3 perlguts - Introduction to the Perl API
7 This document attempts to describe how to use the Perl API, as well as
8 to provide some info on the basic workings of the Perl core. It is far
9 from complete and probably contains many errors. Please refer any
10 questions or comments to the author below.
16 Perl has three typedefs that handle Perl's three main data types:
22 Each typedef has specific routines that manipulate the various data types.
24 =head2 What is an "IV"?
26 Perl uses a special typedef IV which is a simple signed integer type that is
27 guaranteed to be large enough to hold a pointer (as well as an integer).
28 Additionally, there is the UV, which is simply an unsigned IV.
30 Perl also uses two special typedefs, I32 and I16, which will always be at
31 least 32-bits and 16-bits long, respectively. (Again, there are U32 and U16,
32 as well.) They will usually be exactly 32 and 16 bits long, but on Crays
33 they will both be 64 bits.
35 =head2 Working with SVs
37 An SV can be created and loaded with one command. There are five types of
38 values that can be loaded: an integer value (IV), an unsigned integer
39 value (UV), a double (NV), a string (PV), and another scalar (SV).
41 The seven routines are:
46 SV* newSVpv(const char*, STRLEN);
47 SV* newSVpvn(const char*, STRLEN);
48 SV* newSVpvf(const char*, ...);
51 C<STRLEN> is an integer type (Size_t, usually defined as size_t in
52 F<config.h>) guaranteed to be large enough to represent the size of
53 any string that perl can handle.
55 In the unlikely case of a SV requiring more complex initialisation, you
56 can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
57 type NULL is returned, else an SV of type PV is returned with len + 1 (for
58 the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
59 the SV has value undef.
61 SV *sv = newSV(0); /* no storage allocated */
62 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage allocated */
64 To change the value of an I<already-existing> SV, there are eight routines:
66 void sv_setiv(SV*, IV);
67 void sv_setuv(SV*, UV);
68 void sv_setnv(SV*, double);
69 void sv_setpv(SV*, const char*);
70 void sv_setpvn(SV*, const char*, STRLEN)
71 void sv_setpvf(SV*, const char*, ...);
72 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool *);
73 void sv_setsv(SV*, SV*);
75 Notice that you can choose to specify the length of the string to be
76 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
77 allow Perl to calculate the length by using C<sv_setpv> or by specifying
78 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
79 determine the string's length by using C<strlen>, which depends on the
80 string terminating with a NUL character.
82 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
83 formatted output becomes the value.
85 C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
86 either a pointer to a variable argument list or the address and length of
87 an array of SVs. The last argument points to a boolean; on return, if that
88 boolean is true, then locale-specific information has been used to format
89 the string, and the string's contents are therefore untrustworthy (see
90 L<perlsec>). This pointer may be NULL if that information is not
91 important. Note that this function requires you to specify the length of
94 The C<sv_set*()> functions are not generic enough to operate on values
95 that have "magic". See L<Magic Virtual Tables> later in this document.
97 All SVs that contain strings should be terminated with a NUL character.
98 If it is not NUL-terminated there is a risk of
99 core dumps and corruptions from code which passes the string to C
100 functions or system calls which expect a NUL-terminated string.
101 Perl's own functions typically add a trailing NUL for this reason.
102 Nevertheless, you should be very careful when you pass a string stored
103 in an SV to a C function or system call.
105 To access the actual value that an SV points to, you can use the macros:
110 SvPV(SV*, STRLEN len)
113 which will automatically coerce the actual scalar type into an IV, UV, double,
116 In the C<SvPV> macro, the length of the string returned is placed into the
117 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
118 not care what the length of the data is, use the C<SvPV_nolen> macro.
119 Historically the C<SvPV> macro with the global variable C<PL_na> has been
120 used in this case. But that can be quite inefficient because C<PL_na> must
121 be accessed in thread-local storage in threaded Perl. In any case, remember
122 that Perl allows arbitrary strings of data that may both contain NULs and
123 might not be terminated by a NUL.
125 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
126 len);>. It might work with your compiler, but it won't work for everyone.
127 Break this sort of statement up into separate assignments:
135 If you want to know if the scalar value is TRUE, you can use:
139 Although Perl will automatically grow strings for you, if you need to force
140 Perl to allocate more memory for your SV, you can use the macro
142 SvGROW(SV*, STRLEN newlen)
144 which will determine if more memory needs to be allocated. If so, it will
145 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
146 decrease, the allocated memory of an SV and that it does not automatically
147 add a byte for the a trailing NUL (perl's own string functions typically do
148 C<SvGROW(sv, len + 1)>).
150 If you have an SV and want to know what kind of data Perl thinks is stored
151 in it, you can use the following macros to check the type of SV you have.
157 You can get and set the current length of the string stored in an SV with
158 the following macros:
161 SvCUR_set(SV*, I32 val)
163 You can also get a pointer to the end of the string stored in the SV
168 But note that these last three macros are valid only if C<SvPOK()> is true.
170 If you want to append something to the end of string stored in an C<SV*>,
171 you can use the following functions:
173 void sv_catpv(SV*, const char*);
174 void sv_catpvn(SV*, const char*, STRLEN);
175 void sv_catpvf(SV*, const char*, ...);
176 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
177 void sv_catsv(SV*, SV*);
179 The first function calculates the length of the string to be appended by
180 using C<strlen>. In the second, you specify the length of the string
181 yourself. The third function processes its arguments like C<sprintf> and
182 appends the formatted output. The fourth function works like C<vsprintf>.
183 You can specify the address and length of an array of SVs instead of the
184 va_list argument. The fifth function extends the string stored in the first
185 SV with the string stored in the second SV. It also forces the second SV
186 to be interpreted as a string.
188 The C<sv_cat*()> functions are not generic enough to operate on values that
189 have "magic". See L<Magic Virtual Tables> later in this document.
191 If you know the name of a scalar variable, you can get a pointer to its SV
192 by using the following:
194 SV* get_sv("package::varname", FALSE);
196 This returns NULL if the variable does not exist.
198 If you want to know if this variable (or any other SV) is actually C<defined>,
203 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
204 Its address can be used whenever an C<SV*> is needed.
205 However, you have to be careful when using C<&PL_sv_undef> as a value in AVs
206 or HVs (see L<AVs, HVs and undefined values>).
208 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
209 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
210 addresses can be used whenever an C<SV*> is needed.
212 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
216 if (I-am-to-return-a-real-value) {
217 sv = sv_2mortal(newSViv(42));
221 This code tries to return a new SV (which contains the value 42) if it should
222 return a real value, or undef otherwise. Instead it has returned a NULL
223 pointer which, somewhere down the line, will cause a segmentation violation,
224 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
225 first line and all will be well.
227 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
228 call is not necessary (see L<Reference Counts and Mortality>).
232 Perl provides the function C<sv_chop> to efficiently remove characters
233 from the beginning of a string; you give it an SV and a pointer to
234 somewhere inside the PV, and it discards everything before the
235 pointer. The efficiency comes by means of a little hack: instead of
236 actually 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
238 effect, and it puts the number of bytes chopped off into the IV field
239 of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
240 many bytes, and adjusts C<SvCUR> and C<SvLEN>.
242 Hence, at this point, the start of the buffer that we allocated lives
243 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
244 into the middle of this allocated storage.
246 This is best demonstrated by example:
248 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
249 SV = PVIV(0x8128450) at 0x81340f0
251 FLAGS = (POK,OOK,pPOK)
253 PV = 0x8135781 ( "1" . ) "2345"\0
257 Here the number of bytes chopped off (1) is put into IV, and
258 C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
259 portion of the string between the "real" and the "fake" beginnings is
260 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
261 the fake beginning, not the real one.
263 Something similar to the offset hack is performed on AVs to enable
264 efficient shifting and splicing off the beginning of the array; while
265 C<AvARRAY> points to the first element in the array that is visible from
266 Perl, C<AvALLOC> points to the real start of the C array. These are
267 usually the same, but a C<shift> operation can be carried out by
268 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvLEN>.
269 Again, the location of the real start of the C array only comes into
270 play when freeing the array. See C<av_shift> in F<av.c>.
272 =head2 What's Really Stored in an SV?
274 Recall that the usual method of determining the type of scalar you have is
275 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
276 usually these macros will always return TRUE and calling the C<Sv*V>
277 macros will do the appropriate conversion of string to integer/double or
278 integer/double to string.
280 If you I<really> need to know if you have an integer, double, or string
281 pointer in an SV, you can use the following three macros instead:
287 These will tell you if you truly have an integer, double, or string pointer
288 stored in your SV. The "p" stands for private.
290 The are various ways in which the private and public flags may differ.
291 For 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
293 routine rather than directly, so SvIOK is false. Another is when
294 numeric conversion has occured and precision has been lost: only the
295 private flag is set on 'lossy' values. So when an NV is converted to an
296 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
298 In general, though, it's best to use the C<Sv*V> macros.
300 =head2 Working with AVs
302 There are two ways to create and load an AV. The first method creates an
307 The second method both creates the AV and initially populates it with SVs:
309 AV* av_make(I32 num, SV **ptr);
311 The second argument points to an array containing C<num> C<SV*>'s. Once the
312 AV has been created, the SVs can be destroyed, if so desired.
314 Once the AV has been created, the following operations are possible on AVs:
316 void av_push(AV*, SV*);
319 void av_unshift(AV*, I32 num);
321 These should be familiar operations, with the exception of C<av_unshift>.
322 This routine adds C<num> elements at the front of the array with the C<undef>
323 value. You must then use C<av_store> (described below) to assign values
324 to these new elements.
