To access the actual value that an SV points to, you can use the macros:
106
107 SvIV(SV*)
954c1994 108 SvUV(SV*)
a0d0e21e
LW
109 SvNV(SV*)
110 SvPV(SV*, STRLEN len)
1fa8b10d 111 SvPV_nolen(SV*)
a0d0e21e 112
954c1994 113which will automatically coerce the actual scalar type into an IV, UV, double,
a0d0e21e
LW
114or string.
115
116In the C<SvPV> macro, the length of the string returned is placed into the
1fa8b10d
JD
117variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
118not care what the length of the data is, use the C<SvPV_nolen> macro.
119Historically the C<SvPV> macro with the global variable C<PL_na> has been
120used in this case. But that can be quite inefficient because C<PL_na> must
121be accessed in thread-local storage in threaded Perl. In any case, remember
122that Perl allows arbitrary strings of data that may both contain NULs and
123might not be terminated by a NUL.
a0d0e21e 124
ce2f5d8f
KA
125Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
126len);>. It might work with your compiler, but it won't work for everyone.
127Break this sort of statement up into separate assignments:
128
1aa6ea50
JC
129 SV *s;
130 STRLEN len;
131 char * ptr;
132 ptr = SvPV(s, len);
133 foo(ptr, len);
ce2f5d8f 134
07fa94a1 135If you want to know if the scalar value is TRUE, you can use:
a0d0e21e
LW
136
137 SvTRUE(SV*)
138
139Although Perl will automatically grow strings for you, if you need to force
140Perl to allocate more memory for your SV, you can use the macro
141
142 SvGROW(SV*, STRLEN newlen)
143
144which will determine if more memory needs to be allocated. If so, it will
145call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
5f05dabc
PP
146decrease, the allocated memory of an SV and that it does not automatically
147add a byte for the a trailing NUL (perl's own string functions typically do
8ebc5c01 148C<SvGROW(sv, len + 1)>).
a0d0e21e
LW
149
150If you have an SV and want to know what kind of data Perl thinks is stored
151in it, you can use the following macros to check the type of SV you have.
152
153 SvIOK(SV*)
154 SvNOK(SV*)
155 SvPOK(SV*)
156
157You can get and set the current length of the string stored in an SV with
158the following macros:
159
160 SvCUR(SV*)
161 SvCUR_set(SV*, I32 val)
162
cb1a09d0
AD
163You can also get a pointer to the end of the string stored in the SV
164with the macro:
165
166 SvEND(SV*)
167
168But note that these last three macros are valid only if C<SvPOK()> is true.
a0d0e21e 169
d1b91892
AD
170If you want to append something to the end of string stored in an C<SV*>,
171you can use the following functions:
172
08105a92 173 void sv_catpv(SV*, const char*);
e65f3abd 174 void sv_catpvn(SV*, const char*, STRLEN);
46fc3d4c 175 void sv_catpvf(SV*, const char*, ...);
328bf373 176 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
d1b91892
AD
177 void sv_catsv(SV*, SV*);
178
179The first function calculates the length of the string to be appended by
180using C<strlen>. In the second, you specify the length of the string
46fc3d4c 181yourself. The third function processes its arguments like C<sprintf> and
9abd00ed
GS
182appends the formatted output. The fourth function works like C<vsprintf>.
183You can specify the address and length of an array of SVs instead of the
184va_list argument. The fifth function extends the string stored in the first
185SV with the string stored in the second SV. It also forces the second SV
186to be interpreted as a string.
187
188The C<sv_cat*()> functions are not generic enough to operate on values that
189have "magic". See L<Magic Virtual Tables> later in this document.
d1b91892 190
a0d0e21e
LW
191If you know the name of a scalar variable, you can get a pointer to its SV
192by using the following:
193
64ace3f8 194 SV* get_sv("package::varname", 0);
a0d0e21e
LW
195
196This returns NULL if the variable does not exist.
197
d1b91892 198If you want to know if this variable (or any other SV) is actually C<defined>,
a0d0e21e
LW
199you can call:
200
201 SvOK(SV*)
202
06f6df17 203The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
9adebda4
SB
204
205Its address can be used whenever an C<SV*> is needed. Make sure that
206you don't try to compare a random sv with C<&PL_sv_undef>. For example
207when interfacing Perl code, it'll work correctly for:
208
209 foo(undef);
210
211But won't work when called as:
212
213 $x = undef;
214 foo($x);
215
216So to repeat always use SvOK() to check whether an sv is defined.
217
218Also you have to be careful when using C<&PL_sv_undef> as a value in
219AVs or HVs (see L<AVs, HVs and undefined values>).
a0d0e21e 220
06f6df17
RGS
221There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
222boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
223addresses can be used whenever an C<SV*> is needed.
a0d0e21e 224
9cde0e7f 225Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
a0d0e21e
LW
226Take this code:
227
228 SV* sv = (SV*) 0;
229 if (I-am-to-return-a-real-value) {
230 sv = sv_2mortal(newSViv(42));
231 }
232 sv_setsv(ST(0), sv);
233
234This code tries to return a new SV (which contains the value 42) if it should
04343c6d 235return a real value, or undef otherwise. Instead it has returned a NULL
a0d0e21e 236pointer which, somewhere down the line, will cause a segmentation violation,
06f6df17
RGS
237bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
238first line and all will be well.
a0d0e21e
LW
239
240To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
3fe9a6f1 241call is not necessary (see L<Reference Counts and Mortality>).
a0d0e21e 242
94dde4fb
SC
243=head2 Offsets
244
245Perl provides the function C<sv_chop> to efficiently remove characters
246from the beginning of a string; you give it an SV and a pointer to
da75cd15 247somewhere inside the PV, and it discards everything before the
94dde4fb
SC
248pointer. The efficiency comes by means of a little hack: instead of
249actually removing the characters, C<sv_chop> sets the flag C<OOK>
250(offset OK) to signal to other functions that the offset hack is in
251effect, and it puts the number of bytes chopped off into the IV field
252of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
00aadd71 253many bytes, and adjusts C<SvCUR> and C<SvLEN>.
94dde4fb
SC
254
255Hence, at this point, the start of the buffer that we allocated lives
256at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
257into the middle of this allocated storage.
258
259This is best demonstrated by example:
260
261 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
262 SV = PVIV(0x8128450) at 0x81340f0
263 REFCNT = 1
264 FLAGS = (POK,OOK,pPOK)
265 IV = 1 (OFFSET)
266 PV = 0x8135781 ( "1" . ) "2345"\0
267 CUR = 4
268 LEN = 5
269
270Here the number of bytes chopped off (1) is put into IV, and
271C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
272portion of the string between the "real" and the "fake" beginnings is
273shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
274the fake beginning, not the real one.
275
fe854a6f 276Something similar to the offset hack is performed on AVs to enable
319cef53
SC
277efficient shifting and splicing off the beginning of the array; while
278C<AvARRAY> points to the first element in the array that is visible from
279Perl, C<AvALLOC> points to the real start of the C array. These are
280usually the same, but a C<shift> operation can be carried out by
6de131f0 281increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
319cef53
SC
282Again, the location of the real start of the C array only comes into
283play when freeing the array. See C<av_shift> in F<av.c>.
284
d1b91892 285=head2 What's Really Stored in an SV?
a0d0e21e
LW
286
287Recall that the usual method of determining the type of scalar you have is
5f05dabc 288to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 289usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
290macros will do the appropriate conversion of string to integer/double or
291integer/double to string.
292
293If you I<really> need to know if you have an integer, double, or string
294pointer in an SV, you can use the following three macros instead:
295
296 SvIOKp(SV*)
297 SvNOKp(SV*)
298 SvPOKp(SV*)
299
300These will tell you if you truly have an integer, double, or string pointer
d1b91892 301stored in your SV. The "p" stands for private.
a0d0e21e 302
9e9796d6
JH
303The are various ways in which the private and public flags may differ.
304For example, a tied SV may have a valid underlying value in the IV slot
305(so SvIOKp is true), but the data should be accessed via the FETCH
306routine rather than directly, so SvIOK is false. Another is when
d7f8936a 307numeric conversion has occurred and precision has been lost: only the
9e9796d6
JH
308private flag is set on 'lossy' values. So when an NV is converted to an
309IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
310
07fa94a1 311In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 312
54310121 313=head2 Working with AVs
a0d0e21e 314
07fa94a1
JO
315There are two ways to create and load an AV. The first method creates an
316empty AV:
a0d0e21e
LW
317
318 AV* newAV();
319
54310121 320The second method both creates the AV and initially populates it with SVs:
a0d0e21e
LW
321
322 AV* av_make(I32 num, SV **ptr);
323
5f05dabc 324The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 325AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 326
54310121 327Once the AV has been created, the following operations are possible on AVs:
a0d0e21e
LW
328
329 void av_push(AV*, SV*);
330 SV* av_pop(AV*);
331 SV* av_shift(AV*);
332 void av_unshift(AV*, I32 num);
333
334These should be familiar operations, with the exception of C<av_unshift>.
335This routine adds C<num> elements at the front of the array with the C<undef>
336value. You must then use C<av_store> (described below) to assign values
337to these new elements.
338
339Here are some other functions:
340
5f05dabc 341 I32 av_len(AV*);
a0d0e21e 342 SV** av_fetch(AV*, I32 key, I32 lval);
a0d0e21e 343 SV** av_store(AV*, I32 key, SV* val);
a0d0e21e 344
5f05dabc
PP
345The C<av_len> function returns the highest index value in array (just
346like $#array in Perl). If the array is empty, -1 is returned. The
347C<av_fetch> function returns the value at index C<key>, but if C<lval>
348is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
349The C<av_store> function stores the value C<val> at index C<key>, and does
350not increment the reference count of C<val>. Thus the caller is responsible
351for taking care of that, and if C<av_store> returns NULL, the caller will
352have to decrement the reference count to avoid a memory leak. Note that
353C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
354return value.
d1b91892 355
a0d0e21e 356 void av_clear(AV*);
a0d0e21e 357 void av_undef(AV*);
cb1a09d0 358 void av_extend(AV*, I32 key);
5f05dabc
PP
359
360The C<av_clear> function deletes all the elements in the AV* array, but
361does not actually delete the array itself. The C<av_undef> function will
362delete all the elements in the array plus the array itself. The
adc882cf
GS
363C<av_extend> function extends the array so that it contains at least C<key+1>
364elements. If C<key+1> is less than the currently allocated length of the array,
365then nothing is done.
a0d0e21e
LW
366
367If you know the name of an array variable, you can get a pointer to its AV
368by using the following:
369
cbfd0a87 370 AV* get_av("package::varname", 0);
a0d0e21e
LW
371
372This returns NULL if the variable does not exist.
373
04343c6d
GS
374See L<Understanding the Magic of Tied Hashes and Arrays> for more
375information on how to use the array access functions on tied arrays.
376
54310121 377=head2 Working with HVs
a0d0e21e
LW
378
379To create an HV, you use the following routine:
380
381 HV* newHV();
382
54310121 383Once the HV has been created, the following operations are possible on HVs:
a0d0e21e 384
08105a92
GS
385 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
386 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 387
5f05dabc
PP
388The C<klen> parameter is the length of the key being passed in (Note that
389you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
390length of the key). The C<val> argument contains the SV pointer to the
54310121 391scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
392you want C<hv_store> to calculate it for you). The C<lval> parameter
393indicates whether this fetch is actually a part of a store operation, in
394which case a new undefined value will be added to the HV with the supplied
395key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 396
5f05dabc
PP
397Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
398C<SV*>. To access the scalar value, you must first dereference the return
399value. However, you should check to make sure that the return value is
400not NULL before dereferencing it.
a0d0e21e
LW
401
402These two functions check if a hash table entry exists, and deletes it.
403
08105a92
GS
404 bool hv_exists(HV*, const char* key, U32 klen);
405 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 406
5f05dabc
PP
407If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
408create and return a mortal copy of the deleted value.
409
a0d0e21e
LW
410And more miscellaneous functions:
411
412 void hv_clear(HV*);
a0d0e21e 413 void hv_undef(HV*);
5f05dabc
PP
414
415Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
416table but does not actually delete the hash table. The C<hv_undef> deletes
417both the entries and the hash table itself.
a0d0e21e 418
d1b91892
AD
419Perl keeps the actual data in linked list of structures with a typedef of HE.
