This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Mention perl 5.8.2-RC1 in perlhist, as spotted by Abigail.
[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
b2f5ed49 129 SV *s;
ce2f5d8f
KA
130 STRLEN len;
131 char * ptr;
b2f5ed49 132 ptr = SvPV(s, len);
ce2f5d8f
KA
133 foo(ptr, len);
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
4929bf7b 194 SV* get_sv("package::varname", FALSE);
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
RGS
203The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
204Its address can be used whenever an C<SV*> is needed.
a9381218
MHM
205However, you have to be careful when using C<&PL_sv_undef> as a value in AVs
206or HVs (see L<AVs, HVs and undefined values>).
a0d0e21e 207
06f6df17
RGS
208There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
209boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
210addresses can be used whenever an C<SV*> is needed.
a0d0e21e 211
9cde0e7f 212Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
a0d0e21e
LW
213Take this code:
214
215 SV* sv = (SV*) 0;
216 if (I-am-to-return-a-real-value) {
217 sv = sv_2mortal(newSViv(42));
218 }
219 sv_setsv(ST(0), sv);
220
221This code tries to return a new SV (which contains the value 42) if it should
04343c6d 222return a real value, or undef otherwise. Instead it has returned a NULL
a0d0e21e 223pointer which, somewhere down the line, will cause a segmentation violation,
06f6df17
RGS
224bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
225first line and all will be well.
a0d0e21e
LW
226
227To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
3fe9a6f1 228call is not necessary (see L<Reference Counts and Mortality>).
a0d0e21e 229
94dde4fb
SC
230=head2 Offsets
231
232Perl provides the function C<sv_chop> to efficiently remove characters
233from the beginning of a string; you give it an SV and a pointer to
da75cd15 234somewhere inside the PV, and it discards everything before the
94dde4fb
SC
235pointer. The efficiency comes by means of a little hack: instead of
236actually removing the characters, C<sv_chop> sets the flag C<OOK>
237(offset OK) to signal to other functions that the offset hack is in
238effect, and it puts the number of bytes chopped off into the IV field
239of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
00aadd71 240many bytes, and adjusts C<SvCUR> and C<SvLEN>.
94dde4fb
SC
241
242Hence, at this point, the start of the buffer that we allocated lives
243at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
244into the middle of this allocated storage.
245
246This is best demonstrated by example:
247
248 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
249 SV = PVIV(0x8128450) at 0x81340f0
250 REFCNT = 1
251 FLAGS = (POK,OOK,pPOK)
252 IV = 1 (OFFSET)
253 PV = 0x8135781 ( "1" . ) "2345"\0
254 CUR = 4
255 LEN = 5
256
257Here the number of bytes chopped off (1) is put into IV, and
258C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
259portion of the string between the "real" and the "fake" beginnings is
260shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
261the fake beginning, not the real one.
262
fe854a6f 263Something similar to the offset hack is performed on AVs to enable
319cef53
SC
264efficient shifting and splicing off the beginning of the array; while
265C<AvARRAY> points to the first element in the array that is visible from
266Perl, C<AvALLOC> points to the real start of the C array. These are
267usually the same, but a C<shift> operation can be carried out by
268increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvLEN>.
269Again, the location of the real start of the C array only comes into
270play when freeing the array. See C<av_shift> in F<av.c>.
271
d1b91892 272=head2 What's Really Stored in an SV?
a0d0e21e
LW
273
274Recall that the usual method of determining the type of scalar you have is
5f05dabc 275to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 276usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
277macros will do the appropriate conversion of string to integer/double or
278integer/double to string.
279
280If you I<really> need to know if you have an integer, double, or string
281pointer in an SV, you can use the following three macros instead:
282
283 SvIOKp(SV*)
284 SvNOKp(SV*)
285 SvPOKp(SV*)
286
287These will tell you if you truly have an integer, double, or string pointer
d1b91892 288stored in your SV. The "p" stands for private.
a0d0e21e 289
9e9796d6
JH
290The are various ways in which the private and public flags may differ.
291For example, a tied SV may have a valid underlying value in the IV slot
292(so SvIOKp is true), but the data should be accessed via the FETCH
293routine rather than directly, so SvIOK is false. Another is when
294numeric conversion has occured and precision has been lost: only the
295private flag is set on 'lossy' values. So when an NV is converted to an
296IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
297
07fa94a1 298In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 299
54310121 300=head2 Working with AVs
a0d0e21e 301
07fa94a1
JO
302There are two ways to create and load an AV. The first method creates an
303empty AV:
a0d0e21e
LW
304
305 AV* newAV();
306
54310121 307The second method both creates the AV and initially populates it with SVs:
a0d0e21e
LW
308
309 AV* av_make(I32 num, SV **ptr);
310
5f05dabc 311The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 312AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 313
54310121 314Once the AV has been created, the following operations are possible on AVs:
a0d0e21e
LW
315
316 void av_push(AV*, SV*);
317 SV* av_pop(AV*);
318 SV* av_shift(AV*);
319 void av_unshift(AV*, I32 num);
320
321These should be familiar operations, with the exception of C<av_unshift>.
322This routine adds C<num> elements at the front of the array with the C<undef>
323value. You must then use C<av_store> (described below) to assign values
324to these new elements.
325
326Here are some other functions:
327
5f05dabc 328 I32 av_len(AV*);
a0d0e21e 329 SV** av_fetch(AV*, I32 key, I32 lval);
a0d0e21e 330 SV** av_store(AV*, I32 key, SV* val);
a0d0e21e 331
5f05dabc
PP
332The C<av_len> function returns the highest index value in array (just
333like $#array in Perl). If the array is empty, -1 is returned. The
334C<av_fetch> function returns the value at index C<key>, but if C<lval>
335is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
336The C<av_store> function stores the value C<val> at index C<key>, and does
337not increment the reference count of C<val>. Thus the caller is responsible
338for taking care of that, and if C<av_store> returns NULL, the caller will
339have to decrement the reference count to avoid a memory leak. Note that
340C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
341return value.
d1b91892 342
a0d0e21e 343 void av_clear(AV*);
a0d0e21e 344 void av_undef(AV*);
cb1a09d0 345 void av_extend(AV*, I32 key);
5f05dabc
PP
346
347The C<av_clear> function deletes all the elements in the AV* array, but
348does not actually delete the array itself. The C<av_undef> function will
349delete all the elements in the array plus the array itself. The
adc882cf
GS
350C<av_extend> function extends the array so that it contains at least C<key+1>
351elements. If C<key+1> is less than the currently allocated length of the array,
352then nothing is done.
a0d0e21e
LW
353
354If you know the name of an array variable, you can get a pointer to its AV
355by using the following:
356
4929bf7b 357 AV* get_av("package::varname", FALSE);
a0d0e21e
LW
358
359This returns NULL if the variable does not exist.
360
04343c6d
GS
361See L<Understanding the Magic of Tied Hashes and Arrays> for more
362information on how to use the array access functions on tied arrays.
363
54310121 364=head2 Working with HVs
a0d0e21e
LW
365
366To create an HV, you use the following routine:
367
368 HV* newHV();
369
54310121 370Once the HV has been created, the following operations are possible on HVs:
a0d0e21e 371
08105a92
GS
372 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
373 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 374
5f05dabc
PP
375The C<klen> parameter is the length of the key being passed in (Note that
376you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
377length of the key). The C<val> argument contains the SV pointer to the
54310121 378scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
379you want C<hv_store> to calculate it for you). The C<lval> parameter
380indicates whether this fetch is actually a part of a store operation, in
381which case a new undefined value will be added to the HV with the supplied
382key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 383
5f05dabc
PP
384Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
385C<SV*>. To access the scalar value, you must first dereference the return
386value. However, you should check to make sure that the return value is
387not NULL before dereferencing it.
a0d0e21e
LW
388
389These two functions check if a hash table entry exists, and deletes it.
390
08105a92
GS
391 bool hv_exists(HV*, const char* key, U32 klen);
392 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 393
5f05dabc
PP
394If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
395create and return a mortal copy of the deleted value.
396
a0d0e21e
LW
397And more miscellaneous functions:
398
399 void hv_clear(HV*);
a0d0e21e 400 void hv_undef(HV*);
5f05dabc
PP
401
402Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
403table but does not actually delete the hash table. The C<hv_undef> deletes
404both the entries and the hash table itself.
a0d0e21e 405
d1b91892
AD
406Perl keeps the actual data in linked list of structures with a typedef of HE.
407These contain the actual key and value pointers (plus extra administrative
408overhead). The key is a string pointer; the value is an C<SV*>. However,
409once you have an C<HE*>, to get the actual key and value, use the routines
410specified below.
411
a0d0e21e
LW
412 I32 hv_iterinit(HV*);
413 /* Prepares starting point to traverse hash table */
414 HE* hv_iternext(HV*);
415 /* Get the next entry, and return a pointer to a
416 structure that has both the key and value */
417 char* hv_iterkey(HE* entry, I32* retlen);
418 /* Get the key from an HE structure and also return
419 the length of the key string */
cb1a09d0 420 SV* hv_iterval(HV*, HE* entry);
d1be9408 421 /* Return an SV pointer to the value of the HE
a0d0e21e 422 structure */
cb1a09d0 423 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
424 /* This convenience routine combines hv_iternext,
425 hv_iterkey, and hv_iterval. The key and retlen
426 arguments are return values for the key and its
427 length. The value is returned in the SV* argument */
a0d0e21e
LW
428
429If you know the name of a hash variable, you can get a pointer to its HV
430by using the following:
431
4929bf7b 432 HV* get_hv("package::varname", FALSE);
a0d0e21e
LW
433
434This returns NULL if the variable does not exist.
435
8ebc5c01 436The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
a0d0e21e 437
a0d0e21e 438 hash = 0;
ab192400
GS
439 while (klen--)
440 hash = (hash * 33) + *key++;
87275199 441 hash = hash + (hash >> 5); /* after 5.6 */
ab192400 442
87275199 443The last step was added in version 5.6 to improve distribution of
ab192400 444lower bits in the resulting hash value.
a0d0e21e 445
04343c6d
GS
446See L<Understanding the Magic of Tied Hashes and Arrays> for more
447information on how to use the hash access functions on tied hashes.
448
1e422769
PP
449=head2 Hash API Extensions
450
451Beginning with version 5.004, the following functions are also supported:
452
453 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
454 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 455
1e422769
PP
456 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
457 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 458
1e422769
PP
459 SV* hv_iterkeysv (HE* entry);
460
461Note that these functions take C<SV*> keys, which simplifies writing
462of extension code that deals with hash structures. These functions
463also allow passing of C<SV*> keys to C<tie> functions without forcing
464you to stringify the keys (unlike the previous set of functions).