326 Here are some other functions:
329 SV** av_fetch(AV*, I32 key, I32 lval);
330 SV** av_store(AV*, I32 key, SV* val);
332 The C<av_len> function returns the highest index value in array (just
333 like $#array in Perl). If the array is empty, -1 is returned. The
334 C<av_fetch> function returns the value at index C<key>, but if C<lval>
335 is non-zero, then C<av_fetch> will store an undef value at that index.
336 The C<av_store> function stores the value C<val> at index C<key>, and does
337 not increment the reference count of C<val>. Thus the caller is responsible
338 for taking care of that, and if C<av_store> returns NULL, the caller will
339 have to decrement the reference count to avoid a memory leak. Note that
340 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
345 void av_extend(AV*, I32 key);
347 The C<av_clear> function deletes all the elements in the AV* array, but
348 does not actually delete the array itself. The C<av_undef> function will
349 delete all the elements in the array plus the array itself. The
350 C<av_extend> function extends the array so that it contains at least C<key+1>
351 elements. If C<key+1> is less than the currently allocated length of the array,
352 then nothing is done.
354 If you know the name of an array variable, you can get a pointer to its AV
355 by using the following:
357 AV* get_av("package::varname", FALSE);
359 This returns NULL if the variable does not exist.
361 See L<Understanding the Magic of Tied Hashes and Arrays> for more
362 information on how to use the array access functions on tied arrays.
364 =head2 Working with HVs
366 To create an HV, you use the following routine:
370 Once the HV has been created, the following operations are possible on HVs:
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);
375 The C<klen> parameter is the length of the key being passed in (Note that
376 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
377 length of the key). The C<val> argument contains the SV pointer to the
378 scalar being stored, and C<hash> is the precomputed hash value (zero if
379 you want C<hv_store> to calculate it for you). The C<lval> parameter
380 indicates whether this fetch is actually a part of a store operation, in
381 which case a new undefined value will be added to the HV with the supplied
382 key and C<hv_fetch> will return as if the value had already existed.
384 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
385 C<SV*>. To access the scalar value, you must first dereference the return
386 value. However, you should check to make sure that the return value is
387 not NULL before dereferencing it.
389 These two functions check if a hash table entry exists, and deletes it.
391 bool hv_exists(HV*, const char* key, U32 klen);
392 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
394 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
395 create and return a mortal copy of the deleted value.
397 And more miscellaneous functions:
402 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
403 table but does not actually delete the hash table. The C<hv_undef> deletes
404 both the entries and the hash table itself.
406 Perl keeps the actual data in linked list of structures with a typedef of HE.
407 These contain the actual key and value pointers (plus extra administrative
408 overhead). The key is a string pointer; the value is an C<SV*>. However,
409 once you have an C<HE*>, to get the actual key and value, use the routines
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 */
420 SV* hv_iterval(HV*, HE* entry);
421 /* Return an SV pointer to the value of the HE
423 SV* hv_iternextsv(HV*, char** key, I32* retlen);
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 */
429 If you know the name of a hash variable, you can get a pointer to its HV
430 by using the following:
432 HV* get_hv("package::varname", FALSE);
434 This returns NULL if the variable does not exist.
436 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
440 hash = (hash * 33) + *key++;
441 hash = hash + (hash >> 5); /* after 5.6 */
443 The last step was added in version 5.6 to improve distribution of
444 lower bits in the resulting hash value.
446 See L<Understanding the Magic of Tied Hashes and Arrays> for more
447 information on how to use the hash access functions on tied hashes.
449 =head2 Hash API Extensions
451 Beginning with version 5.004, the following functions are also supported:
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);
456 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
457 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
459 SV* hv_iterkeysv (HE* entry);
461 Note that these functions take C<SV*> keys, which simplifies writing
462 of extension code that deals with hash structures. These functions
463 also allow passing of C<SV*> keys to C<tie> functions without forcing
464 you to stringify the keys (unlike the previous set of functions).
466 They also return and accept whole hash entries (C<HE*>), making their
467 use more efficient (since the hash number for a particular string
468 doesn't have to be recomputed every time). See L<perlapi> for detailed
471 The following macros must always be used to access the contents of hash
472 entries. Note that the arguments to these macros must be simple
473 variables, since they may get evaluated more than once. See
474 L<perlapi> for detailed descriptions of these macros.
476 HePV(HE* he, STRLEN len)
480 HeSVKEY_force(HE* he)
481 HeSVKEY_set(HE* he, SV* sv)
483 These two lower level macros are defined, but must only be used when
484 dealing with keys that are not C<SV*>s:
489 Note that both C<hv_store> and C<hv_store_ent> do not increment the
490 reference count of the stored C<val>, which is the caller's responsibility.
491 If these functions return a NULL value, the caller will usually have to
492 decrement the reference count of C<val> to avoid a memory leak.
494 =head2 AVs, HVs and undefined values
496 Sometimes you have to store undefined values in AVs or HVs. Although
497 this may be a rare case, it can be tricky. That's because you're
498 used to using C<&PL_sv_undef> if you need an undefined SV.
500 For example, intuition tells you that this XS code:
503 av_store( av, 0, &PL_sv_undef );
505 is equivalent to this Perl code:
510 Unfortunately, this isn't true. AVs use C<&PL_sv_undef> as a marker
511 for indicating that an array element has not yet been initialized.
512 Thus, C<exists $av[0]> would be true for the above Perl code, but
513 false for the array generated by the XS code.
515 Other problems can occur when storing C<&PL_sv_undef> in HVs:
517 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
519 This will indeed make the value C<undef>, but if you try to modify
520 the value of C<key>, you'll get the following error:
522 Modification of non-creatable hash value attempted
524 In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
525 in restricted hashes. This caused such hash entries not to appear
526 when iterating over the hash or when checking for the keys
527 with the C<hv_exists> function.
529 You can run into similar problems when you store C<&PL_sv_true> or
530 C<&PL_sv_false> into AVs or HVs. Trying to modify such elements
531 will give you the following error:
533 Modification of a read-only value attempted
535 To make a long story short, you can use the special variables
536 C<&PL_sv_undef>, C<&PL_sv_true> and C<&PL_sv_false> with AVs and
537 HVs, but you have to make sure you know what you're doing.
539 Generally, if you want to store an undefined value in an AV
540 or HV, you should not use C<&PL_sv_undef>, but rather create a
541 new undefined value using the C<newSV> function, for example:
543 av_store( av, 42, newSV(0) );
544 hv_store( hv, "foo", 3, newSV(0), 0 );
548 References are a special type of scalar that point to other data types
549 (including references).
551 To create a reference, use either of the following functions:
553 SV* newRV_inc((SV*) thing);
554 SV* newRV_noinc((SV*) thing);
556 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
557 functions are identical except that C<newRV_inc> increments the reference
558 count of the C<thing>, while C<newRV_noinc> does not. For historical
559 reasons, C<newRV> is a synonym for C<newRV_inc>.
561 Once you have a reference, you can use the following macro to dereference
566 then call the appropriate routines, casting the returned C<SV*> to either an
567 C<AV*> or C<HV*>, if required.
569 To determine if an SV is a reference, you can use the following macro:
573 To discover what type of value the reference refers to, use the following
574 macro and then check the return value.
578 The most useful types that will be returned are:
587 SVt_PVGV Glob (possible a file handle)
588 SVt_PVMG Blessed or Magical Scalar
590 See the sv.h header file for more details.
592 =head2 Blessed References and Class Objects
594 References are also used to support object-oriented programming. In perl's
595 OO lexicon, an object is simply a reference that has been blessed into a
596 package (or class). Once blessed, the programmer may now use the reference
597 to access the various methods in the class.
599 A reference can be blessed into a package with the following function:
601 SV* sv_bless(SV* sv, HV* stash);
603 The C<sv> argument must be a reference value. The C<stash> argument
604 specifies which class the reference will belong to. See
605 L<Stashes and Globs> for information on converting class names into stashes.
607 /* Still under construction */
609 Upgrades rv to reference if not already one. Creates new SV for rv to
610 point to. If C<classname> is non-null, the SV is blessed into the specified
611 class. SV is returned.
613 SV* newSVrv(SV* rv, const char* classname);
615 Copies integer, unsigned integer or double into an SV whose reference is C<rv>. SV is blessed
616 if C<classname> is non-null.
618 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
619 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
620 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
622 Copies the pointer value (I<the address, not the string!>) into an SV whose
623 reference is rv. SV is blessed if C<classname> is non-null.
625 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
627 Copies string into an SV whose reference is C<rv>. Set length to 0 to let
628 Perl calculate the string length. SV is blessed if C<classname> is non-null.
630 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
632 Tests whether the SV is blessed into the specified class. It does not
633 check inheritance relationships.
