3 perlguts - Perl's Internal Functions
7 This document attempts to describe some of the internal functions of the
8 Perl executable. It is far from complete and probably contains many errors.
9 Please refer any questions or comments to the author below.
15 Perl has three typedefs that handle Perl's three main data types:
21 Each typedef has specific routines that manipulate the various data types.
23 =head2 What is an "IV"?
25 Perl uses a special typedef IV which is a simple integer type that is
26 guaranteed to be large enough to hold a pointer (as well as an integer).
28 Perl also uses two special typedefs, I32 and I16, which will always be at
29 least 32-bits and 16-bits long, respectively.
31 =head2 Working with SVs
33 An SV can be created and loaded with one command. There are four types of
34 values that can be loaded: an integer value (IV), a double (NV), a string,
35 (PV), and another scalar (SV).
41 SV* newSVpv(const char*, int);
42 SV* newSVpvn(const char*, int);
43 SV* newSVpvf(const char*, ...);
46 To change the value of an *already-existing* SV, there are seven routines:
48 void sv_setiv(SV*, IV);
49 void sv_setuv(SV*, UV);
50 void sv_setnv(SV*, double);
51 void sv_setpv(SV*, const char*);
52 void sv_setpvn(SV*, const char*, int)
53 void sv_setpvf(SV*, const char*, ...);
54 void sv_setpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
55 void sv_setsv(SV*, SV*);
57 Notice that you can choose to specify the length of the string to be
58 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
59 allow Perl to calculate the length by using C<sv_setpv> or by specifying
60 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
61 determine the string's length by using C<strlen>, which depends on the
62 string terminating with a NUL character.
64 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
65 formatted output becomes the value.
67 C<sv_setpvfn> is an analogue of C<vsprintf>, but it allows you to specify
68 either a pointer to a variable argument list or the address and length of
69 an array of SVs. The last argument points to a boolean; on return, if that
70 boolean is true, then locale-specific information has been used to format
71 the string, and the string's contents are therefore untrustworthy (see
72 L<perlsec>). This pointer may be NULL if that information is not
73 important. Note that this function requires you to specify the length of
76 The C<sv_set*()> functions are not generic enough to operate on values
77 that have "magic". See L<Magic Virtual Tables> later in this document.
79 All SVs that contain strings should be terminated with a NUL character.
80 If it is not NUL-terminated there is a risk of
81 core dumps and corruptions from code which passes the string to C
82 functions or system calls which expect a NUL-terminated string.
83 Perl's own functions typically add a trailing NUL for this reason.
84 Nevertheless, you should be very careful when you pass a string stored
85 in an SV to a C function or system call.
87 To access the actual value that an SV points to, you can use the macros:
94 which will automatically coerce the actual scalar type into an IV, double,
97 In the C<SvPV> macro, the length of the string returned is placed into the
98 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
99 not care what the length of the data is, use the C<SvPV_nolen> macro.
100 Historically the C<SvPV> macro with the global variable C<PL_na> has been
101 used in this case. But that can be quite inefficient because C<PL_na> must
102 be accessed in thread-local storage in threaded Perl. In any case, remember
103 that Perl allows arbitrary strings of data that may both contain NULs and
104 might not be terminated by a NUL.
106 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
107 len);>. It might work with your compiler, but it won't work for everyone.
108 Break this sort of statement up into separate assignments:
115 If you want to know if the scalar value is TRUE, you can use:
119 Although Perl will automatically grow strings for you, if you need to force
120 Perl to allocate more memory for your SV, you can use the macro
122 SvGROW(SV*, STRLEN newlen)
124 which will determine if more memory needs to be allocated. If so, it will
125 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
126 decrease, the allocated memory of an SV and that it does not automatically
127 add a byte for the a trailing NUL (perl's own string functions typically do
128 C<SvGROW(sv, len + 1)>).
130 If you have an SV and want to know what kind of data Perl thinks is stored
131 in it, you can use the following macros to check the type of SV you have.
137 You can get and set the current length of the string stored in an SV with
138 the following macros:
141 SvCUR_set(SV*, I32 val)
143 You can also get a pointer to the end of the string stored in the SV
148 But note that these last three macros are valid only if C<SvPOK()> is true.
150 If you want to append something to the end of string stored in an C<SV*>,
151 you can use the following functions:
153 void sv_catpv(SV*, const char*);
154 void sv_catpvn(SV*, const char*, STRLEN);
155 void sv_catpvf(SV*, const char*, ...);
156 void sv_catpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
157 void sv_catsv(SV*, SV*);
159 The first function calculates the length of the string to be appended by
160 using C<strlen>. In the second, you specify the length of the string
161 yourself. The third function processes its arguments like C<sprintf> and
162 appends the formatted output. The fourth function works like C<vsprintf>.
163 You can specify the address and length of an array of SVs instead of the
164 va_list argument. The fifth function extends the string stored in the first
165 SV with the string stored in the second SV. It also forces the second SV
166 to be interpreted as a string.
168 The C<sv_cat*()> functions are not generic enough to operate on values that
169 have "magic". See L<Magic Virtual Tables> later in this document.
171 If you know the name of a scalar variable, you can get a pointer to its SV
172 by using the following:
174 SV* perl_get_sv("package::varname", FALSE);
176 This returns NULL if the variable does not exist.
178 If you want to know if this variable (or any other SV) is actually C<defined>,
183 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>. Its
184 address can be used whenever an C<SV*> is needed.
186 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain Boolean
187 TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their addresses can
188 be used whenever an C<SV*> is needed.
190 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
194 if (I-am-to-return-a-real-value) {
195 sv = sv_2mortal(newSViv(42));
199 This code tries to return a new SV (which contains the value 42) if it should
200 return a real value, or undef otherwise. Instead it has returned a NULL
201 pointer which, somewhere down the line, will cause a segmentation violation,
202 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the first
203 line and all will be well.
205 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
206 call is not necessary (see L<Reference Counts and Mortality>).
208 =head2 What's Really Stored in an SV?
210 Recall that the usual method of determining the type of scalar you have is
211 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
212 usually these macros will always return TRUE and calling the C<Sv*V>
213 macros will do the appropriate conversion of string to integer/double or
214 integer/double to string.
216 If you I<really> need to know if you have an integer, double, or string
217 pointer in an SV, you can use the following three macros instead:
223 These will tell you if you truly have an integer, double, or string pointer
224 stored in your SV. The "p" stands for private.
226 In general, though, it's best to use the C<Sv*V> macros.
228 =head2 Working with AVs
230 There are two ways to create and load an AV. The first method creates an
235 The second method both creates the AV and initially populates it with SVs:
237 AV* av_make(I32 num, SV **ptr);
239 The second argument points to an array containing C<num> C<SV*>'s. Once the
240 AV has been created, the SVs can be destroyed, if so desired.
242 Once the AV has been created, the following operations are possible on AVs:
244 void av_push(AV*, SV*);
247 void av_unshift(AV*, I32 num);
249 These should be familiar operations, with the exception of C<av_unshift>.
250 This routine adds C<num> elements at the front of the array with the C<undef>
251 value. You must then use C<av_store> (described below) to assign values
252 to these new elements.
254 Here are some other functions:
257 SV** av_fetch(AV*, I32 key, I32 lval);
258 SV** av_store(AV*, I32 key, SV* val);
260 The C<av_len> function returns the highest index value in array (just
261 like $#array in Perl). If the array is empty, -1 is returned. The
262 C<av_fetch> function returns the value at index C<key>, but if C<lval>
263 is non-zero, then C<av_fetch> will store an undef value at that index.
264 The C<av_store> function stores the value C<val> at index C<key>, and does
265 not increment the reference count of C<val>. Thus the caller is responsible
266 for taking care of that, and if C<av_store> returns NULL, the caller will
267 have to decrement the reference count to avoid a memory leak. Note that
268 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
273 void av_extend(AV*, I32 key);
275 The C<av_clear> function deletes all the elements in the AV* array, but
276 does not actually delete the array itself. The C<av_undef> function will
277 delete all the elements in the array plus the array itself. The
278 C<av_extend> function extends the array so that it contains at least C<key+1>
279 elements. If C<key+1> is less than the currently allocated length of the array,
280 then nothing is done.