465
466They also return and accept whole hash entries (C<HE*>), making their
467use more efficient (since the hash number for a particular string
4a4eefd0
GS
468doesn't have to be recomputed every time). See L<perlapi> for detailed
469descriptions.
1e422769
PP
470
471The following macros must always be used to access the contents of hash
472entries. Note that the arguments to these macros must be simple
473variables, since they may get evaluated more than once. See
4a4eefd0 474L<perlapi> for detailed descriptions of these macros.
1e422769
PP
475
476 HePV(HE* he, STRLEN len)
477 HeVAL(HE* he)
478 HeHASH(HE* he)
479 HeSVKEY(HE* he)
480 HeSVKEY_force(HE* he)
481 HeSVKEY_set(HE* he, SV* sv)
482
483These two lower level macros are defined, but must only be used when
484dealing with keys that are not C<SV*>s:
485
486 HeKEY(HE* he)
487 HeKLEN(HE* he)
488
04343c6d
GS
489Note that both C<hv_store> and C<hv_store_ent> do not increment the
490reference count of the stored C<val>, which is the caller's responsibility.
491If these functions return a NULL value, the caller will usually have to
492decrement the reference count of C<val> to avoid a memory leak.
1e422769 493
a9381218
MHM
494=head2 AVs, HVs and undefined values
495
496Sometimes you have to store undefined values in AVs or HVs. Although
497this may be a rare case, it can be tricky. That's because you're
498used to using C<&PL_sv_undef> if you need an undefined SV.
499
500For example, intuition tells you that this XS code:
501
502 AV *av = newAV();
503 av_store( av, 0, &PL_sv_undef );
504
505is equivalent to this Perl code:
506
507 my @av;
508 $av[0] = undef;
509
510Unfortunately, this isn't true. AVs use C<&PL_sv_undef> as a marker
511for indicating that an array element has not yet been initialized.
512Thus, C<exists $av[0]> would be true for the above Perl code, but
513false for the array generated by the XS code.
514
515Other problems can occur when storing C<&PL_sv_undef> in HVs:
516
517 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
518
519This will indeed make the value C<undef>, but if you try to modify
520the value of C<key>, you'll get the following error:
521
522 Modification of non-creatable hash value attempted
523
524In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
525in restricted hashes. This caused such hash entries not to appear
526when iterating over the hash or when checking for the keys
527with the C<hv_exists> function.
528
529You can run into similar problems when you store C<&PL_sv_true> or
530C<&PL_sv_false> into AVs or HVs. Trying to modify such elements
531will give you the following error:
532
533 Modification of a read-only value attempted
534
535To make a long story short, you can use the special variables
536C<&PL_sv_undef>, C<&PL_sv_true> and C<&PL_sv_false> with AVs and
537HVs, but you have to make sure you know what you're doing.
538
539Generally, if you want to store an undefined value in an AV
540or HV, you should not use C<&PL_sv_undef>, but rather create a
541new undefined value using the C<newSV> function, for example:
542
543 av_store( av, 42, newSV(0) );
544 hv_store( hv, "foo", 3, newSV(0), 0 );
545
a0d0e21e
LW
546=head2 References
547
d1b91892
AD
548References are a special type of scalar that point to other data types
549(including references).
a0d0e21e 550
07fa94a1 551To create a reference, use either of the following functions:
a0d0e21e 552
5f05dabc
PP
553 SV* newRV_inc((SV*) thing);
554 SV* newRV_noinc((SV*) thing);
a0d0e21e 555
5f05dabc 556The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
557functions are identical except that C<newRV_inc> increments the reference
558count of the C<thing>, while C<newRV_noinc> does not. For historical
559reasons, C<newRV> is a synonym for C<newRV_inc>.
560
561Once you have a reference, you can use the following macro to dereference
562the reference:
a0d0e21e
LW
563
564 SvRV(SV*)
565
566then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 567C<AV*> or C<HV*>, if required.
a0d0e21e 568
d1b91892 569To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
570
571 SvROK(SV*)
572
07fa94a1
JO
573To discover what type of value the reference refers to, use the following
574macro and then check the return value.
d1b91892
AD
575
576 SvTYPE(SvRV(SV*))
577
578The most useful types that will be returned are:
579
580 SVt_IV Scalar
581 SVt_NV Scalar
582 SVt_PV Scalar
5f05dabc 583 SVt_RV Scalar
d1b91892
AD
584 SVt_PVAV Array
585 SVt_PVHV Hash
586 SVt_PVCV Code
5f05dabc
PP
587 SVt_PVGV Glob (possible a file handle)
588 SVt_PVMG Blessed or Magical Scalar
589
590 See the sv.h header file for more details.
d1b91892 591
cb1a09d0
AD
592=head2 Blessed References and Class Objects
593
06f6df17 594References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
595OO lexicon, an object is simply a reference that has been blessed into a
596package (or class). Once blessed, the programmer may now use the reference
597to access the various methods in the class.
598
599A reference can be blessed into a package with the following function:
600
601 SV* sv_bless(SV* sv, HV* stash);
602
06f6df17
RGS
603The C<sv> argument must be a reference value. The C<stash> argument
604specifies which class the reference will belong to. See
2ae324a7 605L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
606
607/* Still under construction */
608
609Upgrades rv to reference if not already one. Creates new SV for rv to
8ebc5c01
PP
610point to. If C<classname> is non-null, the SV is blessed into the specified
611class. SV is returned.
cb1a09d0 612
08105a92 613 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 614
e1c57cef 615Copies integer, unsigned integer or double into an SV whose reference is C<rv>. SV is blessed
8ebc5c01 616if C<classname> is non-null.
cb1a09d0 617
08105a92 618 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 619 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 620 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 621
5f05dabc 622Copies the pointer value (I<the address, not the string!>) into an SV whose
8ebc5c01 623reference is rv. SV is blessed if C<classname> is non-null.
cb1a09d0 624
08105a92 625 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
cb1a09d0 626
8ebc5c01
PP
627Copies string into an SV whose reference is C<rv>. Set length to 0 to let
628Perl calculate the string length. SV is blessed if C<classname> is non-null.
cb1a09d0 629
e65f3abd 630 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
cb1a09d0 631
9abd00ed
GS
632Tests whether the SV is blessed into the specified class. It does not
633check inheritance relationships.
634
08105a92 635 int sv_isa(SV* sv, const char* name);
9abd00ed
GS
636
637Tests whether the SV is a reference to a blessed object.
638
639 int sv_isobject(SV* sv);
640
641Tests whether the SV is derived from the specified class. SV can be either
642a reference to a blessed object or a string containing a class name. This
643is the function implementing the C<UNIVERSAL::isa> functionality.
644
08105a92 645 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 646
00aadd71 647To check if you've got an object derived from a specific class you have
9abd00ed
GS
648to write:
649
650 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 651
5f05dabc 652=head2 Creating New Variables
cb1a09d0 653
5f05dabc
PP
654To create a new Perl variable with an undef value which can be accessed from
655your Perl script, use the following routines, depending on the variable type.
cb1a09d0 656
4929bf7b
GS
657 SV* get_sv("package::varname", TRUE);
658 AV* get_av("package::varname", TRUE);
659 HV* get_hv("package::varname", TRUE);
cb1a09d0
AD
660
661Notice the use of TRUE as the second parameter. The new variable can now
662be set, using the routines appropriate to the data type.
663
5f05dabc
PP
664There are additional macros whose values may be bitwise OR'ed with the
665C<TRUE> argument to enable certain extra features. Those bits are:
cb1a09d0 666
9a68f1db
SB
667=over
668
669=item GV_ADDMULTI
670
671Marks the variable as multiply defined, thus preventing the:
672
673 Name <varname> used only once: possible typo
674
675warning.
676
9a68f1db
SB
677=item GV_ADDWARN
678
679Issues the warning:
680
681 Had to create <varname> unexpectedly
682
683if the variable did not exist before the function was called.
684
685=back
cb1a09d0 686
07fa94a1
JO
687If you do not specify a package name, the variable is created in the current
688package.
cb1a09d0 689
5f05dabc 690=head2 Reference Counts and Mortality
a0d0e21e 691
d1be9408 692Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 693AVs, or HVs (xV for short in the following) start their life with a
55497cff 694reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 695then it will be destroyed and its memory made available for reuse.
55497cff
PP
696
697This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
698undef'ed or the last variable holding a reference to it is changed or
699overwritten. At the internal level, however, reference counts can be
55497cff
PP
700manipulated with the following macros:
701
702 int SvREFCNT(SV* sv);
5f05dabc 703 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
704 void SvREFCNT_dec(SV* sv);
705
706However, there is one other function which manipulates the reference
07fa94a1
JO
707count of its argument. The C<newRV_inc> function, you will recall,
708creates a reference to the specified argument. As a side effect,
709it increments the argument's reference count. If this is not what
710you want, use C<newRV_noinc> instead.
711
712For example, imagine you want to return a reference from an XSUB function.
713Inside the XSUB routine, you create an SV which initially has a reference
714count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
715This returns the reference as a new SV, but the reference count of the
716SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
717return the reference from the XSUB routine and forget about the SV.
718But Perl hasn't! Whenever the returned reference is destroyed, the
719reference count of the original SV is decreased to one and nothing happens.
720The SV will hang around without any way to access it until Perl itself
721terminates. This is a memory leak.
5f05dabc
PP
722
723The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
724C<newRV_inc>. Then, if and when the last reference is destroyed,
725the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 726stopping any memory leak.
55497cff 727
5f05dabc 728There are some convenience functions available that can help with the
54310121 729destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
730An xV that is mortal has had its reference count marked to be decremented,
731but not actually decremented, until "a short time later". Generally the
732term "short time later" means a single Perl statement, such as a call to
54310121 733an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
734reference count decremented depends on two macros, SAVETMPS and FREETMPS.
735See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
736
737"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
738However, if you mortalize a variable twice, the reference count will
739later be decremented twice.
740
00aadd71
NIS
741"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
742For example an SV which is created just to pass a number to a called sub
06f6df17
RGS
743is made mortal to have it cleaned up automatically when it's popped off
744the stack. Similarly, results returned by XSUBs (which are pushed on the
745stack) are often made mortal.
a0d0e21e
LW
746
747To create a mortal variable, use the functions:
748
749 SV* sv_newmortal()
750 SV* sv_2mortal(SV*)
751 SV* sv_mortalcopy(SV*)
752
00aadd71 753The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
754SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
755third creates a mortal copy of an existing SV.
00aadd71 756Because C<sv_newmortal> gives the new SV no value,it must normally be given one
9a68f1db 757via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
758
759 SV *tmp = sv_newmortal();
760 sv_setiv(tmp, an_integer);
761
762As that is multiple C statements it is quite common so see this idiom instead:
763
764 SV *tmp = sv_2mortal(newSViv(an_integer));
765
766
767You should be careful about creating mortal variables. Strange things
768can happen if you make the same value mortal within multiple contexts,
769or if you make a variable mortal multiple times. Thinking of "Mortalization"
770as deferred C<SvREFCNT_dec> should help to minimize such problems.