420These contain the actual key and value pointers (plus extra administrative
421overhead). The key is a string pointer; the value is an C<SV*>. However,
422once you have an C<HE*>, to get the actual key and value, use the routines
423specified below.
424
a0d0e21e
LW
425 I32 hv_iterinit(HV*);
426 /* Prepares starting point to traverse hash table */
427 HE* hv_iternext(HV*);
428 /* Get the next entry, and return a pointer to a
429 structure that has both the key and value */
430 char* hv_iterkey(HE* entry, I32* retlen);
431 /* Get the key from an HE structure and also return
432 the length of the key string */
cb1a09d0 433 SV* hv_iterval(HV*, HE* entry);
d1be9408 434 /* Return an SV pointer to the value of the HE
a0d0e21e 435 structure */
cb1a09d0 436 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
437 /* This convenience routine combines hv_iternext,
438 hv_iterkey, and hv_iterval. The key and retlen
439 arguments are return values for the key and its
440 length. The value is returned in the SV* argument */
a0d0e21e
LW
441
442If you know the name of a hash variable, you can get a pointer to its HV
443by using the following:
444
6673a63c 445 HV* get_hv("package::varname", 0);
a0d0e21e
LW
446
447This returns NULL if the variable does not exist.
448
8ebc5c01 449The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
a0d0e21e 450
a0d0e21e 451 hash = 0;
ab192400
GS
452 while (klen--)
453 hash = (hash * 33) + *key++;
87275199 454 hash = hash + (hash >> 5); /* after 5.6 */
ab192400 455
87275199 456The last step was added in version 5.6 to improve distribution of
ab192400 457lower bits in the resulting hash value.
a0d0e21e 458
04343c6d
GS
459See L<Understanding the Magic of Tied Hashes and Arrays> for more
460information on how to use the hash access functions on tied hashes.
461
1e422769
PP
462=head2 Hash API Extensions
463
464Beginning with version 5.004, the following functions are also supported:
465
466 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
467 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 468
1e422769
PP
469 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
470 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 471
1e422769
PP
472 SV* hv_iterkeysv (HE* entry);
473
474Note that these functions take C<SV*> keys, which simplifies writing
475of extension code that deals with hash structures. These functions
476also allow passing of C<SV*> keys to C<tie> functions without forcing
477you to stringify the keys (unlike the previous set of functions).
478
479They also return and accept whole hash entries (C<HE*>), making their
480use more efficient (since the hash number for a particular string
4a4eefd0
GS
481doesn't have to be recomputed every time). See L<perlapi> for detailed
482descriptions.
1e422769
PP
483
484The following macros must always be used to access the contents of hash
485entries. Note that the arguments to these macros must be simple
486variables, since they may get evaluated more than once. See
4a4eefd0 487L<perlapi> for detailed descriptions of these macros.
1e422769
PP
488
489 HePV(HE* he, STRLEN len)
490 HeVAL(HE* he)
491 HeHASH(HE* he)
492 HeSVKEY(HE* he)
493 HeSVKEY_force(HE* he)
494 HeSVKEY_set(HE* he, SV* sv)
495
496These two lower level macros are defined, but must only be used when
497dealing with keys that are not C<SV*>s:
498
499 HeKEY(HE* he)
500 HeKLEN(HE* he)
501
04343c6d
GS
502Note that both C<hv_store> and C<hv_store_ent> do not increment the
503reference count of the stored C<val>, which is the caller's responsibility.
504If these functions return a NULL value, the caller will usually have to
505decrement the reference count of C<val> to avoid a memory leak.
1e422769 506
a9381218
MHM
507=head2 AVs, HVs and undefined values
508
509Sometimes you have to store undefined values in AVs or HVs. Although
510this may be a rare case, it can be tricky. That's because you're
511used to using C<&PL_sv_undef> if you need an undefined SV.
512
513For example, intuition tells you that this XS code:
514
515 AV *av = newAV();
516 av_store( av, 0, &PL_sv_undef );
517
518is equivalent to this Perl code:
519
520 my @av;
521 $av[0] = undef;
522
523Unfortunately, this isn't true. AVs use C<&PL_sv_undef> as a marker
524for indicating that an array element has not yet been initialized.
525Thus, C<exists $av[0]> would be true for the above Perl code, but
526false for the array generated by the XS code.
527
528Other problems can occur when storing C<&PL_sv_undef> in HVs:
529
530 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
531
532This will indeed make the value C<undef>, but if you try to modify
533the value of C<key>, you'll get the following error:
534
535 Modification of non-creatable hash value attempted
536
537In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
538in restricted hashes. This caused such hash entries not to appear
539when iterating over the hash or when checking for the keys
540with the C<hv_exists> function.
541
8abccac8
AR
542You can run into similar problems when you store C<&PL_sv_yes> or
543C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
544will give you the following error:
545
546 Modification of a read-only value attempted
547
548To make a long story short, you can use the special variables
8abccac8 549C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
550HVs, but you have to make sure you know what you're doing.
551
552Generally, if you want to store an undefined value in an AV
553or HV, you should not use C<&PL_sv_undef>, but rather create a
554new undefined value using the C<newSV> function, for example:
555
556 av_store( av, 42, newSV(0) );
557 hv_store( hv, "foo", 3, newSV(0), 0 );
558
a0d0e21e
LW
559=head2 References
560
d1b91892
AD
561References are a special type of scalar that point to other data types
562(including references).
a0d0e21e 563
07fa94a1 564To create a reference, use either of the following functions:
a0d0e21e 565
5f05dabc
PP
566 SV* newRV_inc((SV*) thing);
567 SV* newRV_noinc((SV*) thing);
a0d0e21e 568
5f05dabc 569The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
570functions are identical except that C<newRV_inc> increments the reference
571count of the C<thing>, while C<newRV_noinc> does not. For historical
572reasons, C<newRV> is a synonym for C<newRV_inc>.
573
574Once you have a reference, you can use the following macro to dereference
575the reference:
a0d0e21e
LW
576
577 SvRV(SV*)
578
579then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 580C<AV*> or C<HV*>, if required.
a0d0e21e 581
d1b91892 582To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
583
584 SvROK(SV*)
585
07fa94a1
JO
586To discover what type of value the reference refers to, use the following
587macro and then check the return value.
d1b91892
AD
588
589 SvTYPE(SvRV(SV*))
590
591The most useful types that will be returned are:
592
593 SVt_IV Scalar
594 SVt_NV Scalar
595 SVt_PV Scalar
5f05dabc 596 SVt_RV Scalar
d1b91892
AD
597 SVt_PVAV Array
598 SVt_PVHV Hash
599 SVt_PVCV Code
5f05dabc
PP
600 SVt_PVGV Glob (possible a file handle)
601 SVt_PVMG Blessed or Magical Scalar
602
3f2c382a 603See the F<sv.h> header file for more details.
d1b91892 604
cb1a09d0
AD
605=head2 Blessed References and Class Objects
606
06f6df17 607References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
608OO lexicon, an object is simply a reference that has been blessed into a
609package (or class). Once blessed, the programmer may now use the reference
610to access the various methods in the class.
611
612A reference can be blessed into a package with the following function:
613
614 SV* sv_bless(SV* sv, HV* stash);
615
06f6df17
RGS
616The C<sv> argument must be a reference value. The C<stash> argument
617specifies which class the reference will belong to. See
2ae324a7 618L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
619
620/* Still under construction */
621
622Upgrades rv to reference if not already one. Creates new SV for rv to
8ebc5c01
PP
623point to. If C<classname> is non-null, the SV is blessed into the specified
624class. SV is returned.
cb1a09d0 625
08105a92 626 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 627
e1c57cef 628Copies integer, unsigned integer or double into an SV whose reference is C<rv>. SV is blessed
8ebc5c01 629if C<classname> is non-null.
cb1a09d0 630
08105a92 631 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 632 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 633 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 634
5f05dabc 635Copies the pointer value (I<the address, not the string!>) into an SV whose
8ebc5c01 636reference is rv. SV is blessed if C<classname> is non-null.
cb1a09d0 637
08105a92 638 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
cb1a09d0 639
8ebc5c01
PP
640Copies string into an SV whose reference is C<rv>. Set length to 0 to let
641Perl calculate the string length. SV is blessed if C<classname> is non-null.
cb1a09d0 642
e65f3abd 643 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
cb1a09d0 644
9abd00ed
GS
645Tests whether the SV is blessed into the specified class. It does not
646check inheritance relationships.
647
08105a92 648 int sv_isa(SV* sv, const char* name);
9abd00ed
GS
649
650Tests whether the SV is a reference to a blessed object.
651
652 int sv_isobject(SV* sv);
653
654Tests whether the SV is derived from the specified class. SV can be either
655a reference to a blessed object or a string containing a class name. This
656is the function implementing the C<UNIVERSAL::isa> functionality.
657
08105a92 658 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 659
00aadd71 660To check if you've got an object derived from a specific class you have
9abd00ed
GS
661to write:
662
663 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 664
5f05dabc 665=head2 Creating New Variables
cb1a09d0 666
5f05dabc
PP
667To create a new Perl variable with an undef value which can be accessed from
668your Perl script, use the following routines, depending on the variable type.
cb1a09d0 669
64ace3f8 670 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 671 AV* get_av("package::varname", GV_ADD);
6673a63c 672 HV* get_hv("package::varname", GV_ADD);
cb1a09d0
AD
673
674Notice the use of TRUE as the second parameter. The new variable can now
675be set, using the routines appropriate to the data type.
676
5f05dabc
PP
677There are additional macros whose values may be bitwise OR'ed with the
678C<TRUE> argument to enable certain extra features. Those bits are:
cb1a09d0 679
9a68f1db
SB
680=over
681
682=item GV_ADDMULTI
683
684Marks the variable as multiply defined, thus preventing the:
685
686 Name <varname> used only once: possible typo
687
688warning.
689
9a68f1db
SB
690=item GV_ADDWARN
691
692Issues the warning:
693
694 Had to create <varname> unexpectedly
695
696if the variable did not exist before the function was called.
697
698=back
cb1a09d0 699
07fa94a1
JO
700If you do not specify a package name, the variable is created in the current
701package.
cb1a09d0 702
5f05dabc 703=head2 Reference Counts and Mortality
a0d0e21e 704
d1be9408 705Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 706AVs, or HVs (xV for short in the following) start their life with a
55497cff 707reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 708then it will be destroyed and its memory made available for reuse.
55497cff
PP
709
710This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
711undef'ed or the last variable holding a reference to it is changed or
712overwritten. At the internal level, however, reference counts can be
55497cff
PP
713manipulated with the following macros:
714
715 int SvREFCNT(SV* sv);
5f05dabc 716 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
717 void SvREFCNT_dec(SV* sv);
718
719However, there is one other function which manipulates the reference
07fa94a1
JO
720count of its argument. The C<newRV_inc> function, you will recall,
721creates a reference to the specified argument. As a side effect,
722it increments the argument's reference count. If this is not what
723you want, use C<newRV_noinc> instead.
724
725For example, imagine you want to return a reference from an XSUB function.
726Inside the XSUB routine, you create an SV which initially has a reference
727count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
728This returns the reference as a new SV, but the reference count of the
729SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
730return the reference from the XSUB routine and forget about the SV.
731But Perl hasn't! Whenever the returned reference is destroyed, the
732reference count of the original SV is decreased to one and nothing happens.
733The SV will hang around without any way to access it until Perl itself
734terminates. This is a memory leak.
5f05dabc
PP
735
736The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
737C<newRV_inc>. Then, if and when the last reference is destroyed,
738the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 739stopping any memory leak.
55497cff 740
5f05dabc 741There are some convenience functions available that can help with the
54310121 742destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
743An xV that is mortal has had its reference count marked to be decremented,
744but not actually decremented, until "a short time later". Generally the
745term "short time later" means a single Perl statement, such as a call to
54310121 746an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
747reference count decremented depends on two macros, SAVETMPS and FREETMPS.
748See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
749
750"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
751However, if you mortalize a variable twice, the reference count will
752later be decremented twice.