635 int sv_isa(SV* sv, const char* name);
637 Tests whether the SV is a reference to a blessed object.
639 int sv_isobject(SV* sv);
641 Tests whether the SV is derived from the specified class. SV can be either
642 a reference to a blessed object or a string containing a class name. This
643 is the function implementing the C<UNIVERSAL::isa> functionality.
645 bool sv_derived_from(SV* sv, const char* name);
647 To check if you've got an object derived from a specific class you have
650 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
652 =head2 Creating New Variables
654 To create a new Perl variable with an undef value which can be accessed from
655 your Perl script, use the following routines, depending on the variable type.
657 SV* get_sv("package::varname", TRUE);
658 AV* get_av("package::varname", TRUE);
659 HV* get_hv("package::varname", TRUE);
661 Notice the use of TRUE as the second parameter. The new variable can now
662 be set, using the routines appropriate to the data type.
664 There are additional macros whose values may be bitwise OR'ed with the
665 C<TRUE> argument to enable certain extra features. Those bits are:
671 Marks the variable as multiply defined, thus preventing the:
673 Name <varname> used only once: possible typo
681 Had to create <varname> unexpectedly
683 if the variable did not exist before the function was called.
687 If you do not specify a package name, the variable is created in the current
690 =head2 Reference Counts and Mortality
692 Perl uses a reference count-driven garbage collection mechanism. SVs,
693 AVs, or HVs (xV for short in the following) start their life with a
694 reference count of 1. If the reference count of an xV ever drops to 0,
695 then it will be destroyed and its memory made available for reuse.
697 This normally doesn't happen at the Perl level unless a variable is
698 undef'ed or the last variable holding a reference to it is changed or
699 overwritten. At the internal level, however, reference counts can be
700 manipulated with the following macros:
702 int SvREFCNT(SV* sv);
703 SV* SvREFCNT_inc(SV* sv);
704 void SvREFCNT_dec(SV* sv);
706 However, there is one other function which manipulates the reference
707 count of its argument. The C<newRV_inc> function, you will recall,
708 creates a reference to the specified argument. As a side effect,
709 it increments the argument's reference count. If this is not what
710 you want, use C<newRV_noinc> instead.
712 For example, imagine you want to return a reference from an XSUB function.
713 Inside the XSUB routine, you create an SV which initially has a reference
714 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
715 This returns the reference as a new SV, but the reference count of the
716 SV you passed to C<newRV_inc> has been incremented to two. Now you
717 return the reference from the XSUB routine and forget about the SV.
718 But Perl hasn't! Whenever the returned reference is destroyed, the
719 reference count of the original SV is decreased to one and nothing happens.
720 The SV will hang around without any way to access it until Perl itself
721 terminates. This is a memory leak.
723 The correct procedure, then, is to use C<newRV_noinc> instead of
724 C<newRV_inc>. Then, if and when the last reference is destroyed,
725 the reference count of the SV will go to zero and it will be destroyed,
726 stopping any memory leak.
728 There are some convenience functions available that can help with the
729 destruction of xVs. These functions introduce the concept of "mortality".
730 An xV that is mortal has had its reference count marked to be decremented,
731 but not actually decremented, until "a short time later". Generally the
732 term "short time later" means a single Perl statement, such as a call to
733 an XSUB function. The actual determinant for when mortal xVs have their
734 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
735 See L<perlcall> and L<perlxs> for more details on these macros.
737 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
738 However, if you mortalize a variable twice, the reference count will
739 later be decremented twice.
741 "Mortal" SVs are mainly used for SVs that are placed on perl's stack.
742 For example an SV which is created just to pass a number to a called sub
743 is made mortal to have it cleaned up automatically when it's popped off
744 the stack. Similarly, results returned by XSUBs (which are pushed on the
745 stack) are often made mortal.
747 To create a mortal variable, use the functions:
751 SV* sv_mortalcopy(SV*)
753 The first call creates a mortal SV (with no value), the second converts an existing
754 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
755 third creates a mortal copy of an existing SV.
756 Because C<sv_newmortal> gives the new SV no value,it must normally be given one
757 via C<sv_setpv>, C<sv_setiv>, etc. :
759 SV *tmp = sv_newmortal();
760 sv_setiv(tmp, an_integer);
762 As that is multiple C statements it is quite common so see this idiom instead:
764 SV *tmp = sv_2mortal(newSViv(an_integer));
767 You should be careful about creating mortal variables. Strange things
768 can happen if you make the same value mortal within multiple contexts,
769 or if you make a variable mortal multiple times. Thinking of "Mortalization"
770 as deferred C<SvREFCNT_dec> should help to minimize such problems.
771 For example if you are passing an SV which you I<know> has high enough REFCNT
772 to survive its use on the stack you need not do any mortalization.
773 If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
774 making a C<sv_mortalcopy> is safer.
776 The mortal routines are not just for SVs -- AVs and HVs can be
777 made mortal by passing their address (type-casted to C<SV*>) to the
778 C<sv_2mortal> or C<sv_mortalcopy> routines.
780 =head2 Stashes and Globs
782 A B<stash> is a hash that contains all variables that are defined
783 within a package. Each key of the stash is a symbol
784 name (shared by all the different types of objects that have the same
785 name), and each value in the hash table is a GV (Glob Value). This GV
786 in turn contains references to the various objects of that name,
787 including (but not limited to) the following:
796 There is a single stash called C<PL_defstash> that holds the items that exist
797 in the C<main> package. To get at the items in other packages, append the
798 string "::" to the package name. The items in the C<Foo> package are in
799 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
800 in the stash C<Baz::> in C<Bar::>'s stash.
802 To get the stash pointer for a particular package, use the function:
804 HV* gv_stashpv(const char* name, I32 create)
805 HV* gv_stashsv(SV*, I32 create)
807 The first function takes a literal string, the second uses the string stored
808 in the SV. Remember that a stash is just a hash table, so you get back an
809 C<HV*>. The C<create> flag will create a new package if it is set.
811 The name that C<gv_stash*v> wants is the name of the package whose symbol table
812 you want. The default package is called C<main>. If you have multiply nested
813 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
816 Alternately, if you have an SV that is a blessed reference, you can find
817 out the stash pointer by using:
819 HV* SvSTASH(SvRV(SV*));
821 then use the following to get the package name itself:
823 char* HvNAME(HV* stash);
825 If you need to bless or re-bless an object you can use the following
828 SV* sv_bless(SV*, HV* stash)
830 where the first argument, an C<SV*>, must be a reference, and the second
831 argument is a stash. The returned C<SV*> can now be used in the same way
834 For more information on references and blessings, consult L<perlref>.
836 =head2 Double-Typed SVs
838 Scalar variables normally contain only one type of value, an integer,
839 double, pointer, or reference. Perl will automatically convert the
840 actual scalar data from the stored type into the requested type.
842 Some scalar variables contain more than one type of scalar data. For
843 example, the variable C<$!> contains either the numeric value of C<errno>
844 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
846 To force multiple data values into an SV, you must do two things: use the
847 C<sv_set*v> routines to add the additional scalar type, then set a flag
848 so that Perl will believe it contains more than one type of data. The
849 four macros to set the flags are:
856 The particular macro you must use depends on which C<sv_set*v> routine
857 you called first. This is because every C<sv_set*v> routine turns on
858 only the bit for the particular type of data being set, and turns off
861 For example, to create a new Perl variable called "dberror" that contains
862 both the numeric and descriptive string error values, you could use the
866 extern char *dberror_list;
868 SV* sv = get_sv("dberror", TRUE);
869 sv_setiv(sv, (IV) dberror);
870 sv_setpv(sv, dberror_list[dberror]);
873 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
874 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
876 =head2 Magic Variables
878 [This section still under construction. Ignore everything here. Post no
879 bills. Everything not permitted is forbidden.]
881 Any SV may be magical, that is, it has special features that a normal
882 SV does not have. These features are stored in the SV structure in a
883 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
896 Note this is current as of patchlevel 0, and could change at any time.
898 =head2 Assigning Magic
900 Perl adds magic to an SV using the sv_magic function:
902 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
904 The C<sv> argument is a pointer to the SV that is to acquire a new magical
907 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
908 convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
909 to the beginning of the linked list of magical features. Any prior entry
910 of the same type of magic is deleted. Note that this can be overridden,
911 and multiple instances of the same type of magic can be associated with an
914 The C<name> and C<namlen> arguments are used to associate a string with
915 the magic, typically the name of a variable. C<namlen> is stored in the
916 C<mg_len> field and if C<name> is non-null and C<namlen> E<gt>= 0 a malloc'd
917 copy of the name is stored in C<mg_ptr> field.
919 The 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.
921 See the L<Magic Virtual Tables> section below. The C<how> argument is also
922 stored in the C<mg_type> field. The value of C<how> should be chosen
923 from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
924 these macros were added, Perl internals used to directly use character
925 literals, so you may occasionally come across old code or documentation
926 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
928 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
929 structure. If it is not the same as the C<sv> argument, the reference
930 count of the C<obj> object is incremented. If it is the same, or if
931 the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
932 then C<obj> is merely stored, without the reference count being incremented.
934 There is also a function to add magic to an C<HV>:
936 void hv_magic(HV *hv, GV *gv, int how);
938 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
940 To remove the magic from an SV, call the function sv_unmagic:
942 void sv_unmagic(SV *sv, int type);
944 The C<type> argument should be equal to the C<how> value when the C<SV>
945 was initially made magical.