771For example if you are passing an SV which you I<know> has high enough REFCNT
772to survive its use on the stack you need not do any mortalization.
773If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
774making a C<sv_mortalcopy> is safer.
a0d0e21e 775
54310121 776The mortal routines are not just for SVs -- AVs and HVs can be
faed5253 777made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 778C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 779
5f05dabc 780=head2 Stashes and Globs
a0d0e21e 781
06f6df17
RGS
782A B<stash> is a hash that contains all variables that are defined
783within a package. Each key of the stash is a symbol
aa689395
PP
784name (shared by all the different types of objects that have the same
785name), and each value in the hash table is a GV (Glob Value). This GV
786in turn contains references to the various objects of that name,
787including (but not limited to) the following:
cb1a09d0 788
a0d0e21e
LW
789 Scalar Value
790 Array Value
791 Hash Value
a3cb178b 792 I/O Handle
a0d0e21e
LW
793 Format
794 Subroutine
795
06f6df17
RGS
796There is a single stash called C<PL_defstash> that holds the items that exist
797in the C<main> package. To get at the items in other packages, append the
798string "::" to the package name. The items in the C<Foo> package are in
799the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
800in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 801
d1b91892 802To get the stash pointer for a particular package, use the function:
a0d0e21e 803
08105a92 804 HV* gv_stashpv(const char* name, I32 create)
a0d0e21e
LW
805 HV* gv_stashsv(SV*, I32 create)
806
807The first function takes a literal string, the second uses the string stored
d1b91892 808in the SV. Remember that a stash is just a hash table, so you get back an
cb1a09d0 809C<HV*>. The C<create> flag will create a new package if it is set.
a0d0e21e
LW
810
811The name that C<gv_stash*v> wants is the name of the package whose symbol table
812you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
813packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
814language itself.
a0d0e21e
LW
815
816Alternately, if you have an SV that is a blessed reference, you can find
817out the stash pointer by using:
818
819 HV* SvSTASH(SvRV(SV*));
820
821then use the following to get the package name itself:
822
823 char* HvNAME(HV* stash);
824
5f05dabc
PP
825If you need to bless or re-bless an object you can use the following
826function:
a0d0e21e
LW
827
828 SV* sv_bless(SV*, HV* stash)
829
830where the first argument, an C<SV*>, must be a reference, and the second
831argument is a stash. The returned C<SV*> can now be used in the same way
832as any other SV.
833
d1b91892
AD
834For more information on references and blessings, consult L<perlref>.
835
54310121 836=head2 Double-Typed SVs
0a753a76
PP
837
838Scalar variables normally contain only one type of value, an integer,
839double, pointer, or reference. Perl will automatically convert the
840actual scalar data from the stored type into the requested type.
841
842Some scalar variables contain more than one type of scalar data. For
843example, the variable C<$!> contains either the numeric value of C<errno>
844or its string equivalent from either C<strerror> or C<sys_errlist[]>.
845
846To force multiple data values into an SV, you must do two things: use the
847C<sv_set*v> routines to add the additional scalar type, then set a flag
848so that Perl will believe it contains more than one type of data. The
849four macros to set the flags are:
850
851 SvIOK_on
852 SvNOK_on
853 SvPOK_on
854 SvROK_on
855
856The particular macro you must use depends on which C<sv_set*v> routine
857you called first. This is because every C<sv_set*v> routine turns on
858only the bit for the particular type of data being set, and turns off
859all the rest.
860
861For example, to create a new Perl variable called "dberror" that contains
862both the numeric and descriptive string error values, you could use the
863following code:
864
865 extern int dberror;
866 extern char *dberror_list;
867
4929bf7b 868 SV* sv = get_sv("dberror", TRUE);
0a753a76
PP
869 sv_setiv(sv, (IV) dberror);
870 sv_setpv(sv, dberror_list[dberror]);
871 SvIOK_on(sv);
872
873If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
874macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
875
876=head2 Magic Variables
a0d0e21e 877
d1b91892
AD
878[This section still under construction. Ignore everything here. Post no
879bills. Everything not permitted is forbidden.]
880
d1b91892
AD
881Any SV may be magical, that is, it has special features that a normal
882SV does not have. These features are stored in the SV structure in a
5f05dabc 883linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
884
885 struct magic {
886 MAGIC* mg_moremagic;
887 MGVTBL* mg_virtual;
888 U16 mg_private;
889 char mg_type;
890 U8 mg_flags;
891 SV* mg_obj;
892 char* mg_ptr;
893 I32 mg_len;
894 };
895
896Note this is current as of patchlevel 0, and could change at any time.
897
898=head2 Assigning Magic
899
900Perl adds magic to an SV using the sv_magic function:
901
08105a92 902 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
903
904The C<sv> argument is a pointer to the SV that is to acquire a new magical
905feature.
906
907If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
645c22ef
DM
908convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
909to the beginning of the linked list of magical features. Any prior entry
910of the same type of magic is deleted. Note that this can be overridden,
911and multiple instances of the same type of magic can be associated with an
912SV.
d1b91892 913
54310121
PP
914The C<name> and C<namlen> arguments are used to associate a string with
915the magic, typically the name of a variable. C<namlen> is stored in the
9b5bb84f 916C<mg_len> field and if C<name> is non-null and C<namlen> E<gt>= 0 a malloc'd
d1b91892
AD
917copy of the name is stored in C<mg_ptr> field.
918
919The sv_magic function uses C<how> to determine which, if any, predefined
920"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 921See the L<Magic Virtual Tables> section below. The C<how> argument is also
14befaf4 922stored in the C<mg_type> field. The value of C<how> should be chosen
06f6df17 923from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 924these macros were added, Perl internals used to directly use character
14befaf4 925literals, so you may occasionally come across old code or documentation
75d0f26d 926referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
927
928The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
929structure. If it is not the same as the C<sv> argument, the reference
930count of the C<obj> object is incremented. If it is the same, or if
645c22ef 931the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 932then C<obj> is merely stored, without the reference count being incremented.
d1b91892 933
cb1a09d0
AD
934There is also a function to add magic to an C<HV>:
935
936 void hv_magic(HV *hv, GV *gv, int how);
937
938This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
939
940To remove the magic from an SV, call the function sv_unmagic:
941
942 void sv_unmagic(SV *sv, int type);
943
944The C<type> argument should be equal to the C<how> value when the C<SV>
945was initially made magical.
946
d1b91892
AD
947=head2 Magic Virtual Tables
948
d1be9408 949The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
950C<MGVTBL>, which is a structure of function pointers and stands for
951"Magic Virtual Table" to handle the various operations that might be
952applied to that variable.
953
954The C<MGVTBL> has five pointers to the following routine types:
955
956 int (*svt_get)(SV* sv, MAGIC* mg);
957 int (*svt_set)(SV* sv, MAGIC* mg);
958 U32 (*svt_len)(SV* sv, MAGIC* mg);
959 int (*svt_clear)(SV* sv, MAGIC* mg);
960 int (*svt_free)(SV* sv, MAGIC* mg);
961
06f6df17 962This MGVTBL structure is set at compile-time in F<perl.h> and there are
d1b91892
AD
963currently 19 types (or 21 with overloading turned on). These different
964structures contain pointers to various routines that perform additional
965actions depending on which function is being called.
966
967 Function pointer Action taken
968 ---------------- ------------
8b0711c3 969 svt_get Do something before the value of the SV is retrieved.
d1b91892
AD
970 svt_set Do something after the SV is assigned a value.
971 svt_len Report on the SV's length.
972 svt_clear Clear something the SV represents.
973 svt_free Free any extra storage associated with the SV.
974
975For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 976to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
977
978 { magic_get, magic_set, magic_len, 0, 0 }
979
14befaf4
DM
980Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
981if a get operation is being performed, the routine C<magic_get> is
982called. All the various routines for the various magical types begin
983with C<magic_>. NOTE: the magic routines are not considered part of
984the Perl API, and may not be exported by the Perl library.
d1b91892
AD
985
986The current kinds of Magic Virtual Tables are:
987
14befaf4
DM
988 mg_type
989 (old-style char and macro) MGVTBL Type of magic
990 -------------------------- ------ ----------------------------
991 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
992 A PERL_MAGIC_overload vtbl_amagic %OVERLOAD hash
993 a PERL_MAGIC_overload_elem vtbl_amagicelem %OVERLOAD hash element
994 c PERL_MAGIC_overload_table (none) Holds overload table (AMT)
995 on stash
996 B PERL_MAGIC_bm vtbl_bm Boyer-Moore (fast string search)
997 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
998 (@+ and @- vars)
999 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1000 element
1001 E PERL_MAGIC_env vtbl_env %ENV hash
1002 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1003 f PERL_MAGIC_fm vtbl_fm Formline ('compiled' format)
1004 g PERL_MAGIC_regex_global vtbl_mglob m//g target / study()ed string
1005 I PERL_MAGIC_isa vtbl_isa @ISA array
1006 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1007 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1008 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1009 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename element
1010 m PERL_MAGIC_mutex vtbl_mutex ???
645c22ef 1011 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale collate transformation
14befaf4
DM
1012 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1013 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1014 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1015 r PERL_MAGIC_qr vtbl_qr precompiled qr// regex
1016 S PERL_MAGIC_sig vtbl_sig %SIG hash
1017 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1018 t PERL_MAGIC_taint vtbl_taint Taintedness
1019 U PERL_MAGIC_uvar vtbl_uvar Available for use by extensions
1020 v PERL_MAGIC_vec vtbl_vec vec() lvalue
92f0c265 1021 V PERL_MAGIC_vstring (none) v-string scalars
836995da 1022 w PERL_MAGIC_utf8 vtbl_utf8 UTF-8 length+offset cache
14befaf4
DM
1023 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1024 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1025 variable / smart parameter
1026 vivification
1027 * PERL_MAGIC_glob vtbl_glob GV (typeglob)
1028 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1029 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1030 < PERL_MAGIC_backref vtbl_backref ???
1031 ~ PERL_MAGIC_ext (none) Available for use by extensions
d1b91892 1032
68dc0745 1033When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1034uppercase letter is typically used to represent some kind of composite type
1035(a list or a hash), and the lowercase letter is used to represent an element
1036of that composite type. Some internals code makes use of this case
1037relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1038
1039The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1040specifically for use by extensions and will not be used by perl itself.
1041Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1042to variables (typically objects). This is especially useful because
1043there is no way for normal perl code to corrupt this private information
1044(unlike using extra elements of a hash object).