282 If you know the name of an array variable, you can get a pointer to its AV
283 by using the following:
285 AV* perl_get_av("package::varname", FALSE);
287 This returns NULL if the variable does not exist.
289 See L<Understanding the Magic of Tied Hashes and Arrays> for more
290 information on how to use the array access functions on tied arrays.
292 =head2 Working with HVs
294 To create an HV, you use the following routine:
298 Once the HV has been created, the following operations are possible on HVs:
300 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
301 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
303 The C<klen> parameter is the length of the key being passed in (Note that
304 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
305 length of the key). The C<val> argument contains the SV pointer to the
306 scalar being stored, and C<hash> is the precomputed hash value (zero if
307 you want C<hv_store> to calculate it for you). The C<lval> parameter
308 indicates whether this fetch is actually a part of a store operation, in
309 which case a new undefined value will be added to the HV with the supplied
310 key and C<hv_fetch> will return as if the value had already existed.
312 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
313 C<SV*>. To access the scalar value, you must first dereference the return
314 value. However, you should check to make sure that the return value is
315 not NULL before dereferencing it.
317 These two functions check if a hash table entry exists, and deletes it.
319 bool hv_exists(HV*, const char* key, U32 klen);
320 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
322 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
323 create and return a mortal copy of the deleted value.
325 And more miscellaneous functions:
330 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
331 table but does not actually delete the hash table. The C<hv_undef> deletes
332 both the entries and the hash table itself.
334 Perl keeps the actual data in linked list of structures with a typedef of HE.
335 These contain the actual key and value pointers (plus extra administrative
336 overhead). The key is a string pointer; the value is an C<SV*>. However,
337 once you have an C<HE*>, to get the actual key and value, use the routines
340 I32 hv_iterinit(HV*);
341 /* Prepares starting point to traverse hash table */
342 HE* hv_iternext(HV*);
343 /* Get the next entry, and return a pointer to a
344 structure that has both the key and value */
345 char* hv_iterkey(HE* entry, I32* retlen);
346 /* Get the key from an HE structure and also return
347 the length of the key string */
348 SV* hv_iterval(HV*, HE* entry);
349 /* Return a SV pointer to the value of the HE
351 SV* hv_iternextsv(HV*, char** key, I32* retlen);
352 /* This convenience routine combines hv_iternext,
353 hv_iterkey, and hv_iterval. The key and retlen
354 arguments are return values for the key and its
355 length. The value is returned in the SV* argument */
357 If you know the name of a hash variable, you can get a pointer to its HV
358 by using the following:
360 HV* perl_get_hv("package::varname", FALSE);
362 This returns NULL if the variable does not exist.
364 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
368 hash = (hash * 33) + *key++;
369 hash = hash + (hash >> 5); /* after 5.6 */
371 The last step was added in version 5.6 to improve distribution of
372 lower bits in the resulting hash value.
374 See L<Understanding the Magic of Tied Hashes and Arrays> for more
375 information on how to use the hash access functions on tied hashes.
377 =head2 Hash API Extensions
379 Beginning with version 5.004, the following functions are also supported:
381 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
382 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
384 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
385 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
387 SV* hv_iterkeysv (HE* entry);
389 Note that these functions take C<SV*> keys, which simplifies writing
390 of extension code that deals with hash structures. These functions
391 also allow passing of C<SV*> keys to C<tie> functions without forcing
392 you to stringify the keys (unlike the previous set of functions).
394 They also return and accept whole hash entries (C<HE*>), making their
395 use more efficient (since the hash number for a particular string
396 doesn't have to be recomputed every time). See L<API LISTING> later in
397 this document for detailed descriptions.
399 The following macros must always be used to access the contents of hash
400 entries. Note that the arguments to these macros must be simple
401 variables, since they may get evaluated more than once. See
402 L<API LISTING> later in this document for detailed descriptions of these
405 HePV(HE* he, STRLEN len)
409 HeSVKEY_force(HE* he)
410 HeSVKEY_set(HE* he, SV* sv)
412 These two lower level macros are defined, but must only be used when
413 dealing with keys that are not C<SV*>s:
418 Note that both C<hv_store> and C<hv_store_ent> do not increment the
419 reference count of the stored C<val>, which is the caller's responsibility.
420 If these functions return a NULL value, the caller will usually have to
421 decrement the reference count of C<val> to avoid a memory leak.
425 References are a special type of scalar that point to other data types
426 (including references).
428 To create a reference, use either of the following functions:
430 SV* newRV_inc((SV*) thing);
431 SV* newRV_noinc((SV*) thing);
433 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
434 functions are identical except that C<newRV_inc> increments the reference
435 count of the C<thing>, while C<newRV_noinc> does not. For historical
436 reasons, C<newRV> is a synonym for C<newRV_inc>.
438 Once you have a reference, you can use the following macro to dereference
443 then call the appropriate routines, casting the returned C<SV*> to either an
444 C<AV*> or C<HV*>, if required.
446 To determine if an SV is a reference, you can use the following macro:
450 To discover what type of value the reference refers to, use the following
451 macro and then check the return value.
455 The most useful types that will be returned are:
464 SVt_PVGV Glob (possible a file handle)
465 SVt_PVMG Blessed or Magical Scalar
467 See the sv.h header file for more details.
469 =head2 Blessed References and Class Objects
471 References are also used to support object-oriented programming. In the
472 OO lexicon, an object is simply a reference that has been blessed into a
473 package (or class). Once blessed, the programmer may now use the reference
474 to access the various methods in the class.
476 A reference can be blessed into a package with the following function:
478 SV* sv_bless(SV* sv, HV* stash);
480 The C<sv> argument must be a reference. The C<stash> argument specifies
481 which class the reference will belong to. See
482 L<Stashes and Globs> for information on converting class names into stashes.
484 /* Still under construction */
486 Upgrades rv to reference if not already one. Creates new SV for rv to
487 point to. If C<classname> is non-null, the SV is blessed into the specified
488 class. SV is returned.
490 SV* newSVrv(SV* rv, const char* classname);
492 Copies integer or double into an SV whose reference is C<rv>. SV is blessed
493 if C<classname> is non-null.
495 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
496 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
498 Copies the pointer value (I<the address, not the string!>) into an SV whose
499 reference is rv. SV is blessed if C<classname> is non-null.
501 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
503 Copies string into an SV whose reference is C<rv>. Set length to 0 to let
504 Perl calculate the string length. SV is blessed if C<classname> is non-null.
506 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
508 Tests whether the SV is blessed into the specified class. It does not
509 check inheritance relationships.
511 int sv_isa(SV* sv, const char* name);
513 Tests whether the SV is a reference to a blessed object.
515 int sv_isobject(SV* sv);
517 Tests whether the SV is derived from the specified class. SV can be either
518 a reference to a blessed object or a string containing a class name. This
519 is the function implementing the C<UNIVERSAL::isa> functionality.
521 bool sv_derived_from(SV* sv, const char* name);
523 To check if you've got an object derived from a specific class you have
526 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
528 =head2 Creating New Variables
530 To create a new Perl variable with an undef value which can be accessed from
531 your Perl script, use the following routines, depending on the variable type.
533 SV* perl_get_sv("package::varname", TRUE);
534 AV* perl_get_av("package::varname", TRUE);
535 HV* perl_get_hv("package::varname", TRUE);
537 Notice the use of TRUE as the second parameter. The new variable can now
538 be set, using the routines appropriate to the data type.
540 There are additional macros whose values may be bitwise OR'ed with the
541 C<TRUE> argument to enable certain extra features. Those bits are:
543 GV_ADDMULTI Marks the variable as multiply defined, thus preventing the
544 "Name <varname> used only once: possible typo" warning.
545 GV_ADDWARN Issues the warning "Had to create <varname> unexpectedly" if
546 the variable did not exist before the function was called.
548 If you do not specify a package name, the variable is created in the current
551 =head2 Reference Counts and Mortality
553 Perl uses an reference count-driven garbage collection mechanism. SVs,
554 AVs, or HVs (xV for short in the following) start their life with a
555 reference count of 1. If the reference count of an xV ever drops to 0,
556 then it will be destroyed and its memory made available for reuse.