947 =head2 Magic Virtual Tables
949 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
950 C<MGVTBL>, which is a structure of function pointers and stands for
951 "Magic Virtual Table" to handle the various operations that might be
952 applied to that variable.
954 The C<MGVTBL> has five pointers to the following routine types:
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);
962 This MGVTBL structure is set at compile-time in F<perl.h> and there are
963 currently 19 types (or 21 with overloading turned on). These different
964 structures contain pointers to various routines that perform additional
965 actions depending on which function is being called.
967 Function pointer Action taken
968 ---------------- ------------
969 svt_get Do something before the value of the SV is retrieved.
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.
975 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
976 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
978 { magic_get, magic_set, magic_len, 0, 0 }
980 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
981 if a get operation is being performed, the routine C<magic_get> is
982 called. All the various routines for the various magical types begin
983 with C<magic_>. NOTE: the magic routines are not considered part of
984 the Perl API, and may not be exported by the Perl library.
986 The current kinds of Magic Virtual Tables are:
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)
996 B PERL_MAGIC_bm vtbl_bm Boyer-Moore (fast string search)
997 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
999 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
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 ???
1011 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale collate transformation
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
1021 V PERL_MAGIC_vstring (none) v-string scalars
1022 w PERL_MAGIC_utf8 vtbl_utf8 UTF-8 length+offset cache
1023 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1024 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1025 variable / smart parameter
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
1033 When an uppercase and lowercase letter both exist in the table, then the
1034 uppercase 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
1036 of that composite type. Some internals code makes use of this case
1037 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
1039 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1040 specifically for use by extensions and will not be used by perl itself.
1041 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1042 to variables (typically objects). This is especially useful because
1043 there is no way for normal perl code to corrupt this private information
1044 (unlike using extra elements of a hash object).
1046 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1047 C function any time a scalar's value is used or changed. The C<MAGIC>'s
1048 C<mg_ptr> field points to a C<ufuncs> structure:
1051 I32 (*uf_val)(pTHX_ IV, SV*);
1052 I32 (*uf_set)(pTHX_ IV, SV*);
1056 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1057 function will be called with C<uf_index> as the first arg and a pointer to
1058 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1059 magic is shown below. Note that the ufuncs structure is copied by
1060 sv_magic, so you can safely allocate it on the stack.
1068 uf.uf_val = &my_get_fn;
1069 uf.uf_set = &my_set_fn;
1071 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1073 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1074 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1075 extra care to avoid conflict. Typically only using the magic on
1076 objects blessed into the same class as the extension is sufficient.
1077 For 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.
1080 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1081 earlier do B<not> invoke 'set' magic on their targets. This must
1082 be done by the user either by calling the C<SvSETMAGIC()> macro after
1083 calling these functions, or by using one of the C<sv_set*_mg()> or
1084 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1085 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1086 obtained from external sources in functions that don't handle magic.
1087 See L<perlapi> for a description of these functions.
1088 For example, calls to the C<sv_cat*()> functions typically need to be
1089 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1090 since their implementation handles 'get' magic.
1092 =head2 Finding Magic
1094 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
1096 This routine returns a pointer to the C<MAGIC> structure stored in the SV.
1097 If the SV does not have that magical feature, C<NULL> is returned. Also,
1098 if the SV is not of type SVt_PVMG, Perl may core dump.
1100 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1102 This routine checks to see what types of magic C<sv> has. If the mg_type
1103 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1104 the mg_type field is changed to be the lowercase letter.
1106 =head2 Understanding the Magic of Tied Hashes and Arrays
1108 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1111 WARNING: As of the 5.004 release, proper usage of the array and hash
1112 access functions requires understanding a few caveats. Some
1113 of these caveats are actually considered bugs in the API, to be fixed
1114 in later releases, and are bracketed with [MAYCHANGE] below. If
1115 you find yourself actually applying such information in this section, be
1116 aware that the behavior may change in the future, umm, without warning.
1118 The perl tie function associates a variable with an object that implements
1119 the various GET, SET, etc methods. To perform the equivalent of the perl
1120 tie function from an XSUB, you must mimic this behaviour. The code below
1121 carries out the necessary steps - firstly it creates a new hash, and then
1122 creates a second hash which it blesses into the class which will implement
1123 the tie methods. Lastly it ties the two hashes together, and returns a
1124 reference to the new tied hash. Note that the code below does NOT call the
1125 TIEHASH method in the MyTie class -
1126 see L<Calling Perl Routines from within C Programs> for details on how
1137 tie = newRV_noinc((SV*)newHV());
1138 stash = gv_stashpv("MyTie", TRUE);
1139 sv_bless(tie, stash);
1140 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1141 RETVAL = newRV_noinc(hash);
1145 The C<av_store> function, when given a tied array argument, merely
1146 copies the magic of the array onto the value to be "stored", using
1147 C<mg_copy>. It may also return NULL, indicating that the value did not
1148 actually need to be stored in the array. [MAYCHANGE] After a call to
1149 C<av_store> on a tied array, the caller will usually need to call
1150 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1151 TIEARRAY object. If C<av_store> did return NULL, a call to
1152 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1155 The previous paragraph is applicable verbatim to tied hash access using the
1156 C<hv_store> and C<hv_store_ent> functions as well.
1158 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1159 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1160 has been initialized using C<mg_copy>. Note the value so returned does not
1161 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1162 need to call C<mg_get()> on the returned value in order to actually invoke
1163 the perl level "FETCH" method on the underlying TIE object. Similarly,
1164 you may also call C<mg_set()> on the return value after possibly assigning
1165 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1166 method on the TIE object. [/MAYCHANGE]
1169 In other words, the array or hash fetch/store functions don't really
1170 fetch and store actual values in the case of tied arrays and hashes. They
1171 merely 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
1173 do the job of invoking the TIE methods on the underlying objects. Thus
1174 the magic mechanism currently implements a kind of lazy access to arrays
1177 Currently (as of perl version 5.004), use of the hash and array access
1178 functions requires the user to be aware of whether they are operating on
1179 "normal" hashes and arrays, or on their tied variants. The API may be
1180 changed to provide more transparent access to both tied and normal data
1181 types in future versions.
1184 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1185 are mere sugar to invoke some perl method calls while using the uniform hash
1186 and array syntax. The use of this sugar imposes some overhead (typically
1187 about two to four extra opcodes per FETCH/STORE operation, in addition to
1188 the creation of all the mortal variables required to invoke the methods).
1189 This overhead will be comparatively small if the TIE methods are themselves
1190 substantial, but if they are only a few statements long, the overhead
1191 will not be insignificant.
1193 =head2 Localizing changes
1195 Perl has a very handy construction
1202 This construction is I<approximately> equivalent to
1211 The biggest difference is that the first construction would
1212 reinstate the initial value of $var, irrespective of how control exits
1213 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1214 more efficient as well.
1216 There is a way to achieve a similar task from C via Perl API: create a
1217 I<pseudo-block>, and arrange for some changes to be automatically
1218 undone at the end of it, either explicit, or via a non-local exit (via
1219 die()). A I<block>-like construct is created by a pair of
1220 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1221 Such a construct may be created specially for some important localized
1222 task, or an existing one (like boundaries of enclosing Perl
1223 subroutine/block, or an existing pair for freeing TMPs) may be
1224 used. (In the second case the overhead of additional localization must
1225 be almost negligible.) Note that any XSUB is automatically enclosed in
1226 an C<ENTER>/C<LEAVE> pair.
1228 Inside such a I<pseudo-block> the following service is available:
1232 =item C<SAVEINT(int i)>
1234 =item C<SAVEIV(IV i)>
1236 =item C<SAVEI32(I32 i)>
1238 =item C<SAVELONG(long i)>
1240 These macros arrange things to restore the value of integer variable
1241 C<i> at the end of enclosing I<pseudo-block>.
1243 =item C<SAVESPTR(s)>
1245 =item C<SAVEPPTR(p)>
1247 These macros arrange things to restore the value of pointers C<s> and
1248 C<p>. C<s> must be a pointer of a type which survives conversion to
1249 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1252 =item C<SAVEFREESV(SV *sv)>
1254 The refcount of C<sv> would be decremented at the end of
1255 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1256 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1257 extends the lifetime of C<sv> until the beginning of the next statement,
1258 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1259 lifetimes can be wildly different.
1261 Also compare C<SAVEMORTALIZESV>.
1263 =item C<SAVEMORTALIZESV(SV *sv)>
1265 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1266 scope instead of decrementing its reference count. This usually has the
1267 effect of keeping C<sv> alive until the statement that called the currently
1268 live scope has finished executing.
1270 =item C<SAVEFREEOP(OP *op)>
1272 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1274 =item C<SAVEFREEPV(p)>
1276 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1277 end of I<pseudo-block>.
1279 =item C<SAVECLEARSV(SV *sv)>
1281 Clears a slot in the current scratchpad which corresponds to C<sv> at
1282 the end of I<pseudo-block>.
1284 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1286 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1287 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1288 short-lived storage, the corresponding string may be reallocated like
1291 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1293 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1295 At the end of I<pseudo-block> the function C<f> is called with the
1298 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1300 At the end of I<pseudo-block> the function C<f> is called with the
1301 implicit context argument (if any), and C<p>.