1045
1046Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1047C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1048C<mg_ptr> field points to a C<ufuncs> structure:
1049
1050 struct ufuncs {
a9402793
AB
1051 I32 (*uf_val)(pTHX_ IV, SV*);
1052 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1053 IV uf_index;
1054 };
1055
1056When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1057function will be called with C<uf_index> as the first arg and a pointer to
1058the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1059magic is shown below. Note that the ufuncs structure is copied by
1060sv_magic, so you can safely allocate it on the stack.
1061
1062 void
1063 Umagic(sv)
1064 SV *sv;
1065 PREINIT:
1066 struct ufuncs uf;
1067 CODE:
1068 uf.uf_val = &my_get_fn;
1069 uf.uf_set = &my_set_fn;
1070 uf.uf_index = 0;
14befaf4 1071 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1072
14befaf4
DM
1073Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1074or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1075extra care to avoid conflict. Typically only using the magic on
1076objects blessed into the same class as the extension is sufficient.
1077For C<PERL_MAGIC_ext> magic, it may also be appropriate to add an I32
1078'signature' at the top of the private data area and check that.
5f05dabc 1079
ef50df4b
GS
1080Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1081earlier do B<not> invoke 'set' magic on their targets. This must
1082be done by the user either by calling the C<SvSETMAGIC()> macro after
1083calling these functions, or by using one of the C<sv_set*_mg()> or
1084C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1085C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1086obtained from external sources in functions that don't handle magic.
4a4eefd0 1087See L<perlapi> for a description of these functions.
189b2af5
GS
1088For example, calls to the C<sv_cat*()> functions typically need to be
1089followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1090since their implementation handles 'get' magic.
1091
d1b91892
AD
1092=head2 Finding Magic
1093
1094 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
1095
1096This routine returns a pointer to the C<MAGIC> structure stored in the SV.
1097If the SV does not have that magical feature, C<NULL> is returned. Also,
54310121 1098if the SV is not of type SVt_PVMG, Perl may core dump.
d1b91892 1099
08105a92 1100 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1101
1102This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1103field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1104the mg_type field is changed to be the lowercase letter.
a0d0e21e 1105
04343c6d
GS
1106=head2 Understanding the Magic of Tied Hashes and Arrays
1107
14befaf4
DM
1108Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1109magic type.
9edb2b46
GS
1110
1111WARNING: As of the 5.004 release, proper usage of the array and hash
1112access functions requires understanding a few caveats. Some
1113of these caveats are actually considered bugs in the API, to be fixed
1114in later releases, and are bracketed with [MAYCHANGE] below. If
1115you find yourself actually applying such information in this section, be
1116aware that the behavior may change in the future, umm, without warning.
04343c6d 1117
1526ead6 1118The perl tie function associates a variable with an object that implements
9a68f1db 1119the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1120tie function from an XSUB, you must mimic this behaviour. The code below
1121carries out the necessary steps - firstly it creates a new hash, and then
1122creates a second hash which it blesses into the class which will implement
1123the tie methods. Lastly it ties the two hashes together, and returns a
1124reference to the new tied hash. Note that the code below does NOT call the
1125TIEHASH method in the MyTie class -
1126see L<Calling Perl Routines from within C Programs> for details on how
1127to do this.
1128
1129 SV*
1130 mytie()
1131 PREINIT:
1132 HV *hash;
1133 HV *stash;
1134 SV *tie;
1135 CODE:
1136 hash = newHV();
1137 tie = newRV_noinc((SV*)newHV());
1138 stash = gv_stashpv("MyTie", TRUE);
1139 sv_bless(tie, stash);
899e16d0 1140 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1141 RETVAL = newRV_noinc(hash);
1142 OUTPUT:
1143 RETVAL
1144
04343c6d
GS
1145The C<av_store> function, when given a tied array argument, merely
1146copies the magic of the array onto the value to be "stored", using
1147C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1148actually need to be stored in the array. [MAYCHANGE] After a call to
1149C<av_store> on a tied array, the caller will usually need to call
1150C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1151TIEARRAY object. If C<av_store> did return NULL, a call to
1152C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1153leak. [/MAYCHANGE]
04343c6d
GS
1154
1155The previous paragraph is applicable verbatim to tied hash access using the
1156C<hv_store> and C<hv_store_ent> functions as well.
1157
1158C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1159C<hv_fetch_ent> actually return an undefined mortal value whose magic
1160has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1161need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1162need to call C<mg_get()> on the returned value in order to actually invoke
1163the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1164you may also call C<mg_set()> on the return value after possibly assigning
1165a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1166method on the TIE object. [/MAYCHANGE]
04343c6d 1167
9edb2b46 1168[MAYCHANGE]
04343c6d
GS
1169In other words, the array or hash fetch/store functions don't really
1170fetch and store actual values in the case of tied arrays and hashes. They
1171merely call C<mg_copy> to attach magic to the values that were meant to be
1172"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1173do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1174the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1175and hashes.
1176
1177Currently (as of perl version 5.004), use of the hash and array access
1178functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1179"normal" hashes and arrays, or on their tied variants. The API may be
1180changed to provide more transparent access to both tied and normal data
1181types in future versions.
1182[/MAYCHANGE]
04343c6d
GS
1183
1184You would do well to understand that the TIEARRAY and TIEHASH interfaces
1185are mere sugar to invoke some perl method calls while using the uniform hash
1186and array syntax. The use of this sugar imposes some overhead (typically
1187about two to four extra opcodes per FETCH/STORE operation, in addition to
1188the creation of all the mortal variables required to invoke the methods).
1189This overhead will be comparatively small if the TIE methods are themselves
1190substantial, but if they are only a few statements long, the overhead
1191will not be insignificant.
1192
d1c897a1
IZ
1193=head2 Localizing changes
1194
1195Perl has a very handy construction
1196
1197 {
1198 local $var = 2;
1199 ...
1200 }
1201
1202This construction is I<approximately> equivalent to
1203
1204 {
1205 my $oldvar = $var;
1206 $var = 2;
1207 ...
1208 $var = $oldvar;
1209 }
1210
1211The biggest difference is that the first construction would
1212reinstate the initial value of $var, irrespective of how control exits
9a68f1db 1213the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1214more efficient as well.
1215
1216There is a way to achieve a similar task from C via Perl API: create a
1217I<pseudo-block>, and arrange for some changes to be automatically
1218undone at the end of it, either explicit, or via a non-local exit (via
1219die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1220C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1221Such a construct may be created specially for some important localized
1222task, or an existing one (like boundaries of enclosing Perl
1223subroutine/block, or an existing pair for freeing TMPs) may be
1224used. (In the second case the overhead of additional localization must
1225be almost negligible.) Note that any XSUB is automatically enclosed in
1226an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1227
1228Inside such a I<pseudo-block> the following service is available:
1229
13a2d996 1230=over 4
d1c897a1
IZ
1231
1232=item C<SAVEINT(int i)>
1233
1234=item C<SAVEIV(IV i)>
1235
1236=item C<SAVEI32(I32 i)>
1237
1238=item C<SAVELONG(long i)>
1239
1240These macros arrange things to restore the value of integer variable
1241C<i> at the end of enclosing I<pseudo-block>.
1242
1243=item C<SAVESPTR(s)>
1244
1245=item C<SAVEPPTR(p)>
1246
1247These macros arrange things to restore the value of pointers C<s> and
1248C<p>. C<s> must be a pointer of a type which survives conversion to
1249C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1250and back.
1251
1252=item C<SAVEFREESV(SV *sv)>
1253
1254The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1255I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1256mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1257extends the lifetime of C<sv> until the beginning of the next statement,
1258C<SAVEFREESV> extends it until the end of the enclosing scope. These
1259lifetimes can be wildly different.
1260
1261Also compare C<SAVEMORTALIZESV>.
1262
1263=item C<SAVEMORTALIZESV(SV *sv)>
1264
1265Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1266scope instead of decrementing its reference count. This usually has the
1267effect of keeping C<sv> alive until the statement that called the currently
1268live scope has finished executing.
d1c897a1
IZ
1269
1270=item C<SAVEFREEOP(OP *op)>
1271
1272The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1273
1274=item C<SAVEFREEPV(p)>
1275
1276The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1277end of I<pseudo-block>.
1278
1279=item C<SAVECLEARSV(SV *sv)>
1280
1281Clears a slot in the current scratchpad which corresponds to C<sv> at
1282the end of I<pseudo-block>.
1283
1284=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1285
1286The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1287string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1288short-lived storage, the corresponding string may be reallocated like
1289this:
1290
9cde0e7f 1291 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1292
c76ac1ee 1293=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1294
1295At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1296only argument C<p>.
1297
1298=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1299
1300At the end of I<pseudo-block> the function C<f> is called with the
1301implicit context argument (if any), and C<p>.
d1c897a1
IZ
1302
1303=item C<SAVESTACK_POS()>
1304
1305The current offset on the Perl internal stack (cf. C<SP>) is restored
1306at the end of I<pseudo-block>.
1307
1308=back
1309
1310The following API list contains functions, thus one needs to
1311provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1312or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1313function takes C<int *>.
1314
13a2d996 1315=over 4
d1c897a1
IZ
1316
1317=item C<SV* save_scalar(GV *gv)>
1318
1319Equivalent to Perl code C<local $gv>.
1320
1321=item C<AV* save_ary(GV *gv)>
1322
1323=item C<HV* save_hash(GV *gv)>
1324
1325Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1326
1327=item C<void save_item(SV *item)>
1328
1329Duplicates the current value of C<SV>, on the exit from the current
1330C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1331using the stored value.
1332
1333=item C<void save_list(SV **sarg, I32 maxsarg)>
1334
1335A variant of C<save_item> which takes multiple arguments via an array
1336C<sarg> of C<SV*> of length C<maxsarg>.
1337
1338=item C<SV* save_svref(SV **sptr)>
1339
d1be9408 1340Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1341
1342=item C<void save_aptr(AV **aptr)>
1343
1344=item C<void save_hptr(HV **hptr)>
1345
1346Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1347
1348=back
1349
1350The C<Alias> module implements localization of the basic types within the
1351I<caller's scope>. People who are interested in how to localize things in
1352the containing scope should take a look there too.
1353
0a753a76 1354=head1 Subroutines
a0d0e21e 1355
68dc0745 1356=head2 XSUBs and the Argument Stack
5f05dabc
PP
1357
1358The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1359An XSUB routine will have a stack that contains the arguments from the Perl
1360program, and a way to map from the Perl data structures to a C equivalent.
1361
1362The stack arguments are accessible through the C<ST(n)> macro, which returns
1363the C<n>'th stack argument. Argument 0 is the first argument passed in the
1364Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1365an C<SV*> is used.
1366
1367Most of the time, output from the C routine can be handled through use of
1368the RETVAL and OUTPUT directives. However, there are some cases where the
1369argument stack is not already long enough to handle all the return values.
1370An example is the POSIX tzname() call, which takes no arguments, but returns
1371two, the local time zone's standard and summer time abbreviations.