558 This normally doesn't happen at the Perl level unless a variable is
559 undef'ed or the last variable holding a reference to it is changed or
560 overwritten. At the internal level, however, reference counts can be
561 manipulated with the following macros:
563 int SvREFCNT(SV* sv);
564 SV* SvREFCNT_inc(SV* sv);
565 void SvREFCNT_dec(SV* sv);
567 However, there is one other function which manipulates the reference
568 count of its argument. The C<newRV_inc> function, you will recall,
569 creates a reference to the specified argument. As a side effect,
570 it increments the argument's reference count. If this is not what
571 you want, use C<newRV_noinc> instead.
573 For example, imagine you want to return a reference from an XSUB function.
574 Inside the XSUB routine, you create an SV which initially has a reference
575 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
576 This returns the reference as a new SV, but the reference count of the
577 SV you passed to C<newRV_inc> has been incremented to two. Now you
578 return the reference from the XSUB routine and forget about the SV.
579 But Perl hasn't! Whenever the returned reference is destroyed, the
580 reference count of the original SV is decreased to one and nothing happens.
581 The SV will hang around without any way to access it until Perl itself
582 terminates. This is a memory leak.
584 The correct procedure, then, is to use C<newRV_noinc> instead of
585 C<newRV_inc>. Then, if and when the last reference is destroyed,
586 the reference count of the SV will go to zero and it will be destroyed,
587 stopping any memory leak.
589 There are some convenience functions available that can help with the
590 destruction of xVs. These functions introduce the concept of "mortality".
591 An xV that is mortal has had its reference count marked to be decremented,
592 but not actually decremented, until "a short time later". Generally the
593 term "short time later" means a single Perl statement, such as a call to
594 an XSUB function. The actual determinant for when mortal xVs have their
595 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
596 See L<perlcall> and L<perlxs> for more details on these macros.
598 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
599 However, if you mortalize a variable twice, the reference count will
600 later be decremented twice.
602 You should be careful about creating mortal variables. Strange things
603 can happen if you make the same value mortal within multiple contexts,
604 or if you make a variable mortal multiple times.
606 To create a mortal variable, use the functions:
610 SV* sv_mortalcopy(SV*)
612 The first call creates a mortal SV, the second converts an existing
613 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
614 third creates a mortal copy of an existing SV.
616 The mortal routines are not just for SVs -- AVs and HVs can be
617 made mortal by passing their address (type-casted to C<SV*>) to the
618 C<sv_2mortal> or C<sv_mortalcopy> routines.
620 =head2 Stashes and Globs
622 A "stash" is a hash that contains all of the different objects that
623 are contained within a package. Each key of the stash is a symbol
624 name (shared by all the different types of objects that have the same
625 name), and each value in the hash table is a GV (Glob Value). This GV
626 in turn contains references to the various objects of that name,
627 including (but not limited to) the following:
636 There is a single stash called "PL_defstash" that holds the items that exist
637 in the "main" package. To get at the items in other packages, append the
638 string "::" to the package name. The items in the "Foo" package are in
639 the stash "Foo::" in PL_defstash. The items in the "Bar::Baz" package are
640 in the stash "Baz::" in "Bar::"'s stash.
642 To get the stash pointer for a particular package, use the function:
644 HV* gv_stashpv(const char* name, I32 create)
645 HV* gv_stashsv(SV*, I32 create)
647 The first function takes a literal string, the second uses the string stored
648 in the SV. Remember that a stash is just a hash table, so you get back an
649 C<HV*>. The C<create> flag will create a new package if it is set.
651 The name that C<gv_stash*v> wants is the name of the package whose symbol table
652 you want. The default package is called C<main>. If you have multiply nested
653 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
656 Alternately, if you have an SV that is a blessed reference, you can find
657 out the stash pointer by using:
659 HV* SvSTASH(SvRV(SV*));
661 then use the following to get the package name itself:
663 char* HvNAME(HV* stash);
665 If you need to bless or re-bless an object you can use the following
668 SV* sv_bless(SV*, HV* stash)
670 where the first argument, an C<SV*>, must be a reference, and the second
671 argument is a stash. The returned C<SV*> can now be used in the same way
674 For more information on references and blessings, consult L<perlref>.
676 =head2 Double-Typed SVs
678 Scalar variables normally contain only one type of value, an integer,
679 double, pointer, or reference. Perl will automatically convert the
680 actual scalar data from the stored type into the requested type.
682 Some scalar variables contain more than one type of scalar data. For
683 example, the variable C<$!> contains either the numeric value of C<errno>
684 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
686 To force multiple data values into an SV, you must do two things: use the
687 C<sv_set*v> routines to add the additional scalar type, then set a flag
688 so that Perl will believe it contains more than one type of data. The
689 four macros to set the flags are:
696 The particular macro you must use depends on which C<sv_set*v> routine
697 you called first. This is because every C<sv_set*v> routine turns on
698 only the bit for the particular type of data being set, and turns off
701 For example, to create a new Perl variable called "dberror" that contains
702 both the numeric and descriptive string error values, you could use the
706 extern char *dberror_list;
708 SV* sv = perl_get_sv("dberror", TRUE);
709 sv_setiv(sv, (IV) dberror);
710 sv_setpv(sv, dberror_list[dberror]);
713 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
714 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
716 =head2 Magic Variables
718 [This section still under construction. Ignore everything here. Post no
719 bills. Everything not permitted is forbidden.]
721 Any SV may be magical, that is, it has special features that a normal
722 SV does not have. These features are stored in the SV structure in a
723 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
736 Note this is current as of patchlevel 0, and could change at any time.
738 =head2 Assigning Magic
740 Perl adds magic to an SV using the sv_magic function:
742 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
744 The C<sv> argument is a pointer to the SV that is to acquire a new magical
747 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
748 set the C<SVt_PVMG> flag for the C<sv>. Perl then continues by adding
749 it to the beginning of the linked list of magical features. Any prior
750 entry of the same type of magic is deleted. Note that this can be
751 overridden, and multiple instances of the same type of magic can be
752 associated with an SV.
754 The C<name> and C<namlen> arguments are used to associate a string with
755 the magic, typically the name of a variable. C<namlen> is stored in the
756 C<mg_len> field and if C<name> is non-null and C<namlen> >= 0 a malloc'd
757 copy of the name is stored in C<mg_ptr> field.
759 The sv_magic function uses C<how> to determine which, if any, predefined
760 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
761 See the "Magic Virtual Table" section below. The C<how> argument is also
762 stored in the C<mg_type> field.
764 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
765 structure. If it is not the same as the C<sv> argument, the reference
766 count of the C<obj> object is incremented. If it is the same, or if
767 the C<how> argument is "#", or if it is a NULL pointer, then C<obj> is
768 merely stored, without the reference count being incremented.
770 There is also a function to add magic to an C<HV>:
772 void hv_magic(HV *hv, GV *gv, int how);
774 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
776 To remove the magic from an SV, call the function sv_unmagic:
778 void sv_unmagic(SV *sv, int type);
780 The C<type> argument should be equal to the C<how> value when the C<SV>
781 was initially made magical.
783 =head2 Magic Virtual Tables
785 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to a
786 C<MGVTBL>, which is a structure of function pointers and stands for
787 "Magic Virtual Table" to handle the various operations that might be
788 applied to that variable.
790 The C<MGVTBL> has five pointers to the following routine types:
792 int (*svt_get)(SV* sv, MAGIC* mg);
793 int (*svt_set)(SV* sv, MAGIC* mg);
794 U32 (*svt_len)(SV* sv, MAGIC* mg);
795 int (*svt_clear)(SV* sv, MAGIC* mg);
796 int (*svt_free)(SV* sv, MAGIC* mg);
798 This MGVTBL structure is set at compile-time in C<perl.h> and there are
799 currently 19 types (or 21 with overloading turned on). These different
800 structures contain pointers to various routines that perform additional
801 actions depending on which function is being called.
803 Function pointer Action taken
804 ---------------- ------------
805 svt_get Do something after the value of the SV is retrieved.
806 svt_set Do something after the SV is assigned a value.
807 svt_len Report on the SV's length.
808 svt_clear Clear something the SV represents.
809 svt_free Free any extra storage associated with the SV.
811 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
812 to an C<mg_type> of '\0') contains:
814 { magic_get, magic_set, magic_len, 0, 0 }
816 Thus, when an SV is determined to be magical and of type '\0', if a get
817 operation is being performed, the routine C<magic_get> is called. All
818 the various routines for the various magical types begin with C<magic_>.