1303 =item C<SAVESTACK_POS()>
1305 The current offset on the Perl internal stack (cf. C<SP>) is restored
1306 at the end of I<pseudo-block>.
1310 The following API list contains functions, thus one needs to
1311 provide pointers to the modifiable data explicitly (either C pointers,
1312 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1313 function takes C<int *>.
1317 =item C<SV* save_scalar(GV *gv)>
1319 Equivalent to Perl code C<local $gv>.
1321 =item C<AV* save_ary(GV *gv)>
1323 =item C<HV* save_hash(GV *gv)>
1325 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1327 =item C<void save_item(SV *item)>
1329 Duplicates the current value of C<SV>, on the exit from the current
1330 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1331 using the stored value. It doesn't handle magic. Use C<save_scalar> if
1334 =item C<void save_list(SV **sarg, I32 maxsarg)>
1336 A variant of C<save_item> which takes multiple arguments via an array
1337 C<sarg> of C<SV*> of length C<maxsarg>.
1339 =item C<SV* save_svref(SV **sptr)>
1341 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1343 =item C<void save_aptr(AV **aptr)>
1345 =item C<void save_hptr(HV **hptr)>
1347 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1351 The C<Alias> module implements localization of the basic types within the
1352 I<caller's scope>. People who are interested in how to localize things in
1353 the containing scope should take a look there too.
1357 =head2 XSUBs and the Argument Stack
1359 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1360 An XSUB routine will have a stack that contains the arguments from the Perl
1361 program, and a way to map from the Perl data structures to a C equivalent.
1363 The stack arguments are accessible through the C<ST(n)> macro, which returns
1364 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1365 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1368 Most of the time, output from the C routine can be handled through use of
1369 the RETVAL and OUTPUT directives. However, there are some cases where the
1370 argument stack is not already long enough to handle all the return values.
1371 An example is the POSIX tzname() call, which takes no arguments, but returns
1372 two, the local time zone's standard and summer time abbreviations.
1374 To handle this situation, the PPCODE directive is used and the stack is
1375 extended using the macro:
1379 where C<SP> is the macro that represents the local copy of the stack pointer,
1380 and C<num> is the number of elements the stack should be extended by.
1382 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1383 macro. The pushed values will often need to be "mortal" (See
1384 L</Reference Counts and Mortality>):
1386 PUSHs(sv_2mortal(newSViv(an_integer)))
1387 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1388 PUSHs(sv_2mortal(newSVnv(a_double)))
1389 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1391 And now the Perl program calling C<tzname>, the two values will be assigned
1394 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1396 An alternate (and possibly simpler) method to pushing values on the stack is
1401 This macro automatically adjust the stack for you, if needed. Thus, you
1402 do not need to call C<EXTEND> to extend the stack.
1404 Despite their suggestions in earlier versions of this document the macros
1405 C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1406 For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1407 C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
1409 For more information, consult L<perlxs> and L<perlxstut>.
1411 =head2 Calling Perl Routines from within C Programs
1413 There are four routines that can be used to call a Perl subroutine from
1414 within a C program. These four are:
1416 I32 call_sv(SV*, I32);
1417 I32 call_pv(const char*, I32);
1418 I32 call_method(const char*, I32);
1419 I32 call_argv(const char*, I32, register char**);
1421 The routine most often used is C<call_sv>. The C<SV*> argument
1422 contains either the name of the Perl subroutine to be called, or a
1423 reference to the subroutine. The second argument consists of flags
1424 that control the context in which the subroutine is called, whether
1425 or not the subroutine is being passed arguments, how errors should be
1426 trapped, and how to treat return values.
1428 All four routines return the number of arguments that the subroutine returned
1431 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1432 but those names are now deprecated; macros of the same name are provided for
1435 When using any of these routines (except C<call_argv>), the programmer
1436 must manipulate the Perl stack. These include the following macros and
1451 For a detailed description of calling conventions from C to Perl,
1452 consult L<perlcall>.
1454 =head2 Memory Allocation
1458 All memory meant to be used with the Perl API functions should be manipulated
1459 using the macros described in this section. The macros provide the necessary
1460 transparency between differences in the actual malloc implementation that is
1463 It is suggested that you enable the version of malloc that is distributed
1464 with Perl. It keeps pools of various sizes of unallocated memory in
1465 order to satisfy allocation requests more quickly. However, on some
1466 platforms, it may cause spurious malloc or free errors.
1468 The following three macros are used to initially allocate memory :
1470 New(x, pointer, number, type);
1471 Newc(x, pointer, number, type, cast);
1472 Newz(x, pointer, number, type);
1474 The first argument C<x> was a "magic cookie" that was used to keep track
1475 of who called the macro, to help when debugging memory problems. However,
1476 the current code makes no use of this feature (most Perl developers now
1477 use run-time memory checkers), so this argument can be any number.
1479 The second argument C<pointer> should be the name of a variable that will
1480 point to the newly allocated memory.
1482 The third and fourth arguments C<number> and C<type> specify how many of
1483 the specified type of data structure should be allocated. The argument
1484 C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1485 should be used if the C<pointer> argument is different from the C<type>
1488 Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1489 to zero out all the newly allocated memory.
1493 Renew(pointer, number, type);
1494 Renewc(pointer, number, type, cast);
1497 These three macros are used to change a memory buffer size or to free a
1498 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1499 match those of C<New> and C<Newc> with the exception of not needing the
1500 "magic cookie" argument.
1504 Move(source, dest, number, type);
1505 Copy(source, dest, number, type);
1506 Zero(dest, number, type);
1508 These three macros are used to move, copy, or zero out previously allocated
1509 memory. The C<source> and C<dest> arguments point to the source and
1510 destination starting points. Perl will move, copy, or zero out C<number>
1511 instances of the size of the C<type> data structure (using the C<sizeof>
1516 The most recent development releases of Perl has been experimenting with
1517 removing Perl's dependency on the "normal" standard I/O suite and allowing
1518 other stdio implementations to be used. This involves creating a new
1519 abstraction layer that then calls whichever implementation of stdio Perl
1520 was compiled with. All XSUBs should now use the functions in the PerlIO
1521 abstraction layer and not make any assumptions about what kind of stdio
1524 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1526 =head2 Putting a C value on Perl stack
1528 A lot of opcodes (this is an elementary operation in the internal perl
1529 stack machine) put an SV* on the stack. However, as an optimization
1530 the corresponding SV is (usually) not recreated each time. The opcodes
1531 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1532 not constantly freed/created.
1534 Each of the targets is created only once (but see
1535 L<Scratchpads and recursion> below), and when an opcode needs to put
1536 an integer, a double, or a string on stack, it just sets the
1537 corresponding parts of its I<target> and puts the I<target> on stack.
1539 The macro to put this target on stack is C<PUSHTARG>, and it is
1540 directly used in some opcodes, as well as indirectly in zillions of
1541 others, which use it via C<(X)PUSH[iunp]>.
1543 Because the target is reused, you must be careful when pushing multiple
1544 values on the stack. The following code will not do what you think:
1549 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1550 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1551 At the end of the operation, the stack does not contain the values 10
1552 and 20, but actually contains two pointers to C<TARG>, which we have set
1555 If you need to push multiple different values then you should either use
1556 the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1557 none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1558 SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1559 will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1560 this a little easier to achieve by creating a new mortal for you (via
1561 C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1562 in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1563 Thus, instead of writing this to "fix" the example above:
1565 XPUSHs(sv_2mortal(newSViv(10)))
1566 XPUSHs(sv_2mortal(newSViv(20)))
1568 you can simply write:
1573 On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1574 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
1575 macros can make use of the local variable C<TARG>. See also C<dTARGET>
1580 The question remains on when the SVs which are I<target>s for opcodes
1581 are created. The answer is that they are created when the current unit --
1582 a subroutine or a file (for opcodes for statements outside of
1583 subroutines) -- is compiled. During this time a special anonymous Perl
1584 array is created, which is called a scratchpad for the current
1587 A scratchpad keeps SVs which are lexicals for the current unit and are
1588 targets for opcodes. One can deduce that an SV lives on a scratchpad
1589 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1590 I<target>s have C<SVs_PADTMP> set.
1592 The correspondence between OPs and I<target>s is not 1-to-1. Different
1593 OPs in the compile tree of the unit can use the same target, if this
1594 would not conflict with the expected life of the temporary.
1596 =head2 Scratchpads and recursion
1598 In fact it is not 100% true that a compiled unit contains a pointer to
1599 the scratchpad AV. In fact it contains a pointer to an AV of
1600 (initially) one element, and this element is the scratchpad AV. Why do
1601 we need an extra level of indirection?
1603 The answer is B<recursion>, and maybe B<threads>. Both
1604 these can create several execution pointers going into the same
1605 subroutine. For the subroutine-child not write over the temporaries
1606 for the subroutine-parent (lifespan of which covers the call to the
1607 child), the parent and the child should have different
1608 scratchpads. (I<And> the lexicals should be separate anyway!)
1610 So each subroutine is born with an array of scratchpads (of length 1).
1611 On each entry to the subroutine it is checked that the current
1612 depth of the recursion is not more than the length of this array, and
1613 if it is, new scratchpad is created and pushed into the array.