1372
1373To handle this situation, the PPCODE directive is used and the stack is
1374extended using the macro:
1375
924508f0 1376 EXTEND(SP, num);
5f05dabc 1377
924508f0
GS
1378where C<SP> is the macro that represents the local copy of the stack pointer,
1379and C<num> is the number of elements the stack should be extended by.
5f05dabc 1380
00aadd71 1381Now that there is room on the stack, values can be pushed on it using C<PUSHs>
06f6df17
RGS
1382macro. The pushed values will often need to be "mortal" (See
1383L</Reference Counts and Mortality>).
5f05dabc 1384
00aadd71
NIS
1385 PUSHs(sv_2mortal(newSViv(an_integer)))
1386 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1387 PUSHs(sv_2mortal(newSVnv(3.141592)))
5f05dabc
PP
1388
1389And now the Perl program calling C<tzname>, the two values will be assigned
1390as in:
1391
1392 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1393
1394An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1395to use the macro:
5f05dabc 1396
5f05dabc
PP
1397 XPUSHs(SV*)
1398
00aadd71 1399This macro automatically adjust the stack for you, if needed. Thus, you
5f05dabc 1400do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1401
1402Despite their suggestions in earlier versions of this document the macros
1403C<PUSHi>, C<PUSHn> and C<PUSHp> are I<not> suited to XSUBs which return
1404multiple results, see L</Putting a C value on Perl stack>.
5f05dabc
PP
1405
1406For more information, consult L<perlxs> and L<perlxstut>.
1407
1408=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1409
1410There are four routines that can be used to call a Perl subroutine from
1411within a C program. These four are:
1412
954c1994
GS
1413 I32 call_sv(SV*, I32);
1414 I32 call_pv(const char*, I32);
1415 I32 call_method(const char*, I32);
1416 I32 call_argv(const char*, I32, register char**);
a0d0e21e 1417
954c1994 1418The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1419contains either the name of the Perl subroutine to be called, or a
1420reference to the subroutine. The second argument consists of flags
1421that control the context in which the subroutine is called, whether
1422or not the subroutine is being passed arguments, how errors should be
1423trapped, and how to treat return values.
a0d0e21e
LW
1424
1425All four routines return the number of arguments that the subroutine returned
1426on the Perl stack.
1427
9a68f1db 1428These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1429but those names are now deprecated; macros of the same name are provided for
1430compatibility.
1431
1432When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1433must manipulate the Perl stack. These include the following macros and
1434functions:
a0d0e21e
LW
1435
1436 dSP
924508f0 1437 SP
a0d0e21e
LW
1438 PUSHMARK()
1439 PUTBACK
1440 SPAGAIN
1441 ENTER
1442 SAVETMPS
1443 FREETMPS
1444 LEAVE
1445 XPUSH*()
cb1a09d0 1446 POP*()
a0d0e21e 1447
5f05dabc
PP
1448For a detailed description of calling conventions from C to Perl,
1449consult L<perlcall>.
a0d0e21e 1450
5f05dabc 1451=head2 Memory Allocation
a0d0e21e 1452
06f6df17
RGS
1453=head3 Allocation
1454
86058a2d
GS
1455All memory meant to be used with the Perl API functions should be manipulated
1456using the macros described in this section. The macros provide the necessary
1457transparency between differences in the actual malloc implementation that is
1458used within perl.
1459
1460It is suggested that you enable the version of malloc that is distributed
5f05dabc 1461with Perl. It keeps pools of various sizes of unallocated memory in
07fa94a1
JO
1462order to satisfy allocation requests more quickly. However, on some
1463platforms, it may cause spurious malloc or free errors.
d1b91892 1464
06f6df17
RGS
1465The following three macros are used to initially allocate memory :
1466
d1b91892
AD
1467 New(x, pointer, number, type);
1468 Newc(x, pointer, number, type, cast);
1469 Newz(x, pointer, number, type);
1470
5f05dabc
PP
1471The first argument C<x> was a "magic cookie" that was used to keep track
1472of who called the macro, to help when debugging memory problems. However,
07fa94a1
JO
1473the current code makes no use of this feature (most Perl developers now
1474use run-time memory checkers), so this argument can be any number.
5f05dabc
PP
1475
1476The second argument C<pointer> should be the name of a variable that will
1477point to the newly allocated memory.
d1b91892 1478
d1b91892
AD
1479The third and fourth arguments C<number> and C<type> specify how many of
1480the specified type of data structure should be allocated. The argument
1481C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1482should be used if the C<pointer> argument is different from the C<type>
1483argument.
1484
1485Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1486to zero out all the newly allocated memory.
1487
06f6df17
RGS
1488=head3 Reallocation
1489
d1b91892
AD
1490 Renew(pointer, number, type);
1491 Renewc(pointer, number, type, cast);
1492 Safefree(pointer)
1493
1494These three macros are used to change a memory buffer size or to free a
1495piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1496match those of C<New> and C<Newc> with the exception of not needing the
1497"magic cookie" argument.
1498
06f6df17
RGS
1499=head3 Moving
1500
d1b91892
AD
1501 Move(source, dest, number, type);
1502 Copy(source, dest, number, type);
1503 Zero(dest, number, type);
1504
1505These three macros are used to move, copy, or zero out previously allocated
1506memory. The C<source> and C<dest> arguments point to the source and
1507destination starting points. Perl will move, copy, or zero out C<number>
1508instances of the size of the C<type> data structure (using the C<sizeof>
1509function).
a0d0e21e 1510
5f05dabc 1511=head2 PerlIO
ce3d39e2 1512
5f05dabc
PP
1513The most recent development releases of Perl has been experimenting with
1514removing Perl's dependency on the "normal" standard I/O suite and allowing
1515other stdio implementations to be used. This involves creating a new
1516abstraction layer that then calls whichever implementation of stdio Perl
68dc0745 1517was compiled with. All XSUBs should now use the functions in the PerlIO
5f05dabc
PP
1518abstraction layer and not make any assumptions about what kind of stdio
1519is being used.
1520
1521For a complete description of the PerlIO abstraction, consult L<perlapio>.
1522
8ebc5c01 1523=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1524
1525A lot of opcodes (this is an elementary operation in the internal perl
1526stack machine) put an SV* on the stack. However, as an optimization
1527the corresponding SV is (usually) not recreated each time. The opcodes
1528reuse specially assigned SVs (I<target>s) which are (as a corollary)
1529not constantly freed/created.
1530
0a753a76 1531Each of the targets is created only once (but see
ce3d39e2
IZ
1532L<Scratchpads and recursion> below), and when an opcode needs to put
1533an integer, a double, or a string on stack, it just sets the
1534corresponding parts of its I<target> and puts the I<target> on stack.
1535
1536The macro to put this target on stack is C<PUSHTARG>, and it is
1537directly used in some opcodes, as well as indirectly in zillions of
1538others, which use it via C<(X)PUSH[pni]>.
1539
1bd1c0d5
SC
1540Because the target is reused, you must be careful when pushing multiple
1541values on the stack. The following code will not do what you think:
1542
1543 XPUSHi(10);
1544 XPUSHi(20);
1545
1546This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1547the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1548At the end of the operation, the stack does not contain the values 10
1549and 20, but actually contains two pointers to C<TARG>, which we have set
1550to 20. If you need to push multiple different values, use C<XPUSHs>,
1551which bypasses C<TARG>.
1552
1553On a related note, if you do use C<(X)PUSH[npi]>, then you're going to
1554need a C<dTARG> in your variable declarations so that the C<*PUSH*>
00aadd71 1555macros can make use of the local variable C<TARG>.
1bd1c0d5 1556
8ebc5c01 1557=head2 Scratchpads
ce3d39e2 1558
54310121 1559The question remains on when the SVs which are I<target>s for opcodes
5f05dabc
PP
1560are created. The answer is that they are created when the current unit --
1561a subroutine or a file (for opcodes for statements outside of
1562subroutines) -- is compiled. During this time a special anonymous Perl
ce3d39e2
IZ
1563array is created, which is called a scratchpad for the current
1564unit.
1565
54310121 1566A scratchpad keeps SVs which are lexicals for the current unit and are
ce3d39e2
IZ
1567targets for opcodes. One can deduce that an SV lives on a scratchpad
1568by looking on its flags: lexicals have C<SVs_PADMY> set, and
1569I<target>s have C<SVs_PADTMP> set.
1570
54310121
PP
1571The correspondence between OPs and I<target>s is not 1-to-1. Different
1572OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1573would not conflict with the expected life of the temporary.
1574
2ae324a7 1575=head2 Scratchpads and recursion
ce3d39e2
IZ
1576
1577In fact it is not 100% true that a compiled unit contains a pointer to
1578the scratchpad AV. In fact it contains a pointer to an AV of
1579(initially) one element, and this element is the scratchpad AV. Why do
1580we need an extra level of indirection?
1581
9a68f1db 1582The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2
IZ
1583these can create several execution pointers going into the same
1584subroutine. For the subroutine-child not write over the temporaries
1585for the subroutine-parent (lifespan of which covers the call to the
1586child), the parent and the child should have different
1587scratchpads. (I<And> the lexicals should be separate anyway!)
1588
5f05dabc
PP
1589So each subroutine is born with an array of scratchpads (of length 1).
1590On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1591depth of the recursion is not more than the length of this array, and
1592if it is, new scratchpad is created and pushed into the array.
1593
1594The I<target>s on this scratchpad are C<undef>s, but they are already
1595marked with correct flags.
1596
0a753a76
PP
1597=head1 Compiled code
1598
1599=head2 Code tree
1600
1601Here we describe the internal form your code is converted to by
1602Perl. Start with a simple example:
1603
1604 $a = $b + $c;
1605
1606This is converted to a tree similar to this one:
1607
1608 assign-to
1609 / \
1610 + $a
1611 / \
1612 $b $c
1613
7b8d334a 1614(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1615parsed your code, but has nothing to do with the execution order.
1616There is an additional "thread" going through the nodes of the tree
1617which shows the order of execution of the nodes. In our simplified
1618example above it looks like:
1619
1620 $b ---> $c ---> + ---> $a ---> assign-to
1621
1622But with the actual compile tree for C<$a = $b + $c> it is different:
1623some nodes I<optimized away>. As a corollary, though the actual tree
1624contains more nodes than our simplified example, the execution order
1625is the same as in our example.