820 The current kinds of Magic Virtual Tables are:
822 mg_type MGVTBL Type of magic
823 ------- ------ ----------------------------
824 \0 vtbl_sv Special scalar variable
825 A vtbl_amagic %OVERLOAD hash
826 a vtbl_amagicelem %OVERLOAD hash element
827 c (none) Holds overload table (AMT) on stash
828 B vtbl_bm Boyer-Moore (fast string search)
830 e vtbl_envelem %ENV hash element
831 f vtbl_fm Formline ('compiled' format)
832 g vtbl_mglob m//g target / study()ed string
833 I vtbl_isa @ISA array
834 i vtbl_isaelem @ISA array element
835 k vtbl_nkeys scalar(keys()) lvalue
836 L (none) Debugger %_<filename
837 l vtbl_dbline Debugger %_<filename element
838 o vtbl_collxfrm Locale transformation
839 P vtbl_pack Tied array or hash
840 p vtbl_packelem Tied array or hash element
841 q vtbl_packelem Tied scalar or handle
843 s vtbl_sigelem %SIG hash element
844 t vtbl_taint Taintedness
845 U vtbl_uvar Available for use by extensions
846 v vtbl_vec vec() lvalue
847 x vtbl_substr substr() lvalue
848 y vtbl_defelem Shadow "foreach" iterator variable /
849 smart parameter vivification
850 * vtbl_glob GV (typeglob)
851 # vtbl_arylen Array length ($#ary)
852 . vtbl_pos pos() lvalue
853 ~ (none) Available for use by extensions
855 When an uppercase and lowercase letter both exist in the table, then the
856 uppercase letter is used to represent some kind of composite type (a list
857 or a hash), and the lowercase letter is used to represent an element of
860 The '~' and 'U' magic types are defined specifically for use by
861 extensions and will not be used by perl itself. Extensions can use
862 '~' magic to 'attach' private information to variables (typically
863 objects). This is especially useful because there is no way for
864 normal perl code to corrupt this private information (unlike using
865 extra elements of a hash object).
867 Similarly, 'U' magic can be used much like tie() to call a C function
868 any time a scalar's value is used or changed. The C<MAGIC>'s
869 C<mg_ptr> field points to a C<ufuncs> structure:
872 I32 (*uf_val)(IV, SV*);
873 I32 (*uf_set)(IV, SV*);
877 When the SV is read from or written to, the C<uf_val> or C<uf_set>
878 function will be called with C<uf_index> as the first arg and a
879 pointer to the SV as the second. A simple example of how to add 'U'
880 magic is shown below. Note that the ufuncs structure is copied by
881 sv_magic, so you can safely allocate it on the stack.
889 uf.uf_val = &my_get_fn;
890 uf.uf_set = &my_set_fn;
892 sv_magic(sv, 0, 'U', (char*)&uf, sizeof(uf));
894 Note that because multiple extensions may be using '~' or 'U' magic,
895 it is important for extensions to take extra care to avoid conflict.
896 Typically only using the magic on objects blessed into the same class
897 as the extension is sufficient. For '~' magic, it may also be
898 appropriate to add an I32 'signature' at the top of the private data
901 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
902 earlier do B<not> invoke 'set' magic on their targets. This must
903 be done by the user either by calling the C<SvSETMAGIC()> macro after
904 calling these functions, or by using one of the C<sv_set*_mg()> or
905 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
906 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
907 obtained from external sources in functions that don't handle magic.
908 L<API LISTING> later in this document identifies such functions.
909 For example, calls to the C<sv_cat*()> functions typically need to be
910 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
911 since their implementation handles 'get' magic.
915 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
917 This routine returns a pointer to the C<MAGIC> structure stored in the SV.
918 If the SV does not have that magical feature, C<NULL> is returned. Also,
919 if the SV is not of type SVt_PVMG, Perl may core dump.
921 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
923 This routine checks to see what types of magic C<sv> has. If the mg_type
924 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
925 the mg_type field is changed to be the lowercase letter.
927 =head2 Understanding the Magic of Tied Hashes and Arrays
929 Tied hashes and arrays are magical beasts of the 'P' magic type.
931 WARNING: As of the 5.004 release, proper usage of the array and hash
932 access functions requires understanding a few caveats. Some
933 of these caveats are actually considered bugs in the API, to be fixed
934 in later releases, and are bracketed with [MAYCHANGE] below. If
935 you find yourself actually applying such information in this section, be
936 aware that the behavior may change in the future, umm, without warning.
938 The perl tie function associates a variable with an object that implements
939 the various GET, SET etc methods. To perform the equivalent of the perl
940 tie function from an XSUB, you must mimic this behaviour. The code below
941 carries out the necessary steps - firstly it creates a new hash, and then
942 creates a second hash which it blesses into the class which will implement
943 the tie methods. Lastly it ties the two hashes together, and returns a
944 reference to the new tied hash. Note that the code below does NOT call the
945 TIEHASH method in the MyTie class -
946 see L<Calling Perl Routines from within C Programs> for details on how
957 tie = newRV_noinc((SV*)newHV());
958 stash = gv_stashpv("MyTie", TRUE);
959 sv_bless(tie, stash);
960 hv_magic(hash, tie, 'P');
961 RETVAL = newRV_noinc(hash);
965 The C<av_store> function, when given a tied array argument, merely
966 copies the magic of the array onto the value to be "stored", using
967 C<mg_copy>. It may also return NULL, indicating that the value did not
968 actually need to be stored in the array. [MAYCHANGE] After a call to
969 C<av_store> on a tied array, the caller will usually need to call
970 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
971 TIEARRAY object. If C<av_store> did return NULL, a call to
972 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
975 The previous paragraph is applicable verbatim to tied hash access using the
976 C<hv_store> and C<hv_store_ent> functions as well.
978 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
979 C<hv_fetch_ent> actually return an undefined mortal value whose magic
980 has been initialized using C<mg_copy>. Note the value so returned does not
981 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
982 need to call C<mg_get()> on the returned value in order to actually invoke
983 the perl level "FETCH" method on the underlying TIE object. Similarly,
984 you may also call C<mg_set()> on the return value after possibly assigning
985 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
986 method on the TIE object. [/MAYCHANGE]
989 In other words, the array or hash fetch/store functions don't really
990 fetch and store actual values in the case of tied arrays and hashes. They
991 merely call C<mg_copy> to attach magic to the values that were meant to be
992 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
993 do the job of invoking the TIE methods on the underlying objects. Thus
994 the magic mechanism currently implements a kind of lazy access to arrays
997 Currently (as of perl version 5.004), use of the hash and array access
998 functions requires the user to be aware of whether they are operating on
999 "normal" hashes and arrays, or on their tied variants. The API may be
1000 changed to provide more transparent access to both tied and normal data
1001 types in future versions.
1004 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1005 are mere sugar to invoke some perl method calls while using the uniform hash
1006 and array syntax. The use of this sugar imposes some overhead (typically
1007 about two to four extra opcodes per FETCH/STORE operation, in addition to
1008 the creation of all the mortal variables required to invoke the methods).
1009 This overhead will be comparatively small if the TIE methods are themselves
1010 substantial, but if they are only a few statements long, the overhead
1011 will not be insignificant.
1013 =head2 Localizing changes
1015 Perl has a very handy construction
1022 This construction is I<approximately> equivalent to
1031 The biggest difference is that the first construction would
1032 reinstate the initial value of $var, irrespective of how control exits
1033 the block: C<goto>, C<return>, C<die>/C<eval> etc. It is a little bit
1034 more efficient as well.
1036 There is a way to achieve a similar task from C via Perl API: create a
1037 I<pseudo-block>, and arrange for some changes to be automatically
1038 undone at the end of it, either explicit, or via a non-local exit (via
1039 die()). A I<block>-like construct is created by a pair of
1040 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1041 Such a construct may be created specially for some important localized
1042 task, or an existing one (like boundaries of enclosing Perl
1043 subroutine/block, or an existing pair for freeing TMPs) may be
1044 used. (In the second case the overhead of additional localization must
1045 be almost negligible.) Note that any XSUB is automatically enclosed in
1046 an C<ENTER>/C<LEAVE> pair.