753
00aadd71
NIS
754"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
755For example an SV which is created just to pass a number to a called sub
06f6df17
RGS
756is made mortal to have it cleaned up automatically when it's popped off
757the stack. Similarly, results returned by XSUBs (which are pushed on the
758stack) are often made mortal.
a0d0e21e
LW
759
760To create a mortal variable, use the functions:
761
762 SV* sv_newmortal()
763 SV* sv_2mortal(SV*)
764 SV* sv_mortalcopy(SV*)
765
00aadd71 766The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
767SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
768third creates a mortal copy of an existing SV.
00aadd71 769Because C<sv_newmortal> gives the new SV no value,it must normally be given one
9a68f1db 770via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
771
772 SV *tmp = sv_newmortal();
773 sv_setiv(tmp, an_integer);
774
775As that is multiple C statements it is quite common so see this idiom instead:
776
777 SV *tmp = sv_2mortal(newSViv(an_integer));
778
779
780You should be careful about creating mortal variables. Strange things
781can happen if you make the same value mortal within multiple contexts,
782or if you make a variable mortal multiple times. Thinking of "Mortalization"
783as deferred C<SvREFCNT_dec> should help to minimize such problems.
784For example if you are passing an SV which you I<know> has high enough REFCNT
785to survive its use on the stack you need not do any mortalization.
786If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
787making a C<sv_mortalcopy> is safer.
a0d0e21e 788
ac036724 789The mortal routines are not just for SVs; AVs and HVs can be
faed5253 790made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 791C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 792
5f05dabc 793=head2 Stashes and Globs
a0d0e21e 794
06f6df17
RGS
795A B<stash> is a hash that contains all variables that are defined
796within a package. Each key of the stash is a symbol
aa689395
PP
797name (shared by all the different types of objects that have the same
798name), and each value in the hash table is a GV (Glob Value). This GV
799in turn contains references to the various objects of that name,
800including (but not limited to) the following:
cb1a09d0 801
a0d0e21e
LW
802 Scalar Value
803 Array Value
804 Hash Value
a3cb178b 805 I/O Handle
a0d0e21e
LW
806 Format
807 Subroutine
808
06f6df17
RGS
809There is a single stash called C<PL_defstash> that holds the items that exist
810in the C<main> package. To get at the items in other packages, append the
811string "::" to the package name. The items in the C<Foo> package are in
812the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
813in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 814
d1b91892 815To get the stash pointer for a particular package, use the function:
a0d0e21e 816
da51bb9b
NC
817 HV* gv_stashpv(const char* name, I32 flags)
818 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
819
820The first function takes a literal string, the second uses the string stored
d1b91892 821in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 822C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
823
824The name that C<gv_stash*v> wants is the name of the package whose symbol table
825you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
826packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
827language itself.
a0d0e21e
LW
828
829Alternately, if you have an SV that is a blessed reference, you can find
830out the stash pointer by using:
831
832 HV* SvSTASH(SvRV(SV*));
833
834then use the following to get the package name itself:
835
836 char* HvNAME(HV* stash);
837
5f05dabc
PP
838If you need to bless or re-bless an object you can use the following
839function:
a0d0e21e
LW
840
841 SV* sv_bless(SV*, HV* stash)
842
843where the first argument, an C<SV*>, must be a reference, and the second
844argument is a stash. The returned C<SV*> can now be used in the same way
845as any other SV.
846
d1b91892
AD
847For more information on references and blessings, consult L<perlref>.
848
54310121 849=head2 Double-Typed SVs
0a753a76
PP
850
851Scalar variables normally contain only one type of value, an integer,
852double, pointer, or reference. Perl will automatically convert the
853actual scalar data from the stored type into the requested type.
854
855Some scalar variables contain more than one type of scalar data. For
856example, the variable C<$!> contains either the numeric value of C<errno>
857or its string equivalent from either C<strerror> or C<sys_errlist[]>.
858
859To force multiple data values into an SV, you must do two things: use the
860C<sv_set*v> routines to add the additional scalar type, then set a flag
861so that Perl will believe it contains more than one type of data. The
862four macros to set the flags are:
863
864 SvIOK_on
865 SvNOK_on
866 SvPOK_on
867 SvROK_on
868
869The particular macro you must use depends on which C<sv_set*v> routine
870you called first. This is because every C<sv_set*v> routine turns on
871only the bit for the particular type of data being set, and turns off
872all the rest.
873
874For example, to create a new Perl variable called "dberror" that contains
875both the numeric and descriptive string error values, you could use the
876following code:
877
878 extern int dberror;
879 extern char *dberror_list;
880
64ace3f8 881 SV* sv = get_sv("dberror", GV_ADD);
0a753a76
PP
882 sv_setiv(sv, (IV) dberror);
883 sv_setpv(sv, dberror_list[dberror]);
884 SvIOK_on(sv);
885
886If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
887macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
888
889=head2 Magic Variables
a0d0e21e 890
d1b91892
AD
891[This section still under construction. Ignore everything here. Post no
892bills. Everything not permitted is forbidden.]
893
d1b91892
AD
894Any SV may be magical, that is, it has special features that a normal
895SV does not have. These features are stored in the SV structure in a
5f05dabc 896linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
897
898 struct magic {
899 MAGIC* mg_moremagic;
900 MGVTBL* mg_virtual;
901 U16 mg_private;
902 char mg_type;
903 U8 mg_flags;
b205eb13 904 I32 mg_len;
d1b91892
AD
905 SV* mg_obj;
906 char* mg_ptr;
d1b91892
AD
907 };
908
909Note this is current as of patchlevel 0, and could change at any time.
910
911=head2 Assigning Magic
912
913Perl adds magic to an SV using the sv_magic function:
914
08105a92 915 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
916
917The C<sv> argument is a pointer to the SV that is to acquire a new magical
918feature.
919
920If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
645c22ef
DM
921convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
922to the beginning of the linked list of magical features. Any prior entry
923of the same type of magic is deleted. Note that this can be overridden,
924and multiple instances of the same type of magic can be associated with an
925SV.
d1b91892 926
54310121
PP
927The C<name> and C<namlen> arguments are used to associate a string with
928the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
929C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
930C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
931whether C<namlen> is greater than zero or equal to zero respectively. As a
932special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
933to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
934
935The sv_magic function uses C<how> to determine which, if any, predefined
936"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 937See the L<Magic Virtual Tables> section below. The C<how> argument is also
14befaf4 938stored in the C<mg_type> field. The value of C<how> should be chosen
06f6df17 939from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 940these macros were added, Perl internals used to directly use character
14befaf4 941literals, so you may occasionally come across old code or documentation
75d0f26d 942referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
943
944The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
945structure. If it is not the same as the C<sv> argument, the reference
946count of the C<obj> object is incremented. If it is the same, or if
645c22ef 947the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 948then C<obj> is merely stored, without the reference count being incremented.
d1b91892 949
2d8d5d5a
SH
950See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
951to an SV.
952
cb1a09d0
AD
953There is also a function to add magic to an C<HV>:
954
955 void hv_magic(HV *hv, GV *gv, int how);
956
957This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
958
959To remove the magic from an SV, call the function sv_unmagic:
960
961 void sv_unmagic(SV *sv, int type);
962
963The C<type> argument should be equal to the C<how> value when the C<SV>
964was initially made magical.
965
d1b91892
AD
966=head2 Magic Virtual Tables
967
d1be9408 968The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
969C<MGVTBL>, which is a structure of function pointers and stands for
970"Magic Virtual Table" to handle the various operations that might be
971applied to that variable.
972
301cb7e8
DM
973The C<MGVTBL> has five (or sometimes eight) pointers to the following
974routine types:
d1b91892
AD
975
976 int (*svt_get)(SV* sv, MAGIC* mg);
977 int (*svt_set)(SV* sv, MAGIC* mg);
978 U32 (*svt_len)(SV* sv, MAGIC* mg);
979 int (*svt_clear)(SV* sv, MAGIC* mg);
980 int (*svt_free)(SV* sv, MAGIC* mg);
981
3468c7ea 982 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv, const char *name, I32 namlen);
301cb7e8
DM
983 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
984 int (*svt_local)(SV *nsv, MAGIC *mg);
985
986
06f6df17 987This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
SM
988currently 32 types. These different structures contain pointers to various
989routines that perform additional actions depending on which function is
990being called.
d1b91892
AD
991
992 Function pointer Action taken
993 ---------------- ------------
8b0711c3 994 svt_get Do something before the value of the SV is retrieved.
d1b91892
AD
995 svt_set Do something after the SV is assigned a value.
996 svt_len Report on the SV's length.
805b34a4 997 svt_clear Clear something the SV represents.
d1b91892
AD
998 svt_free Free any extra storage associated with the SV.
999
805b34a4
AB
1000 svt_copy copy tied variable magic to a tied element
1001 svt_dup duplicate a magic structure during thread cloning
1002 svt_local copy magic to local value during 'local'
301cb7e8 1003
d1b91892 1004For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1005to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1006
1007 { magic_get, magic_set, magic_len, 0, 0 }
1008
14befaf4
DM
1009Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1010if a get operation is being performed, the routine C<magic_get> is
1011called. All the various routines for the various magical types begin
1012with C<magic_>. NOTE: the magic routines are not considered part of
1013the Perl API, and may not be exported by the Perl library.
d1b91892 1014
301cb7e8
DM
1015The last three slots are a recent addition, and for source code
1016compatibility they are only checked for if one of the three flags
1017MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags. This means that most
1018code can continue declaring a vtable as a 5-element value. These three are
1019currently used exclusively by the threading code, and are highly subject
1020to change.
1021
d1b91892
AD
1022The current kinds of Magic Virtual Tables are:
1023
14befaf4 1024 mg_type
5633a428
NC
1025 (old-style char and macro) MGVTBL Type of magic
1026 -------------------------- ------ -------------
805b34a4
AB
1027 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1028 A PERL_MAGIC_overload vtbl_amagic %OVERLOAD hash
14befaf4 1029 a PERL_MAGIC_overload_elem vtbl_amagicelem %OVERLOAD hash element
805b34a4
AB
1030 c PERL_MAGIC_overload_table (none) Holds overload table (AMT)
1031 on stash
1032 B PERL_MAGIC_bm vtbl_bm Boyer-Moore (fast string search)
1033 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1034 (@+ and @- vars)
1035 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1036 element
1037 E PERL_MAGIC_env vtbl_env %ENV hash
1038 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1039 f PERL_MAGIC_fm vtbl_fm Formline ('compiled' format)
1040 g PERL_MAGIC_regex_global vtbl_mglob m//g target / study()ed string
f747ebd6 1041 H PERL_MAGIC_hints vtbl_hints %^H hash
805b34a4
AB
1042 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1043 I PERL_MAGIC_isa vtbl_isa @ISA array
1044 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1045 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1046 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1047 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename element
805b34a4
AB
1048 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale collate transformation
1049 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1050 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1051 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1052 r PERL_MAGIC_qr vtbl_qr precompiled qr// regex
1053 S PERL_MAGIC_sig vtbl_sig %SIG hash
1054 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1055 t PERL_MAGIC_taint vtbl_taint Taintedness
1056 U PERL_MAGIC_uvar vtbl_uvar Available for use by extensions
1057 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1058 V PERL_MAGIC_vstring (none) v-string scalars
1059 w PERL_MAGIC_utf8 vtbl_utf8 UTF-8 length+offset cache
1060 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1061 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1062 variable / smart parameter
1063 vivification
1064 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1065 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1066 < PERL_MAGIC_backref vtbl_backref back pointer to a weak ref
1067 ~ PERL_MAGIC_ext (none) Available for use by extensions
1068 : PERL_MAGIC_symtab (none) hash used as symbol table
1069 % PERL_MAGIC_rhash (none) hash used as restricted hash
1070 @ PERL_MAGIC_arylen_p vtbl_arylen_p pointer to $#a from @a
0cbee0a4 1071
d1b91892 1072
68dc0745 1073When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1074uppercase letter is typically used to represent some kind of composite type
1075(a list or a hash), and the lowercase letter is used to represent an element
1076of that composite type. Some internals code makes use of this case
1077relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1078
1079The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1080specifically for use by extensions and will not be used by perl itself.