1615 The I<target>s on this scratchpad are C<undef>s, but they are already
1616 marked with correct flags.
1618 =head1 Compiled code
1622 Here we describe the internal form your code is converted to by
1623 Perl. Start with a simple example:
1627 This is converted to a tree similar to this one:
1635 (but slightly more complicated). This tree reflects the way Perl
1636 parsed your code, but has nothing to do with the execution order.
1637 There is an additional "thread" going through the nodes of the tree
1638 which shows the order of execution of the nodes. In our simplified
1639 example above it looks like:
1641 $b ---> $c ---> + ---> $a ---> assign-to
1643 But with the actual compile tree for C<$a = $b + $c> it is different:
1644 some nodes I<optimized away>. As a corollary, though the actual tree
1645 contains more nodes than our simplified example, the execution order
1646 is the same as in our example.
1648 =head2 Examining the tree
1650 If you have your perl compiled for debugging (usually done with
1651 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
1652 compiled tree by specifying C<-Dx> on the Perl command line. The
1653 output takes several lines per node, and for C<$b+$c> it looks like
1658 FLAGS = (SCALAR,KIDS)
1660 TYPE = null ===> (4)
1662 FLAGS = (SCALAR,KIDS)
1664 3 TYPE = gvsv ===> 4
1670 TYPE = null ===> (5)
1672 FLAGS = (SCALAR,KIDS)
1674 4 TYPE = gvsv ===> 5
1680 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1681 not optimized away (one per number in the left column). The immediate
1682 children of the given node correspond to C<{}> pairs on the same level
1683 of indentation, thus this listing corresponds to the tree:
1691 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1692 4 5 6> (node C<6> is not included into above listing), i.e.,
1693 C<gvsv gvsv add whatever>.
1695 Each of these nodes represents an op, a fundamental operation inside the
1696 Perl core. The code which implements each operation can be found in the
1697 F<pp*.c> files; the function which implements the op with type C<gvsv>
1698 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1699 different numbers of children: C<add> is a binary operator, as one would
1700 expect, and so has two children. To accommodate the various different
1701 numbers of children, there are various types of op data structure, and
1702 they link together in different ways.
1704 The simplest type of op structure is C<OP>: this has no children. Unary
1705 operators, C<UNOP>s, have one child, and this is pointed to by the
1706 C<op_first> field. Binary operators (C<BINOP>s) have not only an
1707 C<op_first> field but also an C<op_last> field. The most complex type of
1708 op is a C<LISTOP>, which has any number of children. In this case, the
1709 first child is pointed to by C<op_first> and the last child by
1710 C<op_last>. The children in between can be found by iteratively
1711 following the C<op_sibling> pointer from the first child to the last.
1713 There are also two other op types: a C<PMOP> holds a regular expression,
1714 and has no children, and a C<LOOP> may or may not have children. If the
1715 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1716 complicate matters, if a C<UNOP> is actually a C<null> op after
1717 optimization (see L</Compile pass 2: context propagation>) it will still
1718 have children in accordance with its former type.
1720 Another way to examine the tree is to use a compiler back-end module, such
1723 =head2 Compile pass 1: check routines
1725 The tree is created by the compiler while I<yacc> code feeds it
1726 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1727 the first pass of perl compilation.
1729 What makes this pass interesting for perl developers is that some
1730 optimization may be performed on this pass. This is optimization by
1731 so-called "check routines". The correspondence between node names
1732 and corresponding check routines is described in F<opcode.pl> (do not
1733 forget to run C<make regen_headers> if you modify this file).
1735 A check routine is called when the node is fully constructed except
1736 for the execution-order thread. Since at this time there are no
1737 back-links to the currently constructed node, one can do most any
1738 operation to the top-level node, including freeing it and/or creating
1739 new nodes above/below it.
1741 The check routine returns the node which should be inserted into the
1742 tree (if the top-level node was not modified, check routine returns
1745 By convention, check routines have names C<ck_*>. They are usually
1746 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1747 called from F<perly.y>).
1749 =head2 Compile pass 1a: constant folding
1751 Immediately after the check routine is called the returned node is
1752 checked for being compile-time executable. If it is (the value is
1753 judged to be constant) it is immediately executed, and a I<constant>
1754 node with the "return value" of the corresponding subtree is
1755 substituted instead. The subtree is deleted.
1757 If constant folding was not performed, the execution-order thread is
1760 =head2 Compile pass 2: context propagation
1762 When a context for a part of compile tree is known, it is propagated
1763 down through the tree. At this time the context can have 5 values
1764 (instead of 2 for runtime context): void, boolean, scalar, list, and
1765 lvalue. In contrast with the pass 1 this pass is processed from top
1766 to bottom: a node's context determines the context for its children.
1768 Additional context-dependent optimizations are performed at this time.
1769 Since at this moment the compile tree contains back-references (via
1770 "thread" pointers), nodes cannot be free()d now. To allow
1771 optimized-away nodes at this stage, such nodes are null()ified instead
1772 of free()ing (i.e. their type is changed to OP_NULL).
1774 =head2 Compile pass 3: peephole optimization
1776 After the compile tree for a subroutine (or for an C<eval> or a file)
1777 is created, an additional pass over the code is performed. This pass
1778 is neither top-down or bottom-up, but in the execution order (with
1779 additional complications for conditionals). These optimizations are
1780 done in the subroutine peep(). Optimizations performed at this stage
1781 are subject to the same restrictions as in the pass 2.
1783 =head2 Pluggable runops
1785 The compile tree is executed in a runops function. There are two runops
1786 functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1787 with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1788 control over the execution of the compile tree it is possible to provide
1789 your own runops function.
1791 It's probably best to copy one of the existing runops functions and
1792 change it to suit your needs. Then, in the BOOT section of your XS
1795 PL_runops = my_runops;
1797 This function should be as efficient as possible to keep your programs
1798 running as fast as possible.
1800 =head1 Examining internal data structures with the C<dump> functions
1802 To aid debugging, the source file F<dump.c> contains a number of
1803 functions which produce formatted output of internal data structures.
1805 The most commonly used of these functions is C<Perl_sv_dump>; it's used
1806 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
1807 C<sv_dump> to produce debugging output from Perl-space, so users of that
1808 module should already be familiar with its format.
1810 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
1811 derivatives, and produces output similar to C<perl -Dx>; in fact,
1812 C<Perl_dump_eval> will dump the main root of the code being evaluated,
1813 exactly like C<-Dx>.
1815 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
1816 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
1817 subroutines in a package like so: (Thankfully, these are all xsubs, so
1818 there is no op tree)
1820 (gdb) print Perl_dump_packsubs(PL_defstash)
1822 SUB attributes::bootstrap = (xsub 0x811fedc 0)
1824 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
1826 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
1828 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
1830 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
1832 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
1833 the op tree of the main root.
1835 =head1 How multiple interpreters and concurrency are supported
1837 =head2 Background and PERL_IMPLICIT_CONTEXT
1839 The Perl interpreter can be regarded as a closed box: it has an API
1840 for feeding it code or otherwise making it do things, but it also has
1841 functions for its own use. This smells a lot like an object, and
1842 there are ways for you to build Perl so that you can have multiple
1843 interpreters, with one interpreter represented either as a C structure,
1844 or inside a thread-specific structure. These structures contain all
1845 the context, the state of that interpreter.
1847 Two macros control the major Perl build flavors: MULTIPLICITY and
1848 USE_5005THREADS. The MULTIPLICITY build has a C structure
1849 that packages all the interpreter state, and there is a similar thread-specific
1850 data structure under USE_5005THREADS. In both cases,
1851 PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1852 support for passing in a "hidden" first argument that represents all three
1855 All this obviously requires a way for the Perl internal functions to be
1856 either subroutines taking some kind of structure as the first
1857 argument, or subroutines taking nothing as the first argument. To
1858 enable these two very different ways of building the interpreter,
1859 the Perl source (as it does in so many other situations) makes heavy
1860 use of macros and subroutine naming conventions.
1862 First problem: deciding which functions will be public API functions and
1863 which will be private. All functions whose names begin C<S_> are private
1864 (think "S" for "secret" or "static"). All other functions begin with
1865 "Perl_", but just because a function begins with "Perl_" does not mean it is
1866 part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
1867 function is part of the API is to find its entry in L<perlapi>.
1868 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
1869 think it should be (i.e., you need it for your extension), send mail via
1870 L<perlbug> explaining why you think it should be.
1872 Second problem: there must be a syntax so that the same subroutine
1873 declarations and calls can pass a structure as their first argument,
1874 or pass nothing. To solve this, the subroutines are named and
1875 declared in a particular way. Here's a typical start of a static
1876 function used within the Perl guts:
1879 S_incline(pTHX_ char *s)
1881 STATIC becomes "static" in C, and may be #define'd to nothing in some
1882 configurations in future.