1048 Inside such a I<pseudo-block> the following service is available:
1052 =item C<SAVEINT(int i)>
1054 =item C<SAVEIV(IV i)>
1056 =item C<SAVEI32(I32 i)>
1058 =item C<SAVELONG(long i)>
1060 These macros arrange things to restore the value of integer variable
1061 C<i> at the end of enclosing I<pseudo-block>.
1063 =item C<SAVESPTR(s)>
1065 =item C<SAVEPPTR(p)>
1067 These macros arrange things to restore the value of pointers C<s> and
1068 C<p>. C<s> must be a pointer of a type which survives conversion to
1069 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1072 =item C<SAVEFREESV(SV *sv)>
1074 The refcount of C<sv> would be decremented at the end of
1075 I<pseudo-block>. This is similar to C<sv_2mortal>, which should (?) be
1078 =item C<SAVEFREEOP(OP *op)>
1080 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1082 =item C<SAVEFREEPV(p)>
1084 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1085 end of I<pseudo-block>.
1087 =item C<SAVECLEARSV(SV *sv)>
1089 Clears a slot in the current scratchpad which corresponds to C<sv> at
1090 the end of I<pseudo-block>.
1092 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1094 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1095 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1096 short-lived storage, the corresponding string may be reallocated like
1099 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1101 =item C<SAVEDESTRUCTOR(f,p)>
1103 At the end of I<pseudo-block> the function C<f> is called with the
1104 only argument (of type C<void*>) C<p>.
1106 =item C<SAVESTACK_POS()>
1108 The current offset on the Perl internal stack (cf. C<SP>) is restored
1109 at the end of I<pseudo-block>.
1113 The following API list contains functions, thus one needs to
1114 provide pointers to the modifiable data explicitly (either C pointers,
1115 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1116 function takes C<int *>.
1120 =item C<SV* save_scalar(GV *gv)>
1122 Equivalent to Perl code C<local $gv>.
1124 =item C<AV* save_ary(GV *gv)>
1126 =item C<HV* save_hash(GV *gv)>
1128 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1130 =item C<void save_item(SV *item)>
1132 Duplicates the current value of C<SV>, on the exit from the current
1133 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1134 using the stored value.
1136 =item C<void save_list(SV **sarg, I32 maxsarg)>
1138 A variant of C<save_item> which takes multiple arguments via an array
1139 C<sarg> of C<SV*> of length C<maxsarg>.
1141 =item C<SV* save_svref(SV **sptr)>
1143 Similar to C<save_scalar>, but will reinstate a C<SV *>.
1145 =item C<void save_aptr(AV **aptr)>
1147 =item C<void save_hptr(HV **hptr)>
1149 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1153 The C<Alias> module implements localization of the basic types within the
1154 I<caller's scope>. People who are interested in how to localize things in
1155 the containing scope should take a look there too.
1159 =head2 XSUBs and the Argument Stack
1161 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1162 An XSUB routine will have a stack that contains the arguments from the Perl
1163 program, and a way to map from the Perl data structures to a C equivalent.
1165 The stack arguments are accessible through the C<ST(n)> macro, which returns
1166 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1167 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1170 Most of the time, output from the C routine can be handled through use of
1171 the RETVAL and OUTPUT directives. However, there are some cases where the
1172 argument stack is not already long enough to handle all the return values.
1173 An example is the POSIX tzname() call, which takes no arguments, but returns
1174 two, the local time zone's standard and summer time abbreviations.
1176 To handle this situation, the PPCODE directive is used and the stack is
1177 extended using the macro:
1181 where C<SP> is the macro that represents the local copy of the stack pointer,
1182 and C<num> is the number of elements the stack should be extended by.
1184 Now that there is room on the stack, values can be pushed on it using the
1185 macros to push IVs, doubles, strings, and SV pointers respectively:
1192 And now the Perl program calling C<tzname>, the two values will be assigned
1195 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1197 An alternate (and possibly simpler) method to pushing values on the stack is
1205 These macros automatically adjust the stack for you, if needed. Thus, you
1206 do not need to call C<EXTEND> to extend the stack.
1208 For more information, consult L<perlxs> and L<perlxstut>.
1210 =head2 Calling Perl Routines from within C Programs
1212 There are four routines that can be used to call a Perl subroutine from
1213 within a C program. These four are:
1215 I32 perl_call_sv(SV*, I32);
1216 I32 perl_call_pv(const char*, I32);
1217 I32 perl_call_method(const char*, I32);
1218 I32 perl_call_argv(const char*, I32, register char**);
1220 The routine most often used is C<perl_call_sv>. The C<SV*> argument
1221 contains either the name of the Perl subroutine to be called, or a
1222 reference to the subroutine. The second argument consists of flags
1223 that control the context in which the subroutine is called, whether
1224 or not the subroutine is being passed arguments, how errors should be
1225 trapped, and how to treat return values.
1227 All four routines return the number of arguments that the subroutine returned
1230 When using any of these routines (except C<perl_call_argv>), the programmer
1231 must manipulate the Perl stack. These include the following macros and
1246 For a detailed description of calling conventions from C to Perl,
1247 consult L<perlcall>.
1249 =head2 Memory Allocation
1251 All memory meant to be used with the Perl API functions should be manipulated
1252 using the macros described in this section. The macros provide the necessary
1253 transparency between differences in the actual malloc implementation that is
1256 It is suggested that you enable the version of malloc that is distributed
1257 with Perl. It keeps pools of various sizes of unallocated memory in
1258 order to satisfy allocation requests more quickly. However, on some
1259 platforms, it may cause spurious malloc or free errors.
1261 New(x, pointer, number, type);
1262 Newc(x, pointer, number, type, cast);
1263 Newz(x, pointer, number, type);
1265 These three macros are used to initially allocate memory.
1267 The first argument C<x> was a "magic cookie" that was used to keep track
1268 of who called the macro, to help when debugging memory problems. However,
1269 the current code makes no use of this feature (most Perl developers now
1270 use run-time memory checkers), so this argument can be any number.
1272 The second argument C<pointer> should be the name of a variable that will
1273 point to the newly allocated memory.
1275 The third and fourth arguments C<number> and C<type> specify how many of
1276 the specified type of data structure should be allocated. The argument
1277 C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1278 should be used if the C<pointer> argument is different from the C<type>
1281 Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1282 to zero out all the newly allocated memory.
1284 Renew(pointer, number, type);
1285 Renewc(pointer, number, type, cast);
1288 These three macros are used to change a memory buffer size or to free a
1289 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1290 match those of C<New> and C<Newc> with the exception of not needing the
1291 "magic cookie" argument.
1293 Move(source, dest, number, type);
1294 Copy(source, dest, number, type);
1295 Zero(dest, number, type);
1297 These three macros are used to move, copy, or zero out previously allocated
1298 memory. The C<source> and C<dest> arguments point to the source and
1299 destination starting points. Perl will move, copy, or zero out C<number>
1300 instances of the size of the C<type> data structure (using the C<sizeof>
1305 The most recent development releases of Perl has been experimenting with
1306 removing Perl's dependency on the "normal" standard I/O suite and allowing
1307 other stdio implementations to be used. This involves creating a new
1308 abstraction layer that then calls whichever implementation of stdio Perl
1309 was compiled with. All XSUBs should now use the functions in the PerlIO
1310 abstraction layer and not make any assumptions about what kind of stdio
1313 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1315 =head2 Putting a C value on Perl stack
1317 A lot of opcodes (this is an elementary operation in the internal perl
1318 stack machine) put an SV* on the stack. However, as an optimization
1319 the corresponding SV is (usually) not recreated each time. The opcodes
1320 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1321 not constantly freed/created.
1323 Each of the targets is created only once (but see
1324 L<Scratchpads and recursion> below), and when an opcode needs to put
1325 an integer, a double, or a string on stack, it just sets the
1326 corresponding parts of its I<target> and puts the I<target> on stack.
1328 The macro to put this target on stack is C<PUSHTARG>, and it is
1329 directly used in some opcodes, as well as indirectly in zillions of
1330 others, which use it via C<(X)PUSH[pni]>.