1081Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1082to variables (typically objects). This is especially useful because
1083there is no way for normal perl code to corrupt this private information
1084(unlike using extra elements of a hash object).
1085
1086Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1087C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1088C<mg_ptr> field points to a C<ufuncs> structure:
1089
1090 struct ufuncs {
a9402793
AB
1091 I32 (*uf_val)(pTHX_ IV, SV*);
1092 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1093 IV uf_index;
1094 };
1095
1096When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1097function will be called with C<uf_index> as the first arg and a pointer to
1098the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1099magic is shown below. Note that the ufuncs structure is copied by
1100sv_magic, so you can safely allocate it on the stack.
1101
1102 void
1103 Umagic(sv)
1104 SV *sv;
1105 PREINIT:
1106 struct ufuncs uf;
1107 CODE:
1108 uf.uf_val = &my_get_fn;
1109 uf.uf_set = &my_set_fn;
1110 uf.uf_index = 0;
14befaf4 1111 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1112
1e73acc8
AS
1113Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1114
1115For hashes there is a specialized hook that gives control over hash
1116keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1117if the "set" function in the C<ufuncs> structure is NULL. The hook
1118is activated whenever the hash is accessed with a key specified as
1119an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1120C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1121through the functions without the C<..._ent> suffix circumvents the
1122hook. See L<Hash::Util::Fieldhash/Guts> for a detailed description.
1123
14befaf4
DM
1124Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1125or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1126extra care to avoid conflict. Typically only using the magic on
1127objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1128For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1129C<MGVTBL>, even if all its fields will be C<0>, so that individual
1130C<MAGIC> pointers can be identified as a particular kind of magic
1131using their C<mg_virtual> field.
1132
1133 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1134
1135 MAGIC *mg;
1136 for (mg = SvMAGIC(sv); mg; mg = mg->mg_moremagic) {
1137 if (mg->mg_type == PERL_MAGIC_ext && mg->mg_virtual == &my_vtbl) {
1138 /* this is really ours, not another module's PERL_MAGIC_ext */
1139 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1140 }
1141 }
5f05dabc 1142
ef50df4b
GS
1143Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1144earlier do B<not> invoke 'set' magic on their targets. This must
1145be done by the user either by calling the C<SvSETMAGIC()> macro after
1146calling these functions, or by using one of the C<sv_set*_mg()> or
1147C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1148C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1149obtained from external sources in functions that don't handle magic.
4a4eefd0 1150See L<perlapi> for a description of these functions.
189b2af5
GS
1151For example, calls to the C<sv_cat*()> functions typically need to be
1152followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1153since their implementation handles 'get' magic.
1154
d1b91892
AD
1155=head2 Finding Magic
1156
1157 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
1158
1159This routine returns a pointer to the C<MAGIC> structure stored in the SV.
1160If the SV does not have that magical feature, C<NULL> is returned. Also,
54310121 1161if the SV is not of type SVt_PVMG, Perl may core dump.
d1b91892 1162
08105a92 1163 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1164
1165This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1166field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1167the mg_type field is changed to be the lowercase letter.
a0d0e21e 1168
04343c6d
GS
1169=head2 Understanding the Magic of Tied Hashes and Arrays
1170
14befaf4
DM
1171Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1172magic type.
9edb2b46
GS
1173
1174WARNING: As of the 5.004 release, proper usage of the array and hash
1175access functions requires understanding a few caveats. Some
1176of these caveats are actually considered bugs in the API, to be fixed
1177in later releases, and are bracketed with [MAYCHANGE] below. If
1178you find yourself actually applying such information in this section, be
1179aware that the behavior may change in the future, umm, without warning.
04343c6d 1180
1526ead6 1181The perl tie function associates a variable with an object that implements
9a68f1db 1182the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1183tie function from an XSUB, you must mimic this behaviour. The code below
1184carries out the necessary steps - firstly it creates a new hash, and then
1185creates a second hash which it blesses into the class which will implement
1186the tie methods. Lastly it ties the two hashes together, and returns a
1187reference to the new tied hash. Note that the code below does NOT call the
1188TIEHASH method in the MyTie class -
1189see L<Calling Perl Routines from within C Programs> for details on how
1190to do this.
1191
1192 SV*
1193 mytie()
1194 PREINIT:
1195 HV *hash;
1196 HV *stash;
1197 SV *tie;
1198 CODE:
1199 hash = newHV();
1200 tie = newRV_noinc((SV*)newHV());
da51bb9b 1201 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1202 sv_bless(tie, stash);
899e16d0 1203 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1204 RETVAL = newRV_noinc(hash);
1205 OUTPUT:
1206 RETVAL
1207
04343c6d
GS
1208The C<av_store> function, when given a tied array argument, merely
1209copies the magic of the array onto the value to be "stored", using
1210C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1211actually need to be stored in the array. [MAYCHANGE] After a call to
1212C<av_store> on a tied array, the caller will usually need to call
1213C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1214TIEARRAY object. If C<av_store> did return NULL, a call to
1215C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1216leak. [/MAYCHANGE]
04343c6d
GS
1217
1218The previous paragraph is applicable verbatim to tied hash access using the
1219C<hv_store> and C<hv_store_ent> functions as well.
1220
1221C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1222C<hv_fetch_ent> actually return an undefined mortal value whose magic
1223has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1224need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1225need to call C<mg_get()> on the returned value in order to actually invoke
1226the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1227you may also call C<mg_set()> on the return value after possibly assigning
1228a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1229method on the TIE object. [/MAYCHANGE]
04343c6d 1230
9edb2b46 1231[MAYCHANGE]
04343c6d
GS
1232In other words, the array or hash fetch/store functions don't really
1233fetch and store actual values in the case of tied arrays and hashes. They
1234merely call C<mg_copy> to attach magic to the values that were meant to be
1235"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1236do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1237the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1238and hashes.
1239
1240Currently (as of perl version 5.004), use of the hash and array access
1241functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1242"normal" hashes and arrays, or on their tied variants. The API may be
1243changed to provide more transparent access to both tied and normal data
1244types in future versions.
1245[/MAYCHANGE]
04343c6d
GS
1246
1247You would do well to understand that the TIEARRAY and TIEHASH interfaces
1248are mere sugar to invoke some perl method calls while using the uniform hash
1249and array syntax. The use of this sugar imposes some overhead (typically
1250about two to four extra opcodes per FETCH/STORE operation, in addition to
1251the creation of all the mortal variables required to invoke the methods).
1252This overhead will be comparatively small if the TIE methods are themselves
1253substantial, but if they are only a few statements long, the overhead
1254will not be insignificant.
1255
d1c897a1
IZ
1256=head2 Localizing changes
1257
1258Perl has a very handy construction
1259
1260 {
1261 local $var = 2;
1262 ...
1263 }
1264
1265This construction is I<approximately> equivalent to
1266
1267 {
1268 my $oldvar = $var;
1269 $var = 2;
1270 ...
1271 $var = $oldvar;
1272 }
1273
1274The biggest difference is that the first construction would
1275reinstate the initial value of $var, irrespective of how control exits
9a68f1db 1276the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1277more efficient as well.
1278
1279There is a way to achieve a similar task from C via Perl API: create a
1280I<pseudo-block>, and arrange for some changes to be automatically
1281undone at the end of it, either explicit, or via a non-local exit (via
1282die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1283C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1284Such a construct may be created specially for some important localized
1285task, or an existing one (like boundaries of enclosing Perl
1286subroutine/block, or an existing pair for freeing TMPs) may be
1287used. (In the second case the overhead of additional localization must
1288be almost negligible.) Note that any XSUB is automatically enclosed in
1289an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1290
1291Inside such a I<pseudo-block> the following service is available:
1292
13a2d996 1293=over 4
d1c897a1
IZ
1294
1295=item C<SAVEINT(int i)>
1296
1297=item C<SAVEIV(IV i)>
1298
1299=item C<SAVEI32(I32 i)>
1300
1301=item C<SAVELONG(long i)>
1302
1303These macros arrange things to restore the value of integer variable
1304C<i> at the end of enclosing I<pseudo-block>.
1305
1306=item C<SAVESPTR(s)>
1307
1308=item C<SAVEPPTR(p)>
1309
1310These macros arrange things to restore the value of pointers C<s> and
1311C<p>. C<s> must be a pointer of a type which survives conversion to
1312C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1313and back.
1314
1315=item C<SAVEFREESV(SV *sv)>
1316
1317The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1318I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1319mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1320extends the lifetime of C<sv> until the beginning of the next statement,
1321C<SAVEFREESV> extends it until the end of the enclosing scope. These
1322lifetimes can be wildly different.
1323
1324Also compare C<SAVEMORTALIZESV>.
1325
1326=item C<SAVEMORTALIZESV(SV *sv)>
1327
1328Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1329scope instead of decrementing its reference count. This usually has the
1330effect of keeping C<sv> alive until the statement that called the currently
1331live scope has finished executing.
d1c897a1
IZ
1332
1333=item C<SAVEFREEOP(OP *op)>
1334
1335The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1336
1337=item C<SAVEFREEPV(p)>
1338
1339The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1340end of I<pseudo-block>.
1341
1342=item C<SAVECLEARSV(SV *sv)>
1343
1344Clears a slot in the current scratchpad which corresponds to C<sv> at
1345the end of I<pseudo-block>.
1346
1347=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1348
1349The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1350string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1351short-lived storage, the corresponding string may be reallocated like
1352this:
1353
9cde0e7f 1354 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1355
c76ac1ee 1356=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1357
1358At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1359only argument C<p>.
1360
1361=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1362
1363At the end of I<pseudo-block> the function C<f> is called with the
1364implicit context argument (if any), and C<p>.
d1c897a1
IZ
1365
1366=item C<SAVESTACK_POS()>
1367
1368The current offset on the Perl internal stack (cf. C<SP>) is restored
1369at the end of I<pseudo-block>.
1370
1371=back
1372
1373The following API list contains functions, thus one needs to
1374provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1375or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1376function takes C<int *>.
1377
13a2d996 1378=over 4
d1c897a1
IZ
1379
1380=item C<SV* save_scalar(GV *gv)>
1381
1382Equivalent to Perl code C<local $gv>.
1383
1384=item C<AV* save_ary(GV *gv)>
1385
1386=item C<HV* save_hash(GV *gv)>
1387
1388Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1389
1390=item C<void save_item(SV *item)>
1391
1392Duplicates the current value of C<SV>, on the exit from the current
1393C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
038fcae3
SB
1394using the stored value. It doesn't handle magic. Use C<save_scalar> if
1395magic is affected.
d1c897a1
IZ
1396
1397=item C<void save_list(SV **sarg, I32 maxsarg)>
1398
1399A variant of C<save_item> which takes multiple arguments via an array
1400C<sarg> of C<SV*> of length C<maxsarg>.
1401
1402=item C<SV* save_svref(SV **sptr)>
1403
d1be9408 1404Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1405
1406=item C<void save_aptr(AV **aptr)>
1407
1408=item C<void save_hptr(HV **hptr)>
1409
1410Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1411
1412=back
1413
1414The C<Alias> module implements localization of the basic types within the
1415I<caller's scope>. People who are interested in how to localize things in
1416the containing scope should take a look there too.
1417
0a753a76 1418=head1 Subroutines
a0d0e21e 1419
68dc0745 1420=head2 XSUBs and the Argument Stack
5f05dabc
PP
1421
1422The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1423An XSUB routine will have a stack that contains the arguments from the Perl
1424program, and a way to map from the Perl data structures to a C equivalent.
1425
1426The stack arguments are accessible through the C<ST(n)> macro, which returns
1427the C<n>'th stack argument. Argument 0 is the first argument passed in the
1428Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1429an C<SV*> is used.
1430
1431Most of the time, output from the C routine can be handled through use of
1432the RETVAL and OUTPUT directives. However, there are some cases where the
1433argument stack is not already long enough to handle all the return values.
1434An example is the POSIX tzname() call, which takes no arguments, but returns
1435two, the local time zone's standard and summer time abbreviations.