1626
1627=head2 Examining the tree
1628
06f6df17
RGS
1629If you have your perl compiled for debugging (usually done with
1630C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1631compiled tree by specifying C<-Dx> on the Perl command line. The
1632output takes several lines per node, and for C<$b+$c> it looks like
1633this:
1634
1635 5 TYPE = add ===> 6
1636 TARG = 1
1637 FLAGS = (SCALAR,KIDS)
1638 {
1639 TYPE = null ===> (4)
1640 (was rv2sv)
1641 FLAGS = (SCALAR,KIDS)
1642 {
1643 3 TYPE = gvsv ===> 4
1644 FLAGS = (SCALAR)
1645 GV = main::b
1646 }
1647 }
1648 {
1649 TYPE = null ===> (5)
1650 (was rv2sv)
1651 FLAGS = (SCALAR,KIDS)
1652 {
1653 4 TYPE = gvsv ===> 5
1654 FLAGS = (SCALAR)
1655 GV = main::c
1656 }
1657 }
1658
1659This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1660not optimized away (one per number in the left column). The immediate
1661children of the given node correspond to C<{}> pairs on the same level
1662of indentation, thus this listing corresponds to the tree:
1663
1664 add
1665 / \
1666 null null
1667 | |
1668 gvsv gvsv
1669
1670The execution order is indicated by C<===E<gt>> marks, thus it is C<3
16714 5 6> (node C<6> is not included into above listing), i.e.,
1672C<gvsv gvsv add whatever>.
1673
9afa14e3
SC
1674Each of these nodes represents an op, a fundamental operation inside the
1675Perl core. The code which implements each operation can be found in the
1676F<pp*.c> files; the function which implements the op with type C<gvsv>
1677is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1678different numbers of children: C<add> is a binary operator, as one would
1679expect, and so has two children. To accommodate the various different
1680numbers of children, there are various types of op data structure, and
1681they link together in different ways.
1682
1683The simplest type of op structure is C<OP>: this has no children. Unary
1684operators, C<UNOP>s, have one child, and this is pointed to by the
1685C<op_first> field. Binary operators (C<BINOP>s) have not only an
1686C<op_first> field but also an C<op_last> field. The most complex type of
1687op is a C<LISTOP>, which has any number of children. In this case, the
1688first child is pointed to by C<op_first> and the last child by
1689C<op_last>. The children in between can be found by iteratively
1690following the C<op_sibling> pointer from the first child to the last.
1691
1692There are also two other op types: a C<PMOP> holds a regular expression,
1693and has no children, and a C<LOOP> may or may not have children. If the
1694C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1695complicate matters, if a C<UNOP> is actually a C<null> op after
1696optimization (see L</Compile pass 2: context propagation>) it will still
1697have children in accordance with its former type.
1698
06f6df17
RGS
1699Another way to examine the tree is to use a compiler back-end module, such
1700as L<B::Concise>.
1701
0a753a76
PP
1702=head2 Compile pass 1: check routines
1703
8870b5c7
GS
1704The tree is created by the compiler while I<yacc> code feeds it
1705the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
1706the first pass of perl compilation.
1707
1708What makes this pass interesting for perl developers is that some
1709optimization may be performed on this pass. This is optimization by
8870b5c7 1710so-called "check routines". The correspondence between node names
0a753a76
PP
1711and corresponding check routines is described in F<opcode.pl> (do not
1712forget to run C<make regen_headers> if you modify this file).
1713
1714A check routine is called when the node is fully constructed except
7b8d334a 1715for the execution-order thread. Since at this time there are no
0a753a76
PP
1716back-links to the currently constructed node, one can do most any
1717operation to the top-level node, including freeing it and/or creating
1718new nodes above/below it.
1719
1720The check routine returns the node which should be inserted into the
1721tree (if the top-level node was not modified, check routine returns
1722its argument).
1723
1724By convention, check routines have names C<ck_*>. They are usually
1725called from C<new*OP> subroutines (or C<convert>) (which in turn are
1726called from F<perly.y>).
1727
1728=head2 Compile pass 1a: constant folding
1729
1730Immediately after the check routine is called the returned node is
1731checked for being compile-time executable. If it is (the value is
1732judged to be constant) it is immediately executed, and a I<constant>
1733node with the "return value" of the corresponding subtree is
1734substituted instead. The subtree is deleted.
1735
1736If constant folding was not performed, the execution-order thread is
1737created.
1738
1739=head2 Compile pass 2: context propagation
1740
1741When a context for a part of compile tree is known, it is propagated
a3cb178b 1742down through the tree. At this time the context can have 5 values
0a753a76
PP
1743(instead of 2 for runtime context): void, boolean, scalar, list, and
1744lvalue. In contrast with the pass 1 this pass is processed from top
1745to bottom: a node's context determines the context for its children.
1746
1747Additional context-dependent optimizations are performed at this time.
1748Since at this moment the compile tree contains back-references (via
1749"thread" pointers), nodes cannot be free()d now. To allow
1750optimized-away nodes at this stage, such nodes are null()ified instead
1751of free()ing (i.e. their type is changed to OP_NULL).
1752
1753=head2 Compile pass 3: peephole optimization
1754
1755After the compile tree for a subroutine (or for an C<eval> or a file)
1756is created, an additional pass over the code is performed. This pass
1757is neither top-down or bottom-up, but in the execution order (with
7b8d334a 1758additional complications for conditionals). These optimizations are
0a753a76
PP
1759done in the subroutine peep(). Optimizations performed at this stage
1760are subject to the same restrictions as in the pass 2.
1761
1ba7f851
PJ
1762=head2 Pluggable runops
1763
1764The compile tree is executed in a runops function. There are two runops
1765functions in F<run.c>. C<Perl_runops_debug> is used with DEBUGGING and
1766C<Perl_runops_standard> is used otherwise. For fine control over the
1767execution of the compile tree it is possible to provide your own runops
1768function.
1769
1770It's probably best to copy one of the existing runops functions and
1771change it to suit your needs. Then, in the BOOT section of your XS
1772file, add the line:
1773
1774 PL_runops = my_runops;
1775
1776This function should be as efficient as possible to keep your programs
1777running as fast as possible.
1778
9afa14e3
SC
1779=head1 Examining internal data structures with the C<dump> functions
1780
1781To aid debugging, the source file F<dump.c> contains a number of
1782functions which produce formatted output of internal data structures.
1783
1784The most commonly used of these functions is C<Perl_sv_dump>; it's used
1785for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
1786C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 1787module should already be familiar with its format.
9afa14e3
SC
1788
1789C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 1790derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
1791C<Perl_dump_eval> will dump the main root of the code being evaluated,
1792exactly like C<-Dx>.
1793
1794Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
1795op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
1796subroutines in a package like so: (Thankfully, these are all xsubs, so
1797there is no op tree)
1798
1799 (gdb) print Perl_dump_packsubs(PL_defstash)
1800
1801 SUB attributes::bootstrap = (xsub 0x811fedc 0)
1802
1803 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
1804
1805 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
1806
1807 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
1808
1809 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
1810
1811and C<Perl_dump_all>, which dumps all the subroutines in the stash and
1812the op tree of the main root.
1813
954c1994 1814=head1 How multiple interpreters and concurrency are supported
ee072b34 1815
ee072b34
GS
1816=head2 Background and PERL_IMPLICIT_CONTEXT
1817
1818The Perl interpreter can be regarded as a closed box: it has an API
1819for feeding it code or otherwise making it do things, but it also has
1820functions for its own use. This smells a lot like an object, and
1821there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
1822interpreters, with one interpreter represented either as a C structure,
1823or inside a thread-specific structure. These structures contain all
1824the context, the state of that interpreter.
1825
9a68f1db 1826Two macros control the major Perl build flavors: MULTIPLICITY and
acfe0abc
GS
1827USE_5005THREADS. The MULTIPLICITY build has a C structure
1828that packages all the interpreter state, and there is a similar thread-specific
1829data structure under USE_5005THREADS. In both cases,
54aff467
GS
1830PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1831support for passing in a "hidden" first argument that represents all three
651a3225 1832data structures.
54aff467
GS
1833
1834All this obviously requires a way for the Perl internal functions to be
acfe0abc 1835either subroutines taking some kind of structure as the first
ee072b34 1836argument, or subroutines taking nothing as the first argument. To
acfe0abc 1837enable these two very different ways of building the interpreter,
ee072b34
GS
1838the Perl source (as it does in so many other situations) makes heavy
1839use of macros and subroutine naming conventions.
1840
54aff467 1841First problem: deciding which functions will be public API functions and
00aadd71 1842which will be private. All functions whose names begin C<S_> are private
954c1994
GS
1843(think "S" for "secret" or "static"). All other functions begin with
1844"Perl_", but just because a function begins with "Perl_" does not mean it is
00aadd71
NIS
1845part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
1846function is part of the API is to find its entry in L<perlapi>.
1847If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
1848think it should be (i.e., you need it for your extension), send mail via
a422fd2d 1849L<perlbug> explaining why you think it should be.
ee072b34
GS
1850
1851Second problem: there must be a syntax so that the same subroutine
1852declarations and calls can pass a structure as their first argument,
1853or pass nothing. To solve this, the subroutines are named and
1854declared in a particular way. Here's a typical start of a static
1855function used within the Perl guts:
1856
1857 STATIC void
1858 S_incline(pTHX_ char *s)
1859
acfe0abc
GS
1860STATIC becomes "static" in C, and may be #define'd to nothing in some
1861configurations in future.
ee072b34 1862
651a3225
GS
1863A public function (i.e. part of the internal API, but not necessarily
1864sanctioned for use in extensions) begins like this:
ee072b34
GS
1865
1866 void
2307c6d0 1867 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34
GS
1868
1869C<pTHX_> is one of a number of macros (in perl.h) that hide the
1870details of the interpreter's context. THX stands for "thread", "this",
1871or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1872The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
1873or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
1874their variants.
ee072b34 1875
a7486cbb
JH
1876When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
1877first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
1878in the pTHX_ macro indicates that the macro expansion needs a comma
1879after the context argument because other arguments follow it. If
1880PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
1881subroutine is not prototyped to take the extra argument. The form of the
1882macro without the trailing underscore is used when there are no additional
ee072b34
GS
1883explicit arguments.
1884
54aff467 1885When a core function calls another, it must pass the context. This
2307c6d0 1886is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
1887something like this:
1888
2307c6d0
SB
1889 #ifdef PERL_IMPLICIT_CONTEXT
1890 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 1891 /* can't do this for vararg functions, see below */
2307c6d0
SB
1892 #else
1893 #define sv_setiv Perl_sv_setiv
1894 #endif
ee072b34
GS
1895
1896This works well, and means that XS authors can gleefully write:
1897
2307c6d0 1898 sv_setiv(foo, bar);
ee072b34
GS
1899
1900and still have it work under all the modes Perl could have been
1901compiled with.
1902
ee072b34
GS
1903This doesn't work so cleanly for varargs functions, though, as macros
1904imply that the number of arguments is known in advance. Instead we
1905either need to spell them out fully, passing C<aTHX_> as the first
1906argument (the Perl core tends to do this with functions like
1907Perl_warner), or use a context-free version.