1334 The question remains on when the SVs which are I<target>s for opcodes
1335 are created. The answer is that they are created when the current unit --
1336 a subroutine or a file (for opcodes for statements outside of
1337 subroutines) -- is compiled. During this time a special anonymous Perl
1338 array is created, which is called a scratchpad for the current
1341 A scratchpad keeps SVs which are lexicals for the current unit and are
1342 targets for opcodes. One can deduce that an SV lives on a scratchpad
1343 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1344 I<target>s have C<SVs_PADTMP> set.
1346 The correspondence between OPs and I<target>s is not 1-to-1. Different
1347 OPs in the compile tree of the unit can use the same target, if this
1348 would not conflict with the expected life of the temporary.
1350 =head2 Scratchpads and recursion
1352 In fact it is not 100% true that a compiled unit contains a pointer to
1353 the scratchpad AV. In fact it contains a pointer to an AV of
1354 (initially) one element, and this element is the scratchpad AV. Why do
1355 we need an extra level of indirection?
1357 The answer is B<recursion>, and maybe (sometime soon) B<threads>. Both
1358 these can create several execution pointers going into the same
1359 subroutine. For the subroutine-child not write over the temporaries
1360 for the subroutine-parent (lifespan of which covers the call to the
1361 child), the parent and the child should have different
1362 scratchpads. (I<And> the lexicals should be separate anyway!)
1364 So each subroutine is born with an array of scratchpads (of length 1).
1365 On each entry to the subroutine it is checked that the current
1366 depth of the recursion is not more than the length of this array, and
1367 if it is, new scratchpad is created and pushed into the array.
1369 The I<target>s on this scratchpad are C<undef>s, but they are already
1370 marked with correct flags.
1372 =head1 Compiled code
1376 Here we describe the internal form your code is converted to by
1377 Perl. Start with a simple example:
1381 This is converted to a tree similar to this one:
1389 (but slightly more complicated). This tree reflects the way Perl
1390 parsed your code, but has nothing to do with the execution order.
1391 There is an additional "thread" going through the nodes of the tree
1392 which shows the order of execution of the nodes. In our simplified
1393 example above it looks like:
1395 $b ---> $c ---> + ---> $a ---> assign-to
1397 But with the actual compile tree for C<$a = $b + $c> it is different:
1398 some nodes I<optimized away>. As a corollary, though the actual tree
1399 contains more nodes than our simplified example, the execution order
1400 is the same as in our example.
1402 =head2 Examining the tree
1404 If you have your perl compiled for debugging (usually done with C<-D
1405 optimize=-g> on C<Configure> command line), you may examine the
1406 compiled tree by specifying C<-Dx> on the Perl command line. The
1407 output takes several lines per node, and for C<$b+$c> it looks like
1412 FLAGS = (SCALAR,KIDS)
1414 TYPE = null ===> (4)
1416 FLAGS = (SCALAR,KIDS)
1418 3 TYPE = gvsv ===> 4
1424 TYPE = null ===> (5)
1426 FLAGS = (SCALAR,KIDS)
1428 4 TYPE = gvsv ===> 5
1434 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1435 not optimized away (one per number in the left column). The immediate
1436 children of the given node correspond to C<{}> pairs on the same level
1437 of indentation, thus this listing corresponds to the tree:
1445 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1446 4 5 6> (node C<6> is not included into above listing), i.e.,
1447 C<gvsv gvsv add whatever>.
1449 =head2 Compile pass 1: check routines
1451 The tree is created by the I<pseudo-compiler> while yacc code feeds it
1452 the constructions it recognizes. Since yacc works bottom-up, so does
1453 the first pass of perl compilation.
1455 What makes this pass interesting for perl developers is that some
1456 optimization may be performed on this pass. This is optimization by
1457 so-called I<check routines>. The correspondence between node names
1458 and corresponding check routines is described in F<opcode.pl> (do not
1459 forget to run C<make regen_headers> if you modify this file).
1461 A check routine is called when the node is fully constructed except
1462 for the execution-order thread. Since at this time there are no
1463 back-links to the currently constructed node, one can do most any
1464 operation to the top-level node, including freeing it and/or creating
1465 new nodes above/below it.
1467 The check routine returns the node which should be inserted into the
1468 tree (if the top-level node was not modified, check routine returns
1471 By convention, check routines have names C<ck_*>. They are usually
1472 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1473 called from F<perly.y>).
1475 =head2 Compile pass 1a: constant folding
1477 Immediately after the check routine is called the returned node is
1478 checked for being compile-time executable. If it is (the value is
1479 judged to be constant) it is immediately executed, and a I<constant>
1480 node with the "return value" of the corresponding subtree is
1481 substituted instead. The subtree is deleted.
1483 If constant folding was not performed, the execution-order thread is
1486 =head2 Compile pass 2: context propagation
1488 When a context for a part of compile tree is known, it is propagated
1489 down through the tree. At this time the context can have 5 values
1490 (instead of 2 for runtime context): void, boolean, scalar, list, and
1491 lvalue. In contrast with the pass 1 this pass is processed from top
1492 to bottom: a node's context determines the context for its children.
1494 Additional context-dependent optimizations are performed at this time.
1495 Since at this moment the compile tree contains back-references (via
1496 "thread" pointers), nodes cannot be free()d now. To allow
1497 optimized-away nodes at this stage, such nodes are null()ified instead
1498 of free()ing (i.e. their type is changed to OP_NULL).
1500 =head2 Compile pass 3: peephole optimization
1502 After the compile tree for a subroutine (or for an C<eval> or a file)
1503 is created, an additional pass over the code is performed. This pass
1504 is neither top-down or bottom-up, but in the execution order (with
1505 additional complications for conditionals). These optimizations are
1506 done in the subroutine peep(). Optimizations performed at this stage
1507 are subject to the same restrictions as in the pass 2.
1509 =head1 The Perl Internal API
1511 WARNING: This information is subject to radical changes prior to
1512 the Perl 5.6 release. Use with caution.
1514 =head2 Background and PERL_IMPLICIT_CONTEXT
1516 The Perl interpreter can be regarded as a closed box: it has an API
1517 for feeding it code or otherwise making it do things, but it also has
1518 functions for its own use. This smells a lot like an object, and
1519 there are ways for you to build Perl so that you can have multiple
1520 interpreters, with one interpreter represented either as a C++ object,
1521 a C structure, or inside a thread. The thread, the C structure, or
1522 the C++ object will contain all the context, the state of that
1525 Three macros control the major Perl build flavors: MULTIPLICITY,
1526 USE_THREADS and PERL_OBJECT. The MULTIPLICITY build has a C structure
1527 that packages all the interpreter state, there is a similar thread-specific
1528 data structure under USE_THREADS, and the PERL_OBJECT build has a C++
1529 class to maintain interpreter state. In all three cases,
1530 PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1531 support for passing in a "hidden" first argument that represents all three
1534 All this obviously requires a way for the Perl internal functions to be
1535 C++ methods, subroutines taking some kind of structure as the first
1536 argument, or subroutines taking nothing as the first argument. To
1537 enable these three very different ways of building the interpreter,
1538 the Perl source (as it does in so many other situations) makes heavy
1539 use of macros and subroutine naming conventions.
1541 First problem: deciding which functions will be public API functions and
1542 which will be private. Those functions whose names begin C<Perl_> are
1543 public, and those whose names begin C<S_> are private (think "S" for
1544 "secret" or "static").
1546 Some functions have no prefix (e.g., restore_rsfp in toke.c). These
1547 are not parts of the object or pseudo-structure because you need to
1548 pass pointers to them to other subroutines.
1550 Second problem: there must be a syntax so that the same subroutine
1551 declarations and calls can pass a structure as their first argument,
1552 or pass nothing. To solve this, the subroutines are named and
1553 declared in a particular way. Here's a typical start of a static
1554 function used within the Perl guts:
1557 S_incline(pTHX_ char *s)
1559 STATIC becomes "static" in C, and is #define'd to nothing in C++.
1561 A public function (i.e. part of the internal API, but not necessarily
1562 sanctioned for use in extensions) begins like this:
1565 Perl_sv_setsv(pTHX_ SV* dsv, SV* ssv)
1567 C<pTHX_> is one of a number of macros (in perl.h) that hide the
1568 details of the interpreter's context. THX stands for "thread", "this",
1569 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1570 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
1571 or 'd' for B<d>eclaration.