1436
1437To handle this situation, the PPCODE directive is used and the stack is
1438extended using the macro:
1439
924508f0 1440 EXTEND(SP, num);
5f05dabc 1441
924508f0
GS
1442where C<SP> is the macro that represents the local copy of the stack pointer,
1443and C<num> is the number of elements the stack should be extended by.
5f05dabc 1444
00aadd71 1445Now that there is room on the stack, values can be pushed on it using C<PUSHs>
06f6df17 1446macro. The pushed values will often need to be "mortal" (See
d82b684c 1447L</Reference Counts and Mortality>):
5f05dabc 1448
00aadd71 1449 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1450 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1451 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1452 PUSHs(sv_2mortal(newSVpv("Some String",0)))
5f05dabc
PP
1453
1454And now the Perl program calling C<tzname>, the two values will be assigned
1455as in:
1456
1457 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1458
1459An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1460to use the macro:
5f05dabc 1461
5f05dabc
PP
1462 XPUSHs(SV*)
1463
00aadd71 1464This macro automatically adjust the stack for you, if needed. Thus, you
5f05dabc 1465do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1466
1467Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1468C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1469For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1470C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc
PP
1471
1472For more information, consult L<perlxs> and L<perlxstut>.
1473
1474=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1475
1476There are four routines that can be used to call a Perl subroutine from
1477within a C program. These four are:
1478
954c1994
GS
1479 I32 call_sv(SV*, I32);
1480 I32 call_pv(const char*, I32);
1481 I32 call_method(const char*, I32);
1482 I32 call_argv(const char*, I32, register char**);
a0d0e21e 1483
954c1994 1484The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1485contains either the name of the Perl subroutine to be called, or a
1486reference to the subroutine. The second argument consists of flags
1487that control the context in which the subroutine is called, whether
1488or not the subroutine is being passed arguments, how errors should be
1489trapped, and how to treat return values.
a0d0e21e
LW
1490
1491All four routines return the number of arguments that the subroutine returned
1492on the Perl stack.
1493
9a68f1db 1494These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1495but those names are now deprecated; macros of the same name are provided for
1496compatibility.
1497
1498When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1499must manipulate the Perl stack. These include the following macros and
1500functions:
a0d0e21e
LW
1501
1502 dSP
924508f0 1503 SP
a0d0e21e
LW
1504 PUSHMARK()
1505 PUTBACK
1506 SPAGAIN
1507 ENTER
1508 SAVETMPS
1509 FREETMPS
1510 LEAVE
1511 XPUSH*()
cb1a09d0 1512 POP*()
a0d0e21e 1513
5f05dabc
PP
1514For a detailed description of calling conventions from C to Perl,
1515consult L<perlcall>.
a0d0e21e 1516
5f05dabc 1517=head2 Memory Allocation
a0d0e21e 1518
06f6df17
RGS
1519=head3 Allocation
1520
86058a2d
GS
1521All memory meant to be used with the Perl API functions should be manipulated
1522using the macros described in this section. The macros provide the necessary
1523transparency between differences in the actual malloc implementation that is
1524used within perl.
1525
1526It is suggested that you enable the version of malloc that is distributed
5f05dabc 1527with Perl. It keeps pools of various sizes of unallocated memory in
07fa94a1
JO
1528order to satisfy allocation requests more quickly. However, on some
1529platforms, it may cause spurious malloc or free errors.
d1b91892 1530
06f6df17
RGS
1531The following three macros are used to initially allocate memory :
1532
9f653bb5
SH
1533 Newx(pointer, number, type);
1534 Newxc(pointer, number, type, cast);
1535 Newxz(pointer, number, type);
d1b91892 1536
9f653bb5 1537The first argument C<pointer> should be the name of a variable that will
5f05dabc 1538point to the newly allocated memory.
d1b91892 1539
9f653bb5 1540The second and third arguments C<number> and C<type> specify how many of
d1b91892 1541the specified type of data structure should be allocated. The argument
9f653bb5 1542C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
d1b91892
AD
1543should be used if the C<pointer> argument is different from the C<type>
1544argument.
1545
9f653bb5 1546Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
d1b91892
AD
1547to zero out all the newly allocated memory.
1548
06f6df17
RGS
1549=head3 Reallocation
1550
d1b91892
AD
1551 Renew(pointer, number, type);
1552 Renewc(pointer, number, type, cast);
1553 Safefree(pointer)
1554
1555These three macros are used to change a memory buffer size or to free a
1556piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1557match those of C<New> and C<Newc> with the exception of not needing the
1558"magic cookie" argument.
1559
06f6df17
RGS
1560=head3 Moving
1561
d1b91892
AD
1562 Move(source, dest, number, type);
1563 Copy(source, dest, number, type);
1564 Zero(dest, number, type);
1565
1566These three macros are used to move, copy, or zero out previously allocated
1567memory. The C<source> and C<dest> arguments point to the source and
1568destination starting points. Perl will move, copy, or zero out C<number>
1569instances of the size of the C<type> data structure (using the C<sizeof>
1570function).
a0d0e21e 1571
5f05dabc 1572=head2 PerlIO
ce3d39e2 1573
5f05dabc
PP
1574The most recent development releases of Perl has been experimenting with
1575removing Perl's dependency on the "normal" standard I/O suite and allowing
1576other stdio implementations to be used. This involves creating a new
1577abstraction layer that then calls whichever implementation of stdio Perl
68dc0745 1578was compiled with. All XSUBs should now use the functions in the PerlIO
5f05dabc
PP
1579abstraction layer and not make any assumptions about what kind of stdio
1580is being used.
1581
1582For a complete description of the PerlIO abstraction, consult L<perlapio>.
1583
8ebc5c01 1584=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1585
1586A lot of opcodes (this is an elementary operation in the internal perl
1587stack machine) put an SV* on the stack. However, as an optimization
1588the corresponding SV is (usually) not recreated each time. The opcodes
1589reuse specially assigned SVs (I<target>s) which are (as a corollary)
1590not constantly freed/created.
1591
0a753a76 1592Each of the targets is created only once (but see
ce3d39e2
IZ
1593L<Scratchpads and recursion> below), and when an opcode needs to put
1594an integer, a double, or a string on stack, it just sets the
1595corresponding parts of its I<target> and puts the I<target> on stack.
1596
1597The macro to put this target on stack is C<PUSHTARG>, and it is
1598directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1599others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1600
1bd1c0d5
SC
1601Because the target is reused, you must be careful when pushing multiple
1602values on the stack. The following code will not do what you think:
1603
1604 XPUSHi(10);
1605 XPUSHi(20);
1606
1607This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1608the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1609At the end of the operation, the stack does not contain the values 10
1610and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1611to 20.
1bd1c0d5 1612
d82b684c
SH
1613If you need to push multiple different values then you should either use
1614the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1615none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1616SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1617will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1618this a little easier to achieve by creating a new mortal for you (via
1619C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1620in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1621Thus, instead of writing this to "fix" the example above:
1622
1623 XPUSHs(sv_2mortal(newSViv(10)))
1624 XPUSHs(sv_2mortal(newSViv(20)))
1625
1626you can simply write:
1627
1628 mXPUSHi(10)
1629 mXPUSHi(20)
1630
1631On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1632need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1633macros can make use of the local variable C<TARG>. See also C<dTARGET>
1634and C<dXSTARG>.
1bd1c0d5 1635
8ebc5c01 1636=head2 Scratchpads
ce3d39e2 1637
54310121 1638The question remains on when the SVs which are I<target>s for opcodes
ac036724 1639are created. The answer is that they are created when the current
1640unit--a subroutine or a file (for opcodes for statements outside of
1641subroutines)--is compiled. During this time a special anonymous Perl
1642array is created, which is called a scratchpad for the current unit.
ce3d39e2 1643
54310121 1644A scratchpad keeps SVs which are lexicals for the current unit and are
ce3d39e2
IZ
1645targets for opcodes. One can deduce that an SV lives on a scratchpad
1646by looking on its flags: lexicals have C<SVs_PADMY> set, and
1647I<target>s have C<SVs_PADTMP> set.
1648
54310121
PP
1649The correspondence between OPs and I<target>s is not 1-to-1. Different
1650OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1651would not conflict with the expected life of the temporary.
1652
2ae324a7 1653=head2 Scratchpads and recursion
ce3d39e2
IZ
1654
1655In fact it is not 100% true that a compiled unit contains a pointer to
1656the scratchpad AV. In fact it contains a pointer to an AV of
1657(initially) one element, and this element is the scratchpad AV. Why do
1658we need an extra level of indirection?
1659
9a68f1db 1660The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2
IZ
1661these can create several execution pointers going into the same
1662subroutine. For the subroutine-child not write over the temporaries
1663for the subroutine-parent (lifespan of which covers the call to the
1664child), the parent and the child should have different
1665scratchpads. (I<And> the lexicals should be separate anyway!)
1666
5f05dabc
PP
1667So each subroutine is born with an array of scratchpads (of length 1).
1668On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1669depth of the recursion is not more than the length of this array, and
1670if it is, new scratchpad is created and pushed into the array.
1671
1672The I<target>s on this scratchpad are C<undef>s, but they are already
1673marked with correct flags.
1674
0a753a76
PP
1675=head1 Compiled code
1676
1677=head2 Code tree
1678
1679Here we describe the internal form your code is converted to by
1680Perl. Start with a simple example:
1681
1682 $a = $b + $c;
1683
1684This is converted to a tree similar to this one:
1685
1686 assign-to
1687 / \
1688 + $a
1689 / \
1690 $b $c
1691
7b8d334a 1692(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1693parsed your code, but has nothing to do with the execution order.
1694There is an additional "thread" going through the nodes of the tree
1695which shows the order of execution of the nodes. In our simplified
1696example above it looks like:
1697
1698 $b ---> $c ---> + ---> $a ---> assign-to
1699
1700But with the actual compile tree for C<$a = $b + $c> it is different:
1701some nodes I<optimized away>. As a corollary, though the actual tree
1702contains more nodes than our simplified example, the execution order
1703is the same as in our example.
1704
1705=head2 Examining the tree
1706
06f6df17
RGS
1707If you have your perl compiled for debugging (usually done with
1708C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1709compiled tree by specifying C<-Dx> on the Perl command line. The
1710output takes several lines per node, and for C<$b+$c> it looks like
1711this:
1712
1713 5 TYPE = add ===> 6
1714 TARG = 1
1715 FLAGS = (SCALAR,KIDS)
1716 {
1717 TYPE = null ===> (4)
1718 (was rv2sv)
1719 FLAGS = (SCALAR,KIDS)
1720 {
1721 3 TYPE = gvsv ===> 4
1722 FLAGS = (SCALAR)
1723 GV = main::b
1724 }
1725 }
1726 {
1727 TYPE = null ===> (5)
1728 (was rv2sv)
1729 FLAGS = (SCALAR,KIDS)
1730 {
1731 4 TYPE = gvsv ===> 5
1732 FLAGS = (SCALAR)
1733 GV = main::c
1734 }
1735 }
1736
1737This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1738not optimized away (one per number in the left column). The immediate
1739children of the given node correspond to C<{}> pairs on the same level
1740of indentation, thus this listing corresponds to the tree:
1741
1742 add
1743 / \
1744 null null
1745 | |
1746 gvsv gvsv
1747
1748The execution order is indicated by C<===E<gt>> marks, thus it is C<3
17494 5 6> (node C<6> is not included into above listing), i.e.,
1750C<gvsv gvsv add whatever>.
1751
9afa14e3
SC
1752Each of these nodes represents an op, a fundamental operation inside the
1753Perl core. The code which implements each operation can be found in the
1754F<pp*.c> files; the function which implements the op with type C<gvsv>
1755is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1756different numbers of children: C<add> is a binary operator, as one would
1757expect, and so has two children. To accommodate the various different
1758numbers of children, there are various types of op data structure, and
1759they link together in different ways.
1760
1761The simplest type of op structure is C<OP>: this has no children. Unary
1762operators, C<UNOP>s, have one child, and this is pointed to by the
1763C<op_first> field. Binary operators (C<BINOP>s) have not only an
1764C<op_first> field but also an C<op_last> field. The most complex type of
1765op is a C<LISTOP>, which has any number of children. In this case, the
1766first child is pointed to by C<op_first> and the last child by
1767C<op_last>. The children in between can be found by iteratively
1768following the C<op_sibling> pointer from the first child to the last.