1884 A public function (i.e. part of the internal API, but not necessarily
1885 sanctioned for use in extensions) begins like this:
1888 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
1890 C<pTHX_> is one of a number of macros (in perl.h) that hide the
1891 details of the interpreter's context. THX stands for "thread", "this",
1892 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1893 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
1894 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
1897 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
1898 first argument containing the interpreter's context. The trailing underscore
1899 in the pTHX_ macro indicates that the macro expansion needs a comma
1900 after the context argument because other arguments follow it. If
1901 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
1902 subroutine is not prototyped to take the extra argument. The form of the
1903 macro without the trailing underscore is used when there are no additional
1906 When a core function calls another, it must pass the context. This
1907 is normally hidden via macros. Consider C<sv_setiv>. It expands into
1908 something like this:
1910 #ifdef PERL_IMPLICIT_CONTEXT
1911 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
1912 /* can't do this for vararg functions, see below */
1914 #define sv_setiv Perl_sv_setiv
1917 This works well, and means that XS authors can gleefully write:
1921 and still have it work under all the modes Perl could have been
1924 This doesn't work so cleanly for varargs functions, though, as macros
1925 imply that the number of arguments is known in advance. Instead we
1926 either need to spell them out fully, passing C<aTHX_> as the first
1927 argument (the Perl core tends to do this with functions like
1928 Perl_warner), or use a context-free version.
1930 The context-free version of Perl_warner is called
1931 Perl_warner_nocontext, and does not take the extra argument. Instead
1932 it does dTHX; to get the context from thread-local storage. We
1933 C<#define warner Perl_warner_nocontext> so that extensions get source
1934 compatibility at the expense of performance. (Passing an arg is
1935 cheaper than grabbing it from thread-local storage.)
1937 You can ignore [pad]THXx when browsing the Perl headers/sources.
1938 Those are strictly for use within the core. Extensions and embedders
1939 need only be aware of [pad]THX.
1941 =head2 So what happened to dTHR?
1943 C<dTHR> was introduced in perl 5.005 to support the older thread model.
1944 The older thread model now uses the C<THX> mechanism to pass context
1945 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
1946 later still have it for backward source compatibility, but it is defined
1949 =head2 How do I use all this in extensions?
1951 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1952 any functions in the Perl API will need to pass the initial context
1953 argument somehow. The kicker is that you will need to write it in
1954 such a way that the extension still compiles when Perl hasn't been
1955 built with PERL_IMPLICIT_CONTEXT enabled.
1957 There are three ways to do this. First, the easy but inefficient way,
1958 which is also the default, in order to maintain source compatibility
1959 with extensions: whenever XSUB.h is #included, it redefines the aTHX
1960 and aTHX_ macros to call a function that will return the context.
1961 Thus, something like:
1965 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
1968 Perl_sv_setiv(Perl_get_context(), sv, num);
1970 or to this otherwise:
1972 Perl_sv_setiv(sv, num);
1974 You have to do nothing new in your extension to get this; since
1975 the Perl library provides Perl_get_context(), it will all just
1978 The second, more efficient way is to use the following template for
1981 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1986 static my_private_function(int arg1, int arg2);
1989 my_private_function(int arg1, int arg2)
1991 dTHX; /* fetch context */
1992 ... call many Perl API functions ...
1997 MODULE = Foo PACKAGE = Foo
2005 my_private_function(arg, 10);
2007 Note that the only two changes from the normal way of writing an
2008 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2009 including the Perl headers, followed by a C<dTHX;> declaration at
2010 the start of every function that will call the Perl API. (You'll
2011 know which functions need this, because the C compiler will complain
2012 that there's an undeclared identifier in those functions.) No changes
2013 are needed for the XSUBs themselves, because the XS() macro is
2014 correctly defined to pass in the implicit context if needed.
2016 The third, even more efficient way is to ape how it is done within
2020 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2025 /* pTHX_ only needed for functions that call Perl API */
2026 static my_private_function(pTHX_ int arg1, int arg2);
2029 my_private_function(pTHX_ int arg1, int arg2)
2031 /* dTHX; not needed here, because THX is an argument */
2032 ... call Perl API functions ...
2037 MODULE = Foo PACKAGE = Foo
2045 my_private_function(aTHX_ arg, 10);
2047 This implementation never has to fetch the context using a function
2048 call, since it is always passed as an extra argument. Depending on
2049 your needs for simplicity or efficiency, you may mix the previous
2050 two approaches freely.
2052 Never add a comma after C<pTHX> yourself--always use the form of the
2053 macro with the underscore for functions that take explicit arguments,
2054 or the form without the argument for functions with no explicit arguments.
2056 =head2 Should I do anything special if I call perl from multiple threads?
2058 If you create interpreters in one thread and then proceed to call them in
2059 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2060 initialized correctly in each of those threads.
2062 The C<perl_alloc> and C<perl_clone> API functions will automatically set
2063 the TLS slot to the interpreter they created, so that there is no need to do
2064 anything special if the interpreter is always accessed in the same thread that
2065 created it, and that thread did not create or call any other interpreters
2066 afterwards. If that is not the case, you have to set the TLS slot of the
2067 thread before calling any functions in the Perl API on that particular
2068 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2069 thread as the first thing you do:
2071 /* do this before doing anything else with some_perl */
2072 PERL_SET_CONTEXT(some_perl);
2074 ... other Perl API calls on some_perl go here ...
2076 =head2 Future Plans and PERL_IMPLICIT_SYS
2078 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2079 that the interpreter knows about itself and pass it around, so too are
2080 there plans to allow the interpreter to bundle up everything it knows
2081 about the environment it's running on. This is enabled with the
2082 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS
2083 and USE_5005THREADS on Windows (see inside iperlsys.h).
2085 This allows the ability to provide an extra pointer (called the "host"
2086 environment) for all the system calls. This makes it possible for
2087 all the system stuff to maintain their own state, broken down into
2088 seven C structures. These are thin wrappers around the usual system
2089 calls (see win32/perllib.c) for the default perl executable, but for a
2090 more ambitious host (like the one that would do fork() emulation) all
2091 the extra work needed to pretend that different interpreters are
2092 actually different "processes", would be done here.
2094 The Perl engine/interpreter and the host are orthogonal entities.
2095 There could be one or more interpreters in a process, and one or
2096 more "hosts", with free association between them.
2098 =head1 Internal Functions
2100 All of Perl's internal functions which will be exposed to the outside
2101 world are prefixed by C<Perl_> so that they will not conflict with XS
2102 functions or functions used in a program in which Perl is embedded.
2103 Similarly, all global variables begin with C<PL_>. (By convention,
2104 static functions start with C<S_>.)
2106 Inside the Perl core, you can get at the functions either with or
2107 without the C<Perl_> prefix, thanks to a bunch of defines that live in
2108 F<embed.h>. This header file is generated automatically from
2109 F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2110 header files for the internal functions, generates the documentation
2111 and a lot of other bits and pieces. It's important that when you add
2112 a new function to the core or change an existing one, you change the
2113 data in the table in F<embed.fnc> as well. Here's a sample entry from
2116 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2118 The second column is the return type, the third column the name. Columns
2119 after that are the arguments. The first column is a set of flags:
2125 This function is a part of the public API.
2129 This function has a C<Perl_> prefix; ie, it is defined as C<Perl_av_fetch>
2133 This function has documentation using the C<apidoc> feature which we'll
2134 look at in a second.
2138 Other available flags are:
2144 This is a static function and is defined as C<S_whatever>, and usually
2145 called within the sources as C<whatever(...)>.
2149 This does not use C<aTHX_> and C<pTHX> to pass interpreter context. (See
2150 L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
2154 This function never returns; C<croak>, C<exit> and friends.
2158 This function takes a variable number of arguments, C<printf> style.
2159 The argument list should end with C<...>, like this:
2161 Afprd |void |croak |const char* pat|...
2165 This function is part of the experimental development API, and may change
2166 or disappear without notice.
2170 This function should not have a compatibility macro to define, say,
2171 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2175 This function isn't exported out of the Perl core.
2179 This is implemented as a macro.
2183 This function is explicitly exported.
2187 This function is visible to extensions included in the Perl core.
2191 Binary backward compatibility; this function is a macro but also has
2192 a C<Perl_> implementation (which is exported).
2196 If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2197 C<make regen_headers> to force a rebuild of F<embed.h> and other
2198 auto-generated files.
2200 =head2 Formatted Printing of IVs, UVs, and NVs
2202 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2203 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2204 following macros for portability
2209 UVxf UV in hexadecimal
2214 These will take care of 64-bit integers and long doubles.
2217 printf("IV is %"IVdf"\n", iv);
2219 The IVdf will expand to whatever is the correct format for the IVs.
2221 If you are printing addresses of pointers, use UVxf combined
2222 with PTR2UV(), do not use %lx or %p.
2224 =head2 Pointer-To-Integer and Integer-To-Pointer
2226 Because pointer size does not necessarily equal integer size,
2227 use the follow macros to do it right.
2232 INT2PTR(pointertotype, integer)
2237 SV *sv = INT2PTR(SV*, iv);
2244 =head2 Source Documentation
2246 There's an effort going on to document the internal functions and
2247 automatically produce reference manuals from them - L<perlapi> is one
2248 such manual which details all the functions which are available to XS
2249 writers. L<perlintern> is the autogenerated manual for the functions
2250 which are not part of the API and are supposedly for internal use only.