1908
1909The context-free version of Perl_warner is called
1910Perl_warner_nocontext, and does not take the extra argument. Instead
1911it does dTHX; to get the context from thread-local storage. We
1912C<#define warner Perl_warner_nocontext> so that extensions get source
1913compatibility at the expense of performance. (Passing an arg is
1914cheaper than grabbing it from thread-local storage.)
1915
acfe0abc 1916You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
1917Those are strictly for use within the core. Extensions and embedders
1918need only be aware of [pad]THX.
1919
a7486cbb
JH
1920=head2 So what happened to dTHR?
1921
1922C<dTHR> was introduced in perl 5.005 to support the older thread model.
1923The older thread model now uses the C<THX> mechanism to pass context
1924pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
1925later still have it for backward source compatibility, but it is defined
1926to be a no-op.
1927
ee072b34
GS
1928=head2 How do I use all this in extensions?
1929
1930When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1931any functions in the Perl API will need to pass the initial context
1932argument somehow. The kicker is that you will need to write it in
1933such a way that the extension still compiles when Perl hasn't been
1934built with PERL_IMPLICIT_CONTEXT enabled.
1935
1936There are three ways to do this. First, the easy but inefficient way,
1937which is also the default, in order to maintain source compatibility
1938with extensions: whenever XSUB.h is #included, it redefines the aTHX
1939and aTHX_ macros to call a function that will return the context.
1940Thus, something like:
1941
2307c6d0 1942 sv_setiv(sv, num);
ee072b34 1943
4375e838 1944in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 1945in effect:
ee072b34 1946
2307c6d0 1947 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 1948
54aff467 1949or to this otherwise:
ee072b34 1950
2307c6d0 1951 Perl_sv_setiv(sv, num);
ee072b34
GS
1952
1953You have to do nothing new in your extension to get this; since
2fa86c13 1954the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
1955work.
1956
1957The second, more efficient way is to use the following template for
1958your Foo.xs:
1959
c52f9dcd
JH
1960 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1961 #include "EXTERN.h"
1962 #include "perl.h"
1963 #include "XSUB.h"
ee072b34
GS
1964
1965 static my_private_function(int arg1, int arg2);
1966
c52f9dcd
JH
1967 static SV *
1968 my_private_function(int arg1, int arg2)
1969 {
1970 dTHX; /* fetch context */
1971 ... call many Perl API functions ...
1972 }
ee072b34
GS
1973
1974 [... etc ...]
1975
c52f9dcd 1976 MODULE = Foo PACKAGE = Foo
ee072b34 1977
c52f9dcd 1978 /* typical XSUB */
ee072b34 1979
c52f9dcd
JH
1980 void
1981 my_xsub(arg)
1982 int arg
1983 CODE:
1984 my_private_function(arg, 10);
ee072b34
GS
1985
1986Note that the only two changes from the normal way of writing an
1987extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
1988including the Perl headers, followed by a C<dTHX;> declaration at
1989the start of every function that will call the Perl API. (You'll
1990know which functions need this, because the C compiler will complain
1991that there's an undeclared identifier in those functions.) No changes
1992are needed for the XSUBs themselves, because the XS() macro is
1993correctly defined to pass in the implicit context if needed.
1994
1995The third, even more efficient way is to ape how it is done within
1996the Perl guts:
1997
1998
c52f9dcd
JH
1999 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2000 #include "EXTERN.h"
2001 #include "perl.h"
2002 #include "XSUB.h"
ee072b34
GS
2003
2004 /* pTHX_ only needed for functions that call Perl API */
2005 static my_private_function(pTHX_ int arg1, int arg2);
2006
c52f9dcd
JH
2007 static SV *
2008 my_private_function(pTHX_ int arg1, int arg2)
2009 {
2010 /* dTHX; not needed here, because THX is an argument */
2011 ... call Perl API functions ...
2012 }
ee072b34
GS
2013
2014 [... etc ...]
2015
c52f9dcd 2016 MODULE = Foo PACKAGE = Foo
ee072b34 2017
c52f9dcd 2018 /* typical XSUB */
ee072b34 2019
c52f9dcd
JH
2020 void
2021 my_xsub(arg)
2022 int arg
2023 CODE:
2024 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2025
2026This implementation never has to fetch the context using a function
2027call, since it is always passed as an extra argument. Depending on
2028your needs for simplicity or efficiency, you may mix the previous
2029two approaches freely.
2030
651a3225
GS
2031Never add a comma after C<pTHX> yourself--always use the form of the
2032macro with the underscore for functions that take explicit arguments,
2033or the form without the argument for functions with no explicit arguments.
ee072b34 2034
a7486cbb
JH
2035=head2 Should I do anything special if I call perl from multiple threads?
2036
2037If you create interpreters in one thread and then proceed to call them in
2038another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2039initialized correctly in each of those threads.
2040
2041The C<perl_alloc> and C<perl_clone> API functions will automatically set
2042the TLS slot to the interpreter they created, so that there is no need to do
2043anything special if the interpreter is always accessed in the same thread that
2044created it, and that thread did not create or call any other interpreters
2045afterwards. If that is not the case, you have to set the TLS slot of the
2046thread before calling any functions in the Perl API on that particular
2047interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2048thread as the first thing you do:
2049
2050 /* do this before doing anything else with some_perl */
2051 PERL_SET_CONTEXT(some_perl);
2052
2053 ... other Perl API calls on some_perl go here ...
2054
ee072b34
GS
2055=head2 Future Plans and PERL_IMPLICIT_SYS
2056
2057Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2058that the interpreter knows about itself and pass it around, so too are
2059there plans to allow the interpreter to bundle up everything it knows
2060about the environment it's running on. This is enabled with the
acfe0abc 2061PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS
4d1ff10f 2062and USE_5005THREADS on Windows (see inside iperlsys.h).
ee072b34
GS
2063
2064This allows the ability to provide an extra pointer (called the "host"
2065environment) for all the system calls. This makes it possible for
2066all the system stuff to maintain their own state, broken down into
2067seven C structures. These are thin wrappers around the usual system
2068calls (see win32/perllib.c) for the default perl executable, but for a
2069more ambitious host (like the one that would do fork() emulation) all
2070the extra work needed to pretend that different interpreters are
2071actually different "processes", would be done here.
2072
2073The Perl engine/interpreter and the host are orthogonal entities.
2074There could be one or more interpreters in a process, and one or
2075more "hosts", with free association between them.
2076
a422fd2d
SC
2077=head1 Internal Functions
2078
2079All of Perl's internal functions which will be exposed to the outside
06f6df17 2080world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d
SC
2081functions or functions used in a program in which Perl is embedded.
2082Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2083static functions start with C<S_>.)
a422fd2d
SC
2084
2085Inside the Perl core, you can get at the functions either with or
2086without the C<Perl_> prefix, thanks to a bunch of defines that live in
2087F<embed.h>. This header file is generated automatically from
2088F<embed.pl>. F<embed.pl> also creates the prototyping header files for
2089the internal functions, generates the documentation and a lot of other
2090bits and pieces. It's important that when you add a new function to the
2091core or change an existing one, you change the data in the table at the
2092end of F<embed.pl> as well. Here's a sample entry from that table:
2093
2094 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2095
2096The second column is the return type, the third column the name. Columns
2097after that are the arguments. The first column is a set of flags:
2098
2099=over 3
2100
2101=item A
2102
2103This function is a part of the public API.
2104
2105=item p
2106
2107This function has a C<Perl_> prefix; ie, it is defined as C<Perl_av_fetch>
2108
2109=item d
2110
2111This function has documentation using the C<apidoc> feature which we'll
2112look at in a second.
2113
2114=back
2115
2116Other available flags are:
2117
2118=over 3
2119
2120=item s
2121
a7486cbb
JH
2122This is a static function and is defined as C<S_whatever>, and usually
2123called within the sources as C<whatever(...)>.
a422fd2d
SC
2124
2125=item n
2126
2127This does not use C<aTHX_> and C<pTHX> to pass interpreter context. (See
2128L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
2129
2130=item r
2131
2132This function never returns; C<croak>, C<exit> and friends.
2133
2134=item f
2135
2136This function takes a variable number of arguments, C<printf> style.
2137The argument list should end with C<...>, like this:
2138
2139 Afprd |void |croak |const char* pat|...
2140
a7486cbb 2141=item M
a422fd2d 2142
00aadd71 2143This function is part of the experimental development API, and may change
a422fd2d
SC
2144or disappear without notice.
2145
2146=item o
2147
2148This function should not have a compatibility macro to define, say,
2149C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2150
2151=item j
2152
2153This function is not a member of C<CPerlObj>. If you don't know
2154what this means, don't use it.
2155
2156=item x
2157
2158This function isn't exported out of the Perl core.
2159
2160=back
2161
2162If you edit F<embed.pl>, you will need to run C<make regen_headers> to
2163force a rebuild of F<embed.h> and other auto-generated files.
2164
6b4667fc 2165=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2166
6b4667fc
A
2167If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2168formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2169following macros for portability
9dd9db0b 2170
c52f9dcd
JH
2171 IVdf IV in decimal
2172 UVuf UV in decimal
2173 UVof UV in octal
2174 UVxf UV in hexadecimal
2175 NVef NV %e-like
2176 NVff NV %f-like
2177 NVgf NV %g-like
9dd9db0b 2178
6b4667fc
A
2179These will take care of 64-bit integers and long doubles.
2180For example:
2181
c52f9dcd 2182 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2183
2184The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2185
8908e76d
JH
2186If you are printing addresses of pointers, use UVxf combined
2187with PTR2UV(), do not use %lx or %p.
2188
2189=head2 Pointer-To-Integer and Integer-To-Pointer
2190
2191Because pointer size does not necessarily equal integer size,
2192use the follow macros to do it right.
2193
c52f9dcd
JH
2194 PTR2UV(pointer)
2195 PTR2IV(pointer)
2196 PTR2NV(pointer)
2197 INT2PTR(pointertotype, integer)
8908e76d
JH
2198
2199For example:
2200
c52f9dcd
JH
2201 IV iv = ...;
2202 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2203
2204and
2205
c52f9dcd
JH
2206 AV *av = ...;
2207 UV uv = PTR2UV(av);
8908e76d 2208
a422fd2d
SC
2209=head2 Source Documentation
2210
2211There's an effort going on to document the internal functions and
2212automatically produce reference manuals from them - L<perlapi> is one
2213such manual which details all the functions which are available to XS
2214writers. L<perlintern> is the autogenerated manual for the functions
2215which are not part of the API and are supposedly for internal use only.
2216
2217Source documentation is created by putting POD comments into the C
2218source, like this:
2219
2220 /*
2221 =for apidoc sv_setiv
2222
2223 Copies an integer into the given SV. Does not handle 'set' magic. See
2224 C<sv_setiv_mg>.
2225
2226 =cut
2227 */
2228
2229Please try and supply some documentation if you add functions to the
2230Perl core.