1769
1770There are also two other op types: a C<PMOP> holds a regular expression,
1771and has no children, and a C<LOOP> may or may not have children. If the
1772C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1773complicate matters, if a C<UNOP> is actually a C<null> op after
1774optimization (see L</Compile pass 2: context propagation>) it will still
1775have children in accordance with its former type.
1776
06f6df17
RGS
1777Another way to examine the tree is to use a compiler back-end module, such
1778as L<B::Concise>.
1779
0a753a76
PP
1780=head2 Compile pass 1: check routines
1781
8870b5c7
GS
1782The tree is created by the compiler while I<yacc> code feeds it
1783the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
1784the first pass of perl compilation.
1785
1786What makes this pass interesting for perl developers is that some
1787optimization may be performed on this pass. This is optimization by
8870b5c7 1788so-called "check routines". The correspondence between node names
0a753a76
PP
1789and corresponding check routines is described in F<opcode.pl> (do not
1790forget to run C<make regen_headers> if you modify this file).
1791
1792A check routine is called when the node is fully constructed except
7b8d334a 1793for the execution-order thread. Since at this time there are no
0a753a76
PP
1794back-links to the currently constructed node, one can do most any
1795operation to the top-level node, including freeing it and/or creating
1796new nodes above/below it.
1797
1798The check routine returns the node which should be inserted into the
1799tree (if the top-level node was not modified, check routine returns
1800its argument).
1801
1802By convention, check routines have names C<ck_*>. They are usually
1803called from C<new*OP> subroutines (or C<convert>) (which in turn are
1804called from F<perly.y>).
1805
1806=head2 Compile pass 1a: constant folding
1807
1808Immediately after the check routine is called the returned node is
1809checked for being compile-time executable. If it is (the value is
1810judged to be constant) it is immediately executed, and a I<constant>
1811node with the "return value" of the corresponding subtree is
1812substituted instead. The subtree is deleted.
1813
1814If constant folding was not performed, the execution-order thread is
1815created.
1816
1817=head2 Compile pass 2: context propagation
1818
1819When a context for a part of compile tree is known, it is propagated
a3cb178b 1820down through the tree. At this time the context can have 5 values
0a753a76
PP
1821(instead of 2 for runtime context): void, boolean, scalar, list, and
1822lvalue. In contrast with the pass 1 this pass is processed from top
1823to bottom: a node's context determines the context for its children.
1824
1825Additional context-dependent optimizations are performed at this time.
1826Since at this moment the compile tree contains back-references (via
1827"thread" pointers), nodes cannot be free()d now. To allow
1828optimized-away nodes at this stage, such nodes are null()ified instead
1829of free()ing (i.e. their type is changed to OP_NULL).
1830
1831=head2 Compile pass 3: peephole optimization
1832
1833After the compile tree for a subroutine (or for an C<eval> or a file)
1834is created, an additional pass over the code is performed. This pass
1835is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
1836additional complications for conditionals). Optimizations performed
1837at this stage are subject to the same restrictions as in the pass 2.
1838
1839Peephole optimizations are done by calling the function pointed to
1840by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1841calls the function pointed to by the global variable C<PL_rpeepp>.
1842By default, that performs some basic op fixups and optimisations along
1843the execution-order op chain, and recursively calls C<PL_rpeepp> for
1844each side chain of ops (resulting from conditionals). Extensions may
1845provide additional optimisations or fixups, hooking into either the
1846per-subroutine or recursive stage, like this:
1847
1848 static peep_t prev_peepp;
1849 static void my_peep(pTHX_ OP *o)
1850 {
1851 /* custom per-subroutine optimisation goes here */
1852 prev_peepp(o);
1853 /* custom per-subroutine optimisation may also go here */
1854 }
1855 BOOT:
1856 prev_peepp = PL_peepp;
1857 PL_peepp = my_peep;
1858
1859 static peep_t prev_rpeepp;
1860 static void my_rpeep(pTHX_ OP *o)
1861 {
1862 OP *orig_o = o;
1863 for(; o; o = o->op_next) {
1864 /* custom per-op optimisation goes here */
1865 }
1866 prev_rpeepp(orig_o);
1867 }
1868 BOOT:
1869 prev_rpeepp = PL_rpeepp;
1870 PL_rpeepp = my_rpeep;
0a753a76 1871
1ba7f851
PJ
1872=head2 Pluggable runops
1873
1874The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
1875functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1876with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1877control over the execution of the compile tree it is possible to provide
1878your own runops function.
1ba7f851
PJ
1879
1880It's probably best to copy one of the existing runops functions and
1881change it to suit your needs. Then, in the BOOT section of your XS
1882file, add the line:
1883
1884 PL_runops = my_runops;
1885
1886This function should be as efficient as possible to keep your programs
1887running as fast as possible.
1888
fd85fad2
BM
1889=head2 Compile-time scope hooks
1890
1891As of perl 5.14 it is possible to hook into the compile-time lexical
1892scope mechanism using C<Perl_blockhook_register>. This is used like
1893this:
1894
1895 STATIC void my_start_hook(pTHX_ int full);
1896 STATIC BHK my_hooks;
1897
1898 BOOT:
a88d97bf 1899 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
1900 Perl_blockhook_register(aTHX_ &my_hooks);
1901
1902This will arrange to have C<my_start_hook> called at the start of
1903compiling every lexical scope. The available hooks are:
1904
1905=over 4
1906
a88d97bf 1907=item C<void bhk_start(pTHX_ int full)>
fd85fad2
BM
1908
1909This is called just after starting a new lexical scope. Note that Perl
1910code like
1911
1912 if ($x) { ... }
1913
1914creates two scopes: the first starts at the C<(> and has C<full == 1>,
1915the second starts at the C<{> and has C<full == 0>. Both end at the
1916C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
1917pushed onto the save stack by this hook will be popped just before the
1918scope ends (between the C<pre_> and C<post_end> hooks, in fact).
1919
a88d97bf 1920=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
1921
1922This is called at the end of a lexical scope, just before unwinding the
1923stack. I<o> is the root of the optree representing the scope; it is a
1924double pointer so you can replace the OP if you need to.
1925
a88d97bf 1926=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
1927
1928This is called at the end of a lexical scope, just after unwinding the
1929stack. I<o> is as above. Note that it is possible for calls to C<pre_>
1930and C<post_end> to nest, if there is something on the save stack that
1931calls string eval.
1932
a88d97bf 1933=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
1934
1935This is called just before starting to compile an C<eval STRING>, C<do
1936FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
1937OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
1938C<OP_DOFILE> or C<OP_REQUIRE>.
1939
1940=back
1941
1942Once you have your hook functions, you need a C<BHK> structure to put
1943them in. It's best to allocate it statically, since there is no way to
1944free it once it's registered. The function pointers should be inserted
1945into this structure using the C<BhkENTRY_set> macro, which will also set
1946flags indicating which entries are valid. If you do need to allocate
1947your C<BHK> dynamically for some reason, be sure to zero it before you
1948start.
1949
1950Once registered, there is no mechanism to switch these hooks off, so if
1951that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
1952is probably the best way, so the effect is lexically scoped; however it
1953is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
1954temporarily switch entries on and off. You should also be aware that
1955generally speaking at least one scope will have opened before your
1956extension is loaded, so you will see some C<pre/post_end> pairs that
1957didn't have a matching C<start>.
fd85fad2 1958
9afa14e3
SC
1959=head1 Examining internal data structures with the C<dump> functions
1960
1961To aid debugging, the source file F<dump.c> contains a number of
1962functions which produce formatted output of internal data structures.
1963
1964The most commonly used of these functions is C<Perl_sv_dump>; it's used
1965for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
1966C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 1967module should already be familiar with its format.
9afa14e3
SC
1968
1969C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 1970derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
1971C<Perl_dump_eval> will dump the main root of the code being evaluated,
1972exactly like C<-Dx>.
1973
1974Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
1975op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
1976subroutines in a package like so: (Thankfully, these are all xsubs, so
1977there is no op tree)
1978
1979 (gdb) print Perl_dump_packsubs(PL_defstash)
1980
1981 SUB attributes::bootstrap = (xsub 0x811fedc 0)
1982
1983 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
1984
1985 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
1986
1987 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
1988
1989 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
1990
1991and C<Perl_dump_all>, which dumps all the subroutines in the stash and
1992the op tree of the main root.
1993
954c1994 1994=head1 How multiple interpreters and concurrency are supported
ee072b34 1995
ee072b34
GS
1996=head2 Background and PERL_IMPLICIT_CONTEXT
1997
1998The Perl interpreter can be regarded as a closed box: it has an API
1999for feeding it code or otherwise making it do things, but it also has
2000functions for its own use. This smells a lot like an object, and
2001there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2002interpreters, with one interpreter represented either as a C structure,
2003or inside a thread-specific structure. These structures contain all
2004the context, the state of that interpreter.
2005
7b52221d
RGS
2006One macro controls the major Perl build flavor: MULTIPLICITY. The
2007MULTIPLICITY build has a C structure that packages all the interpreter
2008state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2009normally defined, and enables the support for passing in a "hidden" first
2010argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2011multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2012to the macro USE_ITHREADS.)
54aff467 2013
27da23d5
JH
2014Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2015PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2016former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2017internal variables of Perl to be wrapped inside a single global struct,
2018struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2019the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2020one step further, there is still a single struct (allocated in main()
2021either from heap or from stack) but there are no global data symbols
2022pointing to it. In either case the global struct should be initialised
2023as the very first thing in main() using Perl_init_global_struct() and
2024correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2025please see F<miniperlmain.c> for usage details. You may also need
2026to use C<dVAR> in your coding to "declare the global variables"
2027when you are using them. dTHX does this for you automatically.
2028
bc028b6b
JH
2029To see whether you have non-const data you can use a BSD-compatible C<nm>:
2030
2031 nm libperl.a | grep -v ' [TURtr] '
2032
2033If this displays any C<D> or C<d> symbols, you have non-const data.
2034
27da23d5
JH
2035For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2036doesn't actually hide all symbols inside a big global struct: some
2037PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2038then hides everything (see how the PERLIO_FUNCS_DECL is used).
2039
54aff467 2040All this obviously requires a way for the Perl internal functions to be
acfe0abc 2041either subroutines taking some kind of structure as the first
ee072b34 2042argument, or subroutines taking nothing as the first argument. To
acfe0abc 2043enable these two very different ways of building the interpreter,
ee072b34
GS
2044the Perl source (as it does in so many other situations) makes heavy
2045use of macros and subroutine naming conventions.
2046
54aff467 2047First problem: deciding which functions will be public API functions and
00aadd71 2048which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2049(think "S" for "secret" or "static"). All other functions begin with
2050"Perl_", but just because a function begins with "Perl_" does not mean it is
00aadd71
NIS
2051part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
2052function is part of the API is to find its entry in L<perlapi>.
2053If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2054think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2055L<perlbug> explaining why you think it should be.
ee072b34
GS
2056
2057Second problem: there must be a syntax so that the same subroutine
2058declarations and calls can pass a structure as their first argument,
2059or pass nothing. To solve this, the subroutines are named and
2060declared in a particular way. Here's a typical start of a static
2061function used within the Perl guts:
2062
2063 STATIC void
2064 S_incline(pTHX_ char *s)
2065
acfe0abc
GS
2066STATIC becomes "static" in C, and may be #define'd to nothing in some
2067configurations in future.
ee072b34 2068
651a3225
GS
2069A public function (i.e. part of the internal API, but not necessarily
2070sanctioned for use in extensions) begins like this:
ee072b34
GS
2071
2072 void
2307c6d0 2073 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2074
0147cd53 2075C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2076details of the interpreter's context. THX stands for "thread", "this",
2077or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2078The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2079or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2080their variants.
ee072b34 2081
a7486cbb
JH
2082When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2083first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2084in the pTHX_ macro indicates that the macro expansion needs a comma
2085after the context argument because other arguments follow it. If
2086PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2087subroutine is not prototyped to take the extra argument. The form of the
2088macro without the trailing underscore is used when there are no additional
ee072b34
GS
2089explicit arguments.