2252 Source documentation is created by putting POD comments into the C
2256 =for apidoc sv_setiv
2258 Copies an integer into the given SV. Does not handle 'set' magic. See
2264 Please try and supply some documentation if you add functions to the
2267 =head1 Unicode Support
2269 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2270 writers to understand this support and make sure that the code they
2271 write does not corrupt Unicode data.
2273 =head2 What B<is> Unicode, anyway?
2275 In the olden, less enlightened times, we all used to use ASCII. Most of
2276 us did, anyway. The big problem with ASCII is that it's American. Well,
2277 no, that's not actually the problem; the problem is that it's not
2278 particularly useful for people who don't use the Roman alphabet. What
2279 used to happen was that particular languages would stick their own
2280 alphabet in the upper range of the sequence, between 128 and 255. Of
2281 course, we then ended up with plenty of variants that weren't quite
2282 ASCII, and the whole point of it being a standard was lost.
2284 Worse still, if you've got a language like Chinese or
2285 Japanese that has hundreds or thousands of characters, then you really
2286 can't fit them into a mere 256, so they had to forget about ASCII
2287 altogether, and build their own systems using pairs of numbers to refer
2290 To fix this, some people formed Unicode, Inc. and
2291 produced a new character set containing all the characters you can
2292 possibly think of and more. There are several ways of representing these
2293 characters, and the one Perl uses is called UTF-8. UTF-8 uses
2294 a variable number of bytes to represent a character, instead of just
2295 one. You can learn more about Unicode at http://www.unicode.org/
2297 =head2 How can I recognise a UTF-8 string?
2299 You can't. This is because UTF-8 data is stored in bytes just like
2300 non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
2301 capital E with a grave accent, is represented by the two bytes
2302 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2303 has that byte sequence as well. So you can't tell just by looking - this
2304 is what makes Unicode input an interesting problem.
2306 The API function C<is_utf8_string> can help; it'll tell you if a string
2307 contains only valid UTF-8 characters. However, it can't do the work for
2308 you. On a character-by-character basis, C<is_utf8_char> will tell you
2309 whether the current character in a string is valid UTF-8.
2311 =head2 How does UTF-8 represent Unicode characters?
2313 As mentioned above, UTF-8 uses a variable number of bytes to store a
2314 character. Characters with values 1...128 are stored in one byte, just
2315 like good ol' ASCII. Character 129 is stored as C<v194.129>; this
2316 continues up to character 191, which is C<v194.191>. Now we've run out of
2317 bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2318 so it goes on, moving to three bytes at character 2048.
2320 Assuming you know you're dealing with a UTF-8 string, you can find out
2321 how long the first character in it is with the C<UTF8SKIP> macro:
2323 char *utf = "\305\233\340\240\201";
2326 len = UTF8SKIP(utf); /* len is 2 here */
2328 len = UTF8SKIP(utf); /* len is 3 here */
2330 Another way to skip over characters in a UTF-8 string is to use
2331 C<utf8_hop>, which takes a string and a number of characters to skip
2332 over. You're on your own about bounds checking, though, so don't use it
2335 All bytes in a multi-byte UTF-8 character will have the high bit set,
2336 so you can test if you need to do something special with this
2337 character like this (the UTF8_IS_INVARIANT() is a macro that tests
2338 whether the byte can be encoded as a single byte even in UTF-8):
2341 UV uv; /* Note: a UV, not a U8, not a char */
2343 if (!UTF8_IS_INVARIANT(*utf))
2344 /* Must treat this as UTF-8 */
2345 uv = utf8_to_uv(utf);
2347 /* OK to treat this character as a byte */
2350 You can also see in that example that we use C<utf8_to_uv> to get the
2351 value of the character; the inverse function C<uv_to_utf8> is available
2352 for putting a UV into UTF-8:
2354 if (!UTF8_IS_INVARIANT(uv))
2355 /* Must treat this as UTF8 */
2356 utf8 = uv_to_utf8(utf8, uv);
2358 /* OK to treat this character as a byte */
2361 You B<must> convert characters to UVs using the above functions if
2362 you're ever in a situation where you have to match UTF-8 and non-UTF-8
2363 characters. You may not skip over UTF-8 characters in this case. If you
2364 do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2365 for instance, if your UTF-8 string contains C<v196.172>, and you skip
2366 that character, you can never match a C<chr(200)> in a non-UTF-8 string.
2369 =head2 How does Perl store UTF-8 strings?
2371 Currently, Perl deals with Unicode strings and non-Unicode strings
2372 slightly differently. If a string has been identified as being UTF-8
2373 encoded, Perl will set a flag in the SV, C<SVf_UTF8>. You can check and
2374 manipulate this flag with the following macros:
2380 This flag has an important effect on Perl's treatment of the string: if
2381 Unicode data is not properly distinguished, regular expressions,
2382 C<length>, C<substr> and other string handling operations will have
2383 undesirable results.
2385 The problem comes when you have, for instance, a string that isn't
2386 flagged is UTF-8, and contains a byte sequence that could be UTF-8 -
2387 especially when combining non-UTF-8 and UTF-8 strings.
2389 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2390 need be sure you don't accidentally knock it off while you're
2391 manipulating SVs. More specifically, you cannot expect to do this:
2400 nsv = newSVpvn(p, len);
2402 The C<char*> string does not tell you the whole story, and you can't
2403 copy or reconstruct an SV just by copying the string value. Check if the
2404 old SV has the UTF-8 flag set, and act accordingly:
2408 nsv = newSVpvn(p, len);
2412 In fact, your C<frobnicate> function should be made aware of whether or
2413 not it's dealing with UTF-8 data, so that it can handle the string
2416 Since just passing an SV to an XS function and copying the data of
2417 the SV is not enough to copy the UTF-8 flags, even less right is just
2418 passing a C<char *> to an XS function.
2420 =head2 How do I convert a string to UTF-8?
2422 If you're mixing UTF-8 and non-UTF-8 strings, you might find it necessary
2423 to upgrade one of the strings to UTF-8. If you've got an SV, the easiest
2426 sv_utf8_upgrade(sv);
2428 However, you must not do this, for example:
2431 sv_utf8_upgrade(left);
2433 If you do this in a binary operator, you will actually change one of the
2434 strings that came into the operator, and, while it shouldn't be noticeable
2435 by the end user, it can cause problems.
2437 Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
2438 string argument. This is useful for having the data available for
2439 comparisons and so on, without harming the original SV. There's also
2440 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2441 the string contains any characters above 255 that can't be represented
2444 =head2 Is there anything else I need to know?
2446 Not really. Just remember these things:
2452 There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2453 is UTF-8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2454 something should be UTF-8. Treat the flag as part of the PV, even though
2455 it's not - if you pass on the PV to somewhere, pass on the flag too.
2459 If a string is UTF-8, B<always> use C<utf8_to_uv> to get at the value,
2460 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
2464 When writing a character C<uv> to a UTF-8 string, B<always> use
2465 C<uv_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2466 you can use C<*s = uv>.
2470 Mixing UTF-8 and non-UTF-8 strings is tricky. Use C<bytes_to_utf8> to get
2471 a new string which is UTF-8 encoded. There are tricks you can use to
2472 delay deciding whether you need to use a UTF-8 string until you get to a
2473 high character - C<HALF_UPGRADE> is one of those.
2477 =head1 Custom Operators
2479 Custom operator support is a new experimental feature that allows you to
2480 define your own ops. This is primarily to allow the building of
2481 interpreters for other languages in the Perl core, but it also allows
2482 optimizations through the creation of "macro-ops" (ops which perform the
2483 functions of multiple ops which are usually executed together, such as
2484 C<gvsv, gvsv, add>.)
2486 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
2487 core does not "know" anything special about this op type, and so it will
2488 not be involved in any optimizations. This also means that you can
2489 define your custom ops to be any op structure - unary, binary, list and
2492 It's important to know what custom operators won't do for you. They
2493 won't let you add new syntax to Perl, directly. They won't even let you
2494 add new keywords, directly. In fact, they won't change the way Perl
2495 compiles a program at all. You have to do those changes yourself, after
2496 Perl has compiled the program. You do this either by manipulating the op
2497 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2498 a custom peephole optimizer with the C<optimize> module.
2500 When you do this, you replace ordinary Perl ops with custom ops by
2501 creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2502 PP function. This should be defined in XS code, and should look like
2503 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2504 takes the appropriate number of values from the stack, and you are
2505 responsible for adding stack marks if necessary.
2507 You should also "register" your op with the Perl interpreter so that it
2508 can produce sensible error and warning messages. Since it is possible to
2509 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2510 Perl uses the value of C<< o->op_ppaddr >> as a key into the
2511 C<PL_custom_op_descs> and C<PL_custom_op_names> hashes. This means you
2512 need to enter a name and description for your op at the appropriate
2513 place in the C<PL_custom_op_names> and C<PL_custom_op_descs> hashes.
2515 Forthcoming versions of C<B::Generate> (version 1.0 and above) should
2516 directly support the creation of custom ops by name; C<Opcodes::Custom>
2517 will provide functions which make it trivial to "register" custom ops to
2518 the Perl interpreter.
2522 Until May 1997, this document was maintained by Jeff Okamoto
2523 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2524 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
2526 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2527 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2528 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2529 Stephen McCamant, and Gurusamy Sarathy.
2533 perlapi(1), perlintern(1), perlxs(1), perlembed(1)