1573 When Perl is built without PERL_IMPLICIT_CONTEXT, there is no first
1574 argument containing the interpreter's context. The trailing underscore
1575 in the pTHX_ macro indicates that the macro expansion needs a comma
1576 after the context argument because other arguments follow it. If
1577 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
1578 subroutine is not prototyped to take the extra argument. The form of the
1579 macro without the trailing underscore is used when there are no additional
1582 When a core function calls another, it must pass the context. This
1583 is normally hidden via macros. Consider C<sv_setsv>. It expands
1584 something like this:
1586 ifdef PERL_IMPLICIT_CONTEXT
1587 define sv_setsv(a,b) Perl_sv_setsv(aTHX_ a, b)
1588 /* can't do this for vararg functions, see below */
1590 define sv_setsv Perl_sv_setsv
1593 This works well, and means that XS authors can gleefully write:
1597 and still have it work under all the modes Perl could have been
1600 Under PERL_OBJECT in the core, that will translate to either:
1602 CPerlObj::Perl_sv_setsv(foo,bar); # in CPerlObj functions,
1603 # C++ takes care of 'this'
1606 pPerl->Perl_sv_setsv(foo,bar); # in truly static functions,
1609 Under PERL_OBJECT in extensions (aka PERL_CAPI), or under
1610 MULTIPLICITY/USE_THREADS w/ PERL_IMPLICIT_CONTEXT in both core
1611 and extensions, it will be:
1613 Perl_sv_setsv(aTHX_ foo, bar); # the canonical Perl "API"
1614 # for all build flavors
1616 This doesn't work so cleanly for varargs functions, though, as macros
1617 imply that the number of arguments is known in advance. Instead we
1618 either need to spell them out fully, passing C<aTHX_> as the first
1619 argument (the Perl core tends to do this with functions like
1620 Perl_warner), or use a context-free version.
1622 The context-free version of Perl_warner is called
1623 Perl_warner_nocontext, and does not take the extra argument. Instead
1624 it does dTHX; to get the context from thread-local storage. We
1625 C<#define warner Perl_warner_nocontext> so that extensions get source
1626 compatibility at the expense of performance. (Passing an arg is
1627 cheaper than grabbing it from thread-local storage.)
1629 You can ignore [pad]THX[xo] when browsing the Perl headers/sources.
1630 Those are strictly for use within the core. Extensions and embedders
1631 need only be aware of [pad]THX.
1633 =head2 How do I use all this in extensions?
1635 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1636 any functions in the Perl API will need to pass the initial context
1637 argument somehow. The kicker is that you will need to write it in
1638 such a way that the extension still compiles when Perl hasn't been
1639 built with PERL_IMPLICIT_CONTEXT enabled.
1641 There are three ways to do this. First, the easy but inefficient way,
1642 which is also the default, in order to maintain source compatibility
1643 with extensions: whenever XSUB.h is #included, it redefines the aTHX
1644 and aTHX_ macros to call a function that will return the context.
1645 Thus, something like:
1649 in your extesion will translate to this when PERL_IMPLICIT_CONTEXT is
1652 Perl_sv_setsv(GetPerlInterpreter(), asv, bsv);
1654 or to this otherwise:
1656 Perl_sv_setsv(asv, bsv);
1658 You have to do nothing new in your extension to get this; since
1659 the Perl library provides GetPerlInterpreter(), it will all just
1662 The second, more efficient way is to use the following template for
1665 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1670 static my_private_function(int arg1, int arg2);
1673 my_private_function(int arg1, int arg2)
1675 dTHX; /* fetch context */
1676 ... call many Perl API functions ...
1681 MODULE = Foo PACKAGE = Foo
1689 my_private_function(arg, 10);
1691 Note that the only two changes from the normal way of writing an
1692 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
1693 including the Perl headers, followed by a C<dTHX;> declaration at
1694 the start of every function that will call the Perl API. (You'll
1695 know which functions need this, because the C compiler will complain
1696 that there's an undeclared identifier in those functions.) No changes
1697 are needed for the XSUBs themselves, because the XS() macro is
1698 correctly defined to pass in the implicit context if needed.
1700 The third, even more efficient way is to ape how it is done within
1704 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1709 /* pTHX_ only needed for functions that call Perl API */
1710 static my_private_function(pTHX_ int arg1, int arg2);
1713 my_private_function(pTHX_ int arg1, int arg2)
1715 /* dTHX; not needed here, because THX is an argument */
1716 ... call Perl API functions ...
1721 MODULE = Foo PACKAGE = Foo
1729 my_private_function(aTHX_ arg, 10);
1731 This implementation never has to fetch the context using a function
1732 call, since it is always passed as an extra argument. Depending on
1733 your needs for simplicity or efficiency, you may mix the previous
1734 two approaches freely.
1736 Never add a comma after C<pTHX> yourself--always use the form of the
1737 macro with the underscore for functions that take explicit arguments,
1738 or the form without the argument for functions with no explicit arguments.
1740 =head2 Future Plans and PERL_IMPLICIT_SYS
1742 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
1743 that the interpreter knows about itself and pass it around, so too are
1744 there plans to allow the interpreter to bundle up everything it knows
1745 about the environment it's running on. This is enabled with the
1746 PERL_IMPLICIT_SYS macro. Currently it only works with PERL_OBJECT,
1747 but is mostly there for MULTIPLICITY and USE_THREADS (see inside
1750 This allows the ability to provide an extra pointer (called the "host"
1751 environment) for all the system calls. This makes it possible for
1752 all the system stuff to maintain their own state, broken down into
1753 seven C structures. These are thin wrappers around the usual system
1754 calls (see win32/perllib.c) for the default perl executable, but for a
1755 more ambitious host (like the one that would do fork() emulation) all
1756 the extra work needed to pretend that different interpreters are
1757 actually different "processes", would be done here.
1759 The Perl engine/interpreter and the host are orthogonal entities.
1760 There could be one or more interpreters in a process, and one or
1761 more "hosts", with free association between them.
1765 This is a listing of functions, macros, flags, and variables that may be
1766 used by extension writers. The interfaces of any functions that are not
1767 listed here are subject to change without notice. For this reason,
1768 blindly using functions listed in proto.h is to be avoided when writing
1771 Note that all Perl API global variables must be referenced with the C<PL_>
1772 prefix. Some macros are provided for compatibility with the older,
1773 unadorned names, but this support may be disabled in a future release.
1775 The sort order of the listing is case insensitive, with any
1776 occurrences of '_' ignored for the purpose of sorting.
1782 Clears an array, making it empty. Does not free the memory used by the
1785 void av_clear (AV* ar)
1789 Pre-extend an array. The C<key> is the index to which the array should be
1792 void av_extend (AV* ar, I32 key)
1796 Returns the SV at the specified index in the array. The C<key> is the
1797 index. If C<lval> is set then the fetch will be part of a store. Check
1798 that the return value is non-null before dereferencing it to a C<SV*>.
1800 See L<Understanding the Magic of Tied Hashes and Arrays> for more
1801 information on how to use this function on tied arrays.
1803 SV** av_fetch (AV* ar, I32 key, I32 lval)
1807 Same as C<av_len()>. Deprecated, use C<av_len()> instead.
1811 Returns the highest index in the array. Returns -1 if the array is empty.
1817 Creates a new AV and populates it with a list of SVs. The SVs are copied
1818 into the array, so they may be freed after the call to av_make. The new AV
1819 will have a reference count of 1.
1821 AV* av_make (I32 size, SV** svp)
1825 Pops an SV off the end of the array. Returns C<&PL_sv_undef> if the array is
1832 Pushes an SV onto the end of the array. The array will grow automatically
1833 to accommodate the addition.
1835 void av_push (AV* ar, SV* val)
1839 Shifts an SV off the beginning of the array.
1841 SV* av_shift (AV* ar)
1845 Stores an SV in an array. The array index is specified as C<key>. The
1846 return value will be NULL if the operation failed or if the value did not
1847 need to be actually stored within the array (as in the case of tied arrays).
1848 Otherwise it can be dereferenced to get the original C<SV*>. Note that the
1849 caller is responsible for suitably incrementing the reference count of C<val>
1850 before the call, and decrementing it if the function returned NULL.