2090
54aff467 2091When a core function calls another, it must pass the context. This
2307c6d0 2092is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2093something like this:
2094
2307c6d0
SB
2095 #ifdef PERL_IMPLICIT_CONTEXT
2096 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2097 /* can't do this for vararg functions, see below */
2307c6d0
SB
2098 #else
2099 #define sv_setiv Perl_sv_setiv
2100 #endif
ee072b34
GS
2101
2102This works well, and means that XS authors can gleefully write:
2103
2307c6d0 2104 sv_setiv(foo, bar);
ee072b34
GS
2105
2106and still have it work under all the modes Perl could have been
2107compiled with.
2108
ee072b34
GS
2109This doesn't work so cleanly for varargs functions, though, as macros
2110imply that the number of arguments is known in advance. Instead we
2111either need to spell them out fully, passing C<aTHX_> as the first
2112argument (the Perl core tends to do this with functions like
2113Perl_warner), or use a context-free version.
2114
2115The context-free version of Perl_warner is called
2116Perl_warner_nocontext, and does not take the extra argument. Instead
2117it does dTHX; to get the context from thread-local storage. We
2118C<#define warner Perl_warner_nocontext> so that extensions get source
2119compatibility at the expense of performance. (Passing an arg is
2120cheaper than grabbing it from thread-local storage.)
2121
acfe0abc 2122You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2123Those are strictly for use within the core. Extensions and embedders
2124need only be aware of [pad]THX.
2125
a7486cbb
JH
2126=head2 So what happened to dTHR?
2127
2128C<dTHR> was introduced in perl 5.005 to support the older thread model.
2129The older thread model now uses the C<THX> mechanism to pass context
2130pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2131later still have it for backward source compatibility, but it is defined
2132to be a no-op.
2133
ee072b34
GS
2134=head2 How do I use all this in extensions?
2135
2136When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2137any functions in the Perl API will need to pass the initial context
2138argument somehow. The kicker is that you will need to write it in
2139such a way that the extension still compiles when Perl hasn't been
2140built with PERL_IMPLICIT_CONTEXT enabled.
2141
2142There are three ways to do this. First, the easy but inefficient way,
2143which is also the default, in order to maintain source compatibility
0147cd53 2144with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2145and aTHX_ macros to call a function that will return the context.
2146Thus, something like:
2147
2307c6d0 2148 sv_setiv(sv, num);
ee072b34 2149
4375e838 2150in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2151in effect:
ee072b34 2152
2307c6d0 2153 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2154
54aff467 2155or to this otherwise:
ee072b34 2156
2307c6d0 2157 Perl_sv_setiv(sv, num);
ee072b34
GS
2158
2159You have to do nothing new in your extension to get this; since
2fa86c13 2160the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2161work.
2162
2163The second, more efficient way is to use the following template for
2164your Foo.xs:
2165
c52f9dcd
JH
2166 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2167 #include "EXTERN.h"
2168 #include "perl.h"
2169 #include "XSUB.h"
ee072b34 2170
fd061412 2171 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2172
fd061412 2173 STATIC void
c52f9dcd
JH
2174 my_private_function(int arg1, int arg2)
2175 {
2176 dTHX; /* fetch context */
2177 ... call many Perl API functions ...
2178 }
ee072b34
GS
2179
2180 [... etc ...]
2181
c52f9dcd 2182 MODULE = Foo PACKAGE = Foo
ee072b34 2183
c52f9dcd 2184 /* typical XSUB */
ee072b34 2185
c52f9dcd
JH
2186 void
2187 my_xsub(arg)
2188 int arg
2189 CODE:
2190 my_private_function(arg, 10);
ee072b34
GS
2191
2192Note that the only two changes from the normal way of writing an
2193extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2194including the Perl headers, followed by a C<dTHX;> declaration at
2195the start of every function that will call the Perl API. (You'll
2196know which functions need this, because the C compiler will complain
2197that there's an undeclared identifier in those functions.) No changes
2198are needed for the XSUBs themselves, because the XS() macro is
2199correctly defined to pass in the implicit context if needed.
2200
2201The third, even more efficient way is to ape how it is done within
2202the Perl guts:
2203
2204
c52f9dcd
JH
2205 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2206 #include "EXTERN.h"
2207 #include "perl.h"
2208 #include "XSUB.h"
ee072b34
GS
2209
2210 /* pTHX_ only needed for functions that call Perl API */
fd061412 2211 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2212
fd061412 2213 STATIC void
c52f9dcd
JH
2214 my_private_function(pTHX_ int arg1, int arg2)
2215 {
2216 /* dTHX; not needed here, because THX is an argument */
2217 ... call Perl API functions ...
2218 }
ee072b34
GS
2219
2220 [... etc ...]
2221
c52f9dcd 2222 MODULE = Foo PACKAGE = Foo
ee072b34 2223
c52f9dcd 2224 /* typical XSUB */
ee072b34 2225
c52f9dcd
JH
2226 void
2227 my_xsub(arg)
2228 int arg
2229 CODE:
2230 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2231
2232This implementation never has to fetch the context using a function
2233call, since it is always passed as an extra argument. Depending on
2234your needs for simplicity or efficiency, you may mix the previous
2235two approaches freely.
2236
651a3225
GS
2237Never add a comma after C<pTHX> yourself--always use the form of the
2238macro with the underscore for functions that take explicit arguments,
2239or the form without the argument for functions with no explicit arguments.
ee072b34 2240
27da23d5
JH
2241If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2242definition is needed if the Perl global variables (see F<perlvars.h>
2243or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2244used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2245the need for C<dVAR> only with the said compile-time define, because
2246otherwise the Perl global variables are visible as-is.
2247
a7486cbb
JH
2248=head2 Should I do anything special if I call perl from multiple threads?
2249
2250If you create interpreters in one thread and then proceed to call them in
2251another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2252initialized correctly in each of those threads.
2253
2254The C<perl_alloc> and C<perl_clone> API functions will automatically set
2255the TLS slot to the interpreter they created, so that there is no need to do
2256anything special if the interpreter is always accessed in the same thread that
2257created it, and that thread did not create or call any other interpreters
2258afterwards. If that is not the case, you have to set the TLS slot of the
2259thread before calling any functions in the Perl API on that particular
2260interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2261thread as the first thing you do:
2262
2263 /* do this before doing anything else with some_perl */
2264 PERL_SET_CONTEXT(some_perl);
2265
2266 ... other Perl API calls on some_perl go here ...
2267
ee072b34
GS
2268=head2 Future Plans and PERL_IMPLICIT_SYS
2269
2270Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2271that the interpreter knows about itself and pass it around, so too are
2272there plans to allow the interpreter to bundle up everything it knows
2273about the environment it's running on. This is enabled with the
7b52221d
RGS
2274PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2275Windows.
ee072b34
GS
2276
2277This allows the ability to provide an extra pointer (called the "host"
2278environment) for all the system calls. This makes it possible for
2279all the system stuff to maintain their own state, broken down into
2280seven C structures. These are thin wrappers around the usual system
0147cd53 2281calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2282more ambitious host (like the one that would do fork() emulation) all
2283the extra work needed to pretend that different interpreters are
2284actually different "processes", would be done here.
2285
2286The Perl engine/interpreter and the host are orthogonal entities.
2287There could be one or more interpreters in a process, and one or
2288more "hosts", with free association between them.
2289
a422fd2d
SC
2290=head1 Internal Functions
2291
2292All of Perl's internal functions which will be exposed to the outside
06f6df17 2293world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d
SC
2294functions or functions used in a program in which Perl is embedded.
2295Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2296static functions start with C<S_>.)
a422fd2d 2297
0972ecdf
DM
2298Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2299either with or without the C<Perl_> prefix, thanks to a bunch of defines
2300that live in F<embed.h>. Note that extension code should I<not> set
2301C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2302breakage of the XS in each new perl release.
2303
2304The file F<embed.h> is generated automatically from
dc9b1d22
MHM
2305F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2306header files for the internal functions, generates the documentation
2307and a lot of other bits and pieces. It's important that when you add
2308a new function to the core or change an existing one, you change the
2309data in the table in F<embed.fnc> as well. Here's a sample entry from
2310that table:
a422fd2d
SC
2311
2312 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2313
2314The second column is the return type, the third column the name. Columns
2315after that are the arguments. The first column is a set of flags:
2316
2317=over 3
2318
2319=item A
2320
1aa6ea50
JC
2321This function is a part of the public API. All such functions should also
2322have 'd', very few do not.
a422fd2d
SC
2323
2324=item p
2325
1aa6ea50
JC
2326This function has a C<Perl_> prefix; i.e. it is defined as
2327C<Perl_av_fetch>.
a422fd2d
SC
2328
2329=item d
2330
2331This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2332look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2333
2334=back
2335
2336Other available flags are:
2337
2338=over 3
2339
2340=item s
2341
1aa6ea50
JC
2342This is a static function and is defined as C<STATIC S_whatever>, and
2343usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2344
2345=item n
2346
1aa6ea50
JC
2347This does not need a interpreter context, so the definition has no
2348C<pTHX>, and it follows that callers don't use C<aTHX>. (See
a422fd2d
SC
2349L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
2350
2351=item r
2352
2353This function never returns; C<croak>, C<exit> and friends.
2354
2355=item f
2356
2357This function takes a variable number of arguments, C<printf> style.
2358The argument list should end with C<...>, like this:
2359
2360 Afprd |void |croak |const char* pat|...
2361
a7486cbb 2362=item M
a422fd2d 2363
00aadd71 2364This function is part of the experimental development API, and may change
a422fd2d
SC
2365or disappear without notice.
2366
2367=item o
2368
2369This function should not have a compatibility macro to define, say,
2370C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2371
a422fd2d
SC
2372=item x
2373
2374This function isn't exported out of the Perl core.
2375
dc9b1d22
MHM
2376=item m
2377
2378This is implemented as a macro.
2379
2380=item X
2381
2382This function is explicitly exported.
2383
2384=item E
2385
2386This function is visible to extensions included in the Perl core.
2387
2388=item b
2389
2390Binary backward compatibility; this function is a macro but also has
2391a C<Perl_> implementation (which is exported).
2392
1aa6ea50
JC
2393=item others
2394
2395See the comments at the top of C<embed.fnc> for others.
2396
a422fd2d
SC
2397=back
2398
dc9b1d22
MHM
2399If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2400C<make regen_headers> to force a rebuild of F<embed.h> and other
2401auto-generated files.
a422fd2d 2402
6b4667fc 2403=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2404
6b4667fc
A
2405If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2406formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2407following macros for portability
9dd9db0b 2408
c52f9dcd
JH
2409 IVdf IV in decimal
2410 UVuf UV in decimal
2411 UVof UV in octal
2412 UVxf UV in hexadecimal
2413 NVef NV %e-like
2414 NVff NV %f-like
2415 NVgf NV %g-like
9dd9db0b 2416
6b4667fc
A
2417These will take care of 64-bit integers and long doubles.
2418For example:
2419
c52f9dcd 2420 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2421
2422The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2423
8908e76d
JH
2424If you are printing addresses of pointers, use UVxf combined
2425with PTR2UV(), do not use %lx or %p.
2426
2427=head2 Pointer-To-Integer and Integer-To-Pointer
2428
2429Because pointer size does not necessarily equal integer size,
2430use the follow macros to do it right.
2431
c52f9dcd
JH
2432 PTR2UV(pointer)
2433 PTR2IV(pointer)
2434 PTR2NV(pointer)
2435 INT2PTR(pointertotype, integer)
8908e76d
JH
2436
2437For example:
2438
c52f9dcd
JH
2439 IV iv = ...;
2440 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2441
2442and
2443
c52f9dcd
JH
2444 AV *av = ...;
2445 UV uv = PTR2UV(av);
8908e76d 2446
0ca3a874
MHM
2447=head2 Exception Handling
2448
9b5c3821
MHM
2449There are a couple of macros to do very basic exception handling in XS
2450modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
2451be able to use these macros:
2452
2453 #define NO_XSLOCKS
2454 #include "XSUB.h"
2455
2456You can use these macros if you call code that may croak, but you need
2457to do some cleanup before giving control back to Perl. For example:
0ca3a874 2458
d7f8936a 2459 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2460
2461 XCPT_TRY_START {
2462 code_that_may_croak();
2463 } XCPT_TRY_END
2464
2465 XCPT_CATCH
2466 {
2467 /* do cleanup here */
2468 XCPT_RETHROW;
2469 }
2470
2471Note that you always have to rethrow an exception that has been
2472caught. Using these macros, it is not possible to just catch the
2473exception and ignore it. If you have to ignore the exception, you
2474have to use the C<call_*> function.