2231
2232=head1 Unicode Support
2233
2234Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2235writers to understand this support and make sure that the code they
2236write does not corrupt Unicode data.
2237
2238=head2 What B<is> Unicode, anyway?
2239
2240In the olden, less enlightened times, we all used to use ASCII. Most of
2241us did, anyway. The big problem with ASCII is that it's American. Well,
2242no, that's not actually the problem; the problem is that it's not
2243particularly useful for people who don't use the Roman alphabet. What
2244used to happen was that particular languages would stick their own
2245alphabet in the upper range of the sequence, between 128 and 255. Of
2246course, we then ended up with plenty of variants that weren't quite
2247ASCII, and the whole point of it being a standard was lost.
2248
2249Worse still, if you've got a language like Chinese or
2250Japanese that has hundreds or thousands of characters, then you really
2251can't fit them into a mere 256, so they had to forget about ASCII
2252altogether, and build their own systems using pairs of numbers to refer
2253to one character.
2254
2255To fix this, some people formed Unicode, Inc. and
2256produced a new character set containing all the characters you can
2257possibly think of and more. There are several ways of representing these
1e54db1a 2258characters, and the one Perl uses is called UTF-8. UTF-8 uses
a422fd2d 2259a variable number of bytes to represent a character, instead of just
b3b6085d 2260one. You can learn more about Unicode at http://www.unicode.org/
a422fd2d 2261
1e54db1a 2262=head2 How can I recognise a UTF-8 string?
a422fd2d 2263
1e54db1a
JH
2264You can't. This is because UTF-8 data is stored in bytes just like
2265non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d
SC
2266capital E with a grave accent, is represented by the two bytes
2267C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2268has that byte sequence as well. So you can't tell just by looking - this
2269is what makes Unicode input an interesting problem.
2270
2271The API function C<is_utf8_string> can help; it'll tell you if a string
1e54db1a 2272contains only valid UTF-8 characters. However, it can't do the work for
a422fd2d 2273you. On a character-by-character basis, C<is_utf8_char> will tell you
1e54db1a 2274whether the current character in a string is valid UTF-8.
a422fd2d 2275
1e54db1a 2276=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2277
1e54db1a 2278As mentioned above, UTF-8 uses a variable number of bytes to store a
a422fd2d
SC
2279character. Characters with values 1...128 are stored in one byte, just
2280like good ol' ASCII. Character 129 is stored as C<v194.129>; this
a31a806a 2281continues up to character 191, which is C<v194.191>. Now we've run out of
a422fd2d
SC
2282bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2283so it goes on, moving to three bytes at character 2048.
2284
1e54db1a 2285Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2286how long the first character in it is with the C<UTF8SKIP> macro:
2287
2288 char *utf = "\305\233\340\240\201";
2289 I32 len;
2290
2291 len = UTF8SKIP(utf); /* len is 2 here */
2292 utf += len;
2293 len = UTF8SKIP(utf); /* len is 3 here */
2294
1e54db1a 2295Another way to skip over characters in a UTF-8 string is to use
a422fd2d
SC
2296C<utf8_hop>, which takes a string and a number of characters to skip
2297over. You're on your own about bounds checking, though, so don't use it
2298lightly.
2299
1e54db1a 2300All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2301so you can test if you need to do something special with this
2302character like this (the UTF8_IS_INVARIANT() is a macro that tests
2303whether the byte can be encoded as a single byte even in UTF-8):
a422fd2d 2304
3a2263fe
RGS
2305 U8 *utf;
2306 UV uv; /* Note: a UV, not a U8, not a char */
a422fd2d 2307
3a2263fe 2308 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2309 /* Must treat this as UTF-8 */
a422fd2d
SC
2310 uv = utf8_to_uv(utf);
2311 else
2312 /* OK to treat this character as a byte */
2313 uv = *utf;
2314
2315You can also see in that example that we use C<utf8_to_uv> to get the
2316value of the character; the inverse function C<uv_to_utf8> is available
1e54db1a 2317for putting a UV into UTF-8:
a422fd2d 2318
3a2263fe 2319 if (!UTF8_IS_INVARIANT(uv))
a422fd2d
SC
2320 /* Must treat this as UTF8 */
2321 utf8 = uv_to_utf8(utf8, uv);
2322 else
2323 /* OK to treat this character as a byte */
2324 *utf8++ = uv;
2325
2326You B<must> convert characters to UVs using the above functions if
1e54db1a
JH
2327you're ever in a situation where you have to match UTF-8 and non-UTF-8
2328characters. You may not skip over UTF-8 characters in this case. If you
2329do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2330for instance, if your UTF-8 string contains C<v196.172>, and you skip
2331that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2332So don't do that!
2333
1e54db1a 2334=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2335
2336Currently, Perl deals with Unicode strings and non-Unicode strings
2337slightly differently. If a string has been identified as being UTF-8
2338encoded, Perl will set a flag in the SV, C<SVf_UTF8>. You can check and
2339manipulate this flag with the following macros:
2340
2341 SvUTF8(sv)
2342 SvUTF8_on(sv)
2343 SvUTF8_off(sv)
2344
2345This flag has an important effect on Perl's treatment of the string: if
2346Unicode data is not properly distinguished, regular expressions,
2347C<length>, C<substr> and other string handling operations will have
2348undesirable results.
2349
2350The problem comes when you have, for instance, a string that isn't
1e54db1a
JH
2351flagged is UTF-8, and contains a byte sequence that could be UTF-8 -
2352especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2353
2354Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2355need be sure you don't accidentally knock it off while you're
2356manipulating SVs. More specifically, you cannot expect to do this:
2357
2358 SV *sv;
2359 SV *nsv;
2360 STRLEN len;
2361 char *p;
2362
2363 p = SvPV(sv, len);
2364 frobnicate(p);
2365 nsv = newSVpvn(p, len);
2366
2367The C<char*> string does not tell you the whole story, and you can't
2368copy or reconstruct an SV just by copying the string value. Check if the
1e54db1a 2369old SV has the UTF-8 flag set, and act accordingly:
a422fd2d
SC
2370
2371 p = SvPV(sv, len);
2372 frobnicate(p);
2373 nsv = newSVpvn(p, len);
2374 if (SvUTF8(sv))
2375 SvUTF8_on(nsv);
2376
2377In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2378not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2379appropriately.
2380
3a2263fe 2381Since just passing an SV to an XS function and copying the data of
1e54db1a 2382the SV is not enough to copy the UTF-8 flags, even less right is just
3a2263fe
RGS
2383passing a C<char *> to an XS function.
2384
1e54db1a 2385=head2 How do I convert a string to UTF-8?
a422fd2d 2386
1e54db1a
JH
2387If you're mixing UTF-8 and non-UTF-8 strings, you might find it necessary
2388to upgrade one of the strings to UTF-8. If you've got an SV, the easiest
a422fd2d
SC
2389way to do this is:
2390
2391 sv_utf8_upgrade(sv);
2392
2393However, you must not do this, for example:
2394
2395 if (!SvUTF8(left))
2396 sv_utf8_upgrade(left);
2397
2398If you do this in a binary operator, you will actually change one of the
b1866b2d 2399strings that came into the operator, and, while it shouldn't be noticeable
a422fd2d
SC
2400by the end user, it can cause problems.
2401
1e54db1a 2402Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
a422fd2d 2403string argument. This is useful for having the data available for
b1866b2d 2404comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2405C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2406the string contains any characters above 255 that can't be represented
2407in a single byte.
2408
2409=head2 Is there anything else I need to know?
2410
2411Not really. Just remember these things:
2412
2413=over 3
2414
2415=item *
2416
1e54db1a
JH
2417There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2418is UTF-8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2419something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2420it's not - if you pass on the PV to somewhere, pass on the flag too.
2421
2422=item *
2423
1e54db1a 2424If a string is UTF-8, B<always> use C<utf8_to_uv> to get at the value,
3a2263fe 2425unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2426
2427=item *
2428
1e54db1a 2429When writing a character C<uv> to a UTF-8 string, B<always> use
3a2263fe
RGS
2430C<uv_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2431you can use C<*s = uv>.
a422fd2d
SC
2432
2433=item *
2434
1e54db1a
JH
2435Mixing UTF-8 and non-UTF-8 strings is tricky. Use C<bytes_to_utf8> to get
2436a new string which is UTF-8 encoded. There are tricks you can use to
2437delay deciding whether you need to use a UTF-8 string until you get to a
a422fd2d
SC
2438high character - C<HALF_UPGRADE> is one of those.
2439
2440=back
2441
53e06cf0
SC
2442=head1 Custom Operators
2443
9a68f1db 2444Custom operator support is a new experimental feature that allows you to
53e06cf0
SC
2445define your own ops. This is primarily to allow the building of
2446interpreters for other languages in the Perl core, but it also allows
2447optimizations through the creation of "macro-ops" (ops which perform the
2448functions of multiple ops which are usually executed together, such as
b7cb320d 2449C<gvsv, gvsv, add>.)
53e06cf0 2450
b455bf3f 2451This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0
SC
2452core does not "know" anything special about this op type, and so it will
2453not be involved in any optimizations. This also means that you can
2454define your custom ops to be any op structure - unary, binary, list and
2455so on - you like.
2456
2457It's important to know what custom operators won't do for you. They
2458won't let you add new syntax to Perl, directly. They won't even let you
2459add new keywords, directly. In fact, they won't change the way Perl
2460compiles a program at all. You have to do those changes yourself, after
2461Perl has compiled the program. You do this either by manipulating the op
2462tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2463a custom peephole optimizer with the C<optimize> module.
2464
2465When you do this, you replace ordinary Perl ops with custom ops by
2466creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2467PP function. This should be defined in XS code, and should look like
2468the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2469takes the appropriate number of values from the stack, and you are
2470responsible for adding stack marks if necessary.
2471
2472You should also "register" your op with the Perl interpreter so that it
2473can produce sensible error and warning messages. Since it is possible to
2474have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2475Perl uses the value of C<< o->op_ppaddr >> as a key into the
2476C<PL_custom_op_descs> and C<PL_custom_op_names> hashes. This means you
2477need to enter a name and description for your op at the appropriate
2478place in the C<PL_custom_op_names> and C<PL_custom_op_descs> hashes.
2479
2480Forthcoming versions of C<B::Generate> (version 1.0 and above) should
2481directly support the creation of custom ops by name; C<Opcodes::Custom>
2482will provide functions which make it trivial to "register" custom ops to
2483the Perl interpreter.
2484
954c1994 2485=head1 AUTHORS
e89caa19 2486
954c1994 2487Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
2488E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2489itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 2490
954c1994
GS
2491With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2492Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2493Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2494Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 2495
954c1994 2496=head1 SEE ALSO
cb1a09d0 2497
954c1994 2498perlapi(1), perlintern(1), perlxs(1), perlembed(1)