1852 See L<Understanding the Magic of Tied Hashes and Arrays> for more
1853 information on how to use this function on tied arrays.
1855 SV** av_store (AV* ar, I32 key, SV* val)
1859 Undefines the array. Frees the memory used by the array itself.
1861 void av_undef (AV* ar)
1865 Unshift the given number of C<undef> values onto the beginning of the
1866 array. The array will grow automatically to accommodate the addition.
1867 You must then use C<av_store> to assign values to these new elements.
1869 void av_unshift (AV* ar, I32 num)
1873 Variable which is setup by C<xsubpp> to indicate the class name for a C++ XS
1874 constructor. This is always a C<char*>. See C<THIS> and
1875 L<perlxs/"Using XS With C++">.
1879 The XSUB-writer's interface to the C C<memcpy> function. The C<s> is the
1880 source, C<d> is the destination, C<n> is the number of items, and C<t> is
1881 the type. May fail on overlapping copies. See also C<Move>.
1883 void Copy( s, d, n, t )
1887 This is the XSUB-writer's interface to Perl's C<die> function. Use this
1888 function the same way you use the C C<printf> function. See C<warn>.
1892 Returns the stash of the CV.
1894 HV* CvSTASH( SV* sv )
1898 When Perl is run in debugging mode, with the B<-d> switch, this SV is a
1899 boolean which indicates whether subs are being single-stepped.
1900 Single-stepping is automatically turned on after every step. This is the C
1901 variable which corresponds to Perl's $DB::single variable. See C<PL_DBsub>.
1905 When Perl is run in debugging mode, with the B<-d> switch, this GV contains
1906 the SV which holds the name of the sub being debugged. This is the C
1907 variable which corresponds to Perl's $DB::sub variable. See C<PL_DBsingle>.
1908 The sub name can be found by
1910 SvPV( GvSV( PL_DBsub ), len )
1914 Trace variable used when Perl is run in debugging mode, with the B<-d>
1915 switch. This is the C variable which corresponds to Perl's $DB::trace
1916 variable. See C<PL_DBsingle>.
1920 Declare a stack marker variable, C<mark>, for the XSUB. See C<MARK> and
1925 Saves the original stack mark for the XSUB. See C<ORIGMARK>.
1929 The C variable which corresponds to Perl's $^W warning variable.
1933 Declares a local copy of perl's stack pointer for the XSUB, available via
1934 the C<SP> macro. See C<SP>.
1938 Sets up stack and mark pointers for an XSUB, calling dSP and dMARK. This is
1939 usually handled automatically by C<xsubpp>. Declares the C<items> variable
1940 to indicate the number of items on the stack.
1944 Sets up the C<ix> variable for an XSUB which has aliases. This is usually
1945 handled automatically by C<xsubpp>.
1949 Switches filehandle to binmode. C<iotype> is what C<IoTYPE(io)> would
1952 do_binmode(fp, iotype, TRUE);
1956 Opening bracket on a callback. See C<LEAVE> and L<perlcall>.
1962 Used to extend the argument stack for an XSUB's return values.
1968 Analyses the string in order to make fast searches on it using fbm_instr() --
1969 the Boyer-Moore algorithm.
1971 void fbm_compile(SV* sv, U32 flags)
1975 Returns the location of the SV in the string delimited by C<str> and
1976 C<strend>. It returns C<Nullch> if the string can't be found. The
1977 C<sv> does not have to be fbm_compiled, but the search will not be as
1980 char* fbm_instr(char *str, char *strend, SV *sv, U32 flags)
1984 Closing bracket for temporaries on a callback. See C<SAVETMPS> and
1991 Used to indicate array context. See C<GIMME_V>, C<GIMME> and L<perlcall>.
1995 Indicates that arguments returned from a callback should be discarded. See
2000 Used to force a Perl C<eval> wrapper around a callback. See L<perlcall>.
2004 A backward-compatible version of C<GIMME_V> which can only return
2005 C<G_SCALAR> or C<G_ARRAY>; in a void context, it returns C<G_SCALAR>.
2009 The XSUB-writer's equivalent to Perl's C<wantarray>. Returns
2010 C<G_VOID>, C<G_SCALAR> or C<G_ARRAY> for void, scalar or array
2011 context, respectively.
2015 Indicates that no arguments are being sent to a callback. See L<perlcall>.
2019 Used to indicate scalar context. See C<GIMME_V>, C<GIMME>, and L<perlcall>.
2023 Returns the glob with the given C<name> and a defined subroutine or
2024 C<NULL>. The glob lives in the given C<stash>, or in the stashes
2025 accessible via @ISA and @UNIVERSAL.
2027 The argument C<level> should be either 0 or -1. If C<level==0>, as a
2028 side-effect creates a glob with the given C<name> in the given
2029 C<stash> which in the case of success contains an alias for the
2030 subroutine, and sets up caching info for this glob. Similarly for all
2031 the searched stashes.
2033 This function grants C<"SUPER"> token as a postfix of the stash name.
2035 The GV returned from C<gv_fetchmeth> may be a method cache entry,
2036 which is not visible to Perl code. So when calling C<perl_call_sv>,
2037 you should not use the GV directly; instead, you should use the
2038 method's CV, which can be obtained from the GV with the C<GvCV> macro.
2040 GV* gv_fetchmeth (HV* stash, const char* name, STRLEN len, I32 level)
2042 =item gv_fetchmethod
2044 =item gv_fetchmethod_autoload
2046 Returns the glob which contains the subroutine to call to invoke the
2047 method on the C<stash>. In fact in the presence of autoloading this may
2048 be the glob for "AUTOLOAD". In this case the corresponding variable
2049 $AUTOLOAD is already setup.
2051 The third parameter of C<gv_fetchmethod_autoload> determines whether AUTOLOAD
2052 lookup is performed if the given method is not present: non-zero means
2053 yes, look for AUTOLOAD; zero means no, don't look for AUTOLOAD. Calling
2054 C<gv_fetchmethod> is equivalent to calling C<gv_fetchmethod_autoload> with a
2055 non-zero C<autoload> parameter.
2057 These functions grant C<"SUPER"> token as a prefix of the method name.
2059 Note that if you want to keep the returned glob for a long time, you
2060 need to check for it being "AUTOLOAD", since at the later time the call
2061 may load a different subroutine due to $AUTOLOAD changing its value.
2062 Use the glob created via a side effect to do this.
2064 These functions have the same side-effects and as C<gv_fetchmeth> with
2065 C<level==0>. C<name> should be writable if contains C<':'> or C<'\''>.
2066 The warning against passing the GV returned by C<gv_fetchmeth> to
2067 C<perl_call_sv> apply equally to these functions.
2069 GV* gv_fetchmethod (HV* stash, const char* name)
2070 GV* gv_fetchmethod_autoload (HV* stash, const char* name, I32 autoload)
2074 Used to indicate void context. See C<GIMME_V> and L<perlcall>.
2078 Returns a pointer to the stash for a specified package. If C<create> is set
2079 then the package will be created if it does not already exist. If C<create>
2080 is not set and the package does not exist then NULL is returned.
2082 HV* gv_stashpv (const char* name, I32 create)
2086 Returns a pointer to the stash for a specified package. See C<gv_stashpv>.
2088 HV* gv_stashsv (SV* sv, I32 create)
2092 Return the SV from the GV.
2096 This flag, used in the length slot of hash entries and magic
2097 structures, specifies the structure contains a C<SV*> pointer where a
2098 C<char*> pointer is to be expected. (For information only--not to be used).
2102 Returns the computed hash stored in the hash entry.
2108 Returns the actual pointer stored in the key slot of the hash entry.
2109 The pointer may be either C<char*> or C<SV*>, depending on the value of
2110 C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros
2111 are usually preferable for finding the value of a key.
2117 If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
2118 holds an C<SV*> key. Otherwise, holds the actual length of the key.
2119 Can be assigned to. The C<HePV()> macro is usually preferable for finding
2126 Returns the key slot of the hash entry as a C<char*> value, doing any
2127 necessary dereferencing of possibly C<SV*> keys. The length of
2128 the string is placed in C<len> (this is a macro, so do I<not> use
2129 C<&len>). If you do not care about what the length of the key is,
2130 you may use the global variable C<