2475
2476The advantage of using the above macros is that you don't have
2477to setup an extra function for C<call_*>, and that using these
2478macros is faster than using C<call_*>.
2479
a422fd2d
SC
2480=head2 Source Documentation
2481
2482There's an effort going on to document the internal functions and
2483automatically produce reference manuals from them - L<perlapi> is one
2484such manual which details all the functions which are available to XS
2485writers. L<perlintern> is the autogenerated manual for the functions
2486which are not part of the API and are supposedly for internal use only.
2487
2488Source documentation is created by putting POD comments into the C
2489source, like this:
2490
2491 /*
2492 =for apidoc sv_setiv
2493
2494 Copies an integer into the given SV. Does not handle 'set' magic. See
2495 C<sv_setiv_mg>.
2496
2497 =cut
2498 */
2499
2500Please try and supply some documentation if you add functions to the
2501Perl core.
2502
0d098d33
MHM
2503=head2 Backwards compatibility
2504
2505The Perl API changes over time. New functions are added or the interfaces
2506of existing functions are changed. The C<Devel::PPPort> module tries to
2507provide compatibility code for some of these changes, so XS writers don't
2508have to code it themselves when supporting multiple versions of Perl.
2509
2510C<Devel::PPPort> generates a C header file F<ppport.h> that can also
2511be run as a Perl script. To generate F<ppport.h>, run:
2512
2513 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2514
2515Besides checking existing XS code, the script can also be used to retrieve
2516compatibility information for various API calls using the C<--api-info>
2517command line switch. For example:
2518
2519 % perl ppport.h --api-info=sv_magicext
2520
2521For details, see C<perldoc ppport.h>.
2522
a422fd2d
SC
2523=head1 Unicode Support
2524
2525Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2526writers to understand this support and make sure that the code they
2527write does not corrupt Unicode data.
2528
2529=head2 What B<is> Unicode, anyway?
2530
2531In the olden, less enlightened times, we all used to use ASCII. Most of
2532us did, anyway. The big problem with ASCII is that it's American. Well,
2533no, that's not actually the problem; the problem is that it's not
2534particularly useful for people who don't use the Roman alphabet. What
2535used to happen was that particular languages would stick their own
2536alphabet in the upper range of the sequence, between 128 and 255. Of
2537course, we then ended up with plenty of variants that weren't quite
2538ASCII, and the whole point of it being a standard was lost.
2539
2540Worse still, if you've got a language like Chinese or
2541Japanese that has hundreds or thousands of characters, then you really
2542can't fit them into a mere 256, so they had to forget about ASCII
2543altogether, and build their own systems using pairs of numbers to refer
2544to one character.
2545
2546To fix this, some people formed Unicode, Inc. and
2547produced a new character set containing all the characters you can
2548possibly think of and more. There are several ways of representing these
1e54db1a 2549characters, and the one Perl uses is called UTF-8. UTF-8 uses
2575c402
JW
2550a variable number of bytes to represent a character. You can learn more
2551about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2552
1e54db1a 2553=head2 How can I recognise a UTF-8 string?
a422fd2d 2554
1e54db1a
JH
2555You can't. This is because UTF-8 data is stored in bytes just like
2556non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d
SC
2557capital E with a grave accent, is represented by the two bytes
2558C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2559has that byte sequence as well. So you can't tell just by looking - this
2560is what makes Unicode input an interesting problem.
2561
2575c402
JW
2562In general, you either have to know what you're dealing with, or you
2563have to guess. The API function C<is_utf8_string> can help; it'll tell
2564you if a string contains only valid UTF-8 characters. However, it can't
2565do the work for you. On a character-by-character basis, C<is_utf8_char>
2566will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2567
1e54db1a 2568=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2569
1e54db1a 2570As mentioned above, UTF-8 uses a variable number of bytes to store a
2575c402
JW
2571character. Characters with values 0...127 are stored in one byte, just
2572like good ol' ASCII. Character 128 is stored as C<v194.128>; this
a31a806a 2573continues up to character 191, which is C<v194.191>. Now we've run out of
a422fd2d
SC
2574bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2575so it goes on, moving to three bytes at character 2048.
2576
1e54db1a 2577Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2578how long the first character in it is with the C<UTF8SKIP> macro:
2579
2580 char *utf = "\305\233\340\240\201";
2581 I32 len;
2582
2583 len = UTF8SKIP(utf); /* len is 2 here */
2584 utf += len;
2585 len = UTF8SKIP(utf); /* len is 3 here */
2586
1e54db1a 2587Another way to skip over characters in a UTF-8 string is to use
a422fd2d
SC
2588C<utf8_hop>, which takes a string and a number of characters to skip
2589over. You're on your own about bounds checking, though, so don't use it
2590lightly.
2591
1e54db1a 2592All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2593so you can test if you need to do something special with this
2594character like this (the UTF8_IS_INVARIANT() is a macro that tests
2595whether the byte can be encoded as a single byte even in UTF-8):
a422fd2d 2596
3a2263fe
RGS
2597 U8 *utf;
2598 UV uv; /* Note: a UV, not a U8, not a char */
a422fd2d 2599
3a2263fe 2600 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2601 /* Must treat this as UTF-8 */
a422fd2d
SC
2602 uv = utf8_to_uv(utf);
2603 else
2604 /* OK to treat this character as a byte */
2605 uv = *utf;
2606
2607You can also see in that example that we use C<utf8_to_uv> to get the
2608value of the character; the inverse function C<uv_to_utf8> is available
1e54db1a 2609for putting a UV into UTF-8:
a422fd2d 2610
3a2263fe 2611 if (!UTF8_IS_INVARIANT(uv))
a422fd2d
SC
2612 /* Must treat this as UTF8 */
2613 utf8 = uv_to_utf8(utf8, uv);
2614 else
2615 /* OK to treat this character as a byte */
2616 *utf8++ = uv;
2617
2618You B<must> convert characters to UVs using the above functions if
1e54db1a
JH
2619you're ever in a situation where you have to match UTF-8 and non-UTF-8
2620characters. You may not skip over UTF-8 characters in this case. If you
2621do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2622for instance, if your UTF-8 string contains C<v196.172>, and you skip
2623that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2624So don't do that!
2625
1e54db1a 2626=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2627
2628Currently, Perl deals with Unicode strings and non-Unicode strings
2575c402
JW
2629slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2630string is internally encoded as UTF-8. Without it, the byte value is the
2631codepoint number and vice versa (in other words, the string is encoded
e1b711da
KW
2632as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
2633semantics). You can check and manipulate this flag with the
2575c402 2634following macros:
a422fd2d
SC
2635
2636 SvUTF8(sv)
2637 SvUTF8_on(sv)
2638 SvUTF8_off(sv)
2639
2640This flag has an important effect on Perl's treatment of the string: if
2641Unicode data is not properly distinguished, regular expressions,
2642C<length>, C<substr> and other string handling operations will have
2643undesirable results.
2644
2645The problem comes when you have, for instance, a string that isn't
2575c402 2646flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
1e54db1a 2647especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2648
2649Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2650need be sure you don't accidentally knock it off while you're
2651manipulating SVs. More specifically, you cannot expect to do this:
2652
2653 SV *sv;
2654 SV *nsv;
2655 STRLEN len;
2656 char *p;
2657
2658 p = SvPV(sv, len);
2659 frobnicate(p);
2660 nsv = newSVpvn(p, len);
2661
2662The C<char*> string does not tell you the whole story, and you can't
2663copy or reconstruct an SV just by copying the string value. Check if the
2575c402 2664old SV has the UTF8 flag set, and act accordingly:
a422fd2d
SC
2665
2666 p = SvPV(sv, len);
2667 frobnicate(p);
2668 nsv = newSVpvn(p, len);
2669 if (SvUTF8(sv))
2670 SvUTF8_on(nsv);
2671
2672In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2673not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2674appropriately.
2675
3a2263fe 2676Since just passing an SV to an XS function and copying the data of
2575c402 2677the SV is not enough to copy the UTF8 flags, even less right is just
3a2263fe
RGS
2678passing a C<char *> to an XS function.
2679
1e54db1a 2680=head2 How do I convert a string to UTF-8?
a422fd2d 2681
2575c402
JW
2682If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
2683one of the strings to UTF-8. If you've got an SV, the easiest way to do
2684this is:
a422fd2d
SC
2685
2686 sv_utf8_upgrade(sv);
2687
2688However, you must not do this, for example:
2689
2690 if (!SvUTF8(left))
2691 sv_utf8_upgrade(left);
2692
2693If you do this in a binary operator, you will actually change one of the
b1866b2d 2694strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2695by the end user, it can cause problems in deficient code.
a422fd2d 2696
1e54db1a 2697Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
a422fd2d 2698string argument. This is useful for having the data available for
b1866b2d 2699comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2700C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2701the string contains any characters above 255 that can't be represented
2702in a single byte.
2703
2704=head2 Is there anything else I need to know?
2705
2706Not really. Just remember these things:
2707
2708=over 3
2709
2710=item *
2711
1e54db1a
JH
2712There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2713is UTF-8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2714something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2715it's not - if you pass on the PV to somewhere, pass on the flag too.
2716
2717=item *
2718
1e54db1a 2719If a string is UTF-8, B<always> use C<utf8_to_uv> to get at the value,
3a2263fe 2720unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2721
2722=item *
2723
1e54db1a 2724When writing a character C<uv> to a UTF-8 string, B<always> use
3a2263fe
RGS
2725C<uv_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2726you can use C<*s = uv>.
a422fd2d
SC
2727
2728=item *
2729
1e54db1a 2730Mixing UTF-8 and non-UTF-8 strings is tricky. Use C<bytes_to_utf8> to get
2bbc8d55 2731a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
2732
2733=back
2734
53e06cf0
SC
2735=head1 Custom Operators
2736
9a68f1db 2737Custom operator support is a new experimental feature that allows you to
53e06cf0
SC
2738define your own ops. This is primarily to allow the building of
2739interpreters for other languages in the Perl core, but it also allows
2740optimizations through the creation of "macro-ops" (ops which perform the
2741functions of multiple ops which are usually executed together, such as
1aa6ea50 2742C<gvsv, gvsv, add>.)
53e06cf0 2743
b455bf3f 2744This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0
SC
2745core does not "know" anything special about this op type, and so it will
2746not be involved in any optimizations. This also means that you can
2747define your custom ops to be any op structure - unary, binary, list and
2748so on - you like.
2749
2750It's important to know what custom operators won't do for you. They
2751won't let you add new syntax to Perl, directly. They won't even let you
2752add new keywords, directly. In fact, they won't change the way Perl
2753compiles a program at all. You have to do those changes yourself, after
2754Perl has compiled the program. You do this either by manipulating the op
2755tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2756a custom peephole optimizer with the C<optimize> module.
2757
2758When you do this, you replace ordinary Perl ops with custom ops by
2759creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2760PP function. This should be defined in XS code, and should look like
2761the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2762takes the appropriate number of values from the stack, and you are
2763responsible for adding stack marks if necessary.
2764
2765You should also "register" your op with the Perl interpreter so that it
2766can produce sensible error and warning messages. Since it is possible to
2767have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2768Perl uses the value of C<< o->op_ppaddr >> as a key into the
2769C<PL_custom_op_descs> and C<PL_custom_op_names> hashes. This means you
2770need to enter a name and description for your op at the appropriate
2771place in the C<PL_custom_op_names> and C<PL_custom_op_descs> hashes.
2772
e7d4c058 2773C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 2774
954c1994 2775=head1 AUTHORS
e89caa19 2776
954c1994 2777Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
2778E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2779itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 2780
954c1994
GS
2781With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2782Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2783Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2784Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 2785
954c1994 2786=head1 SEE ALSO
cb1a09d0 2787
ba555bf5 2788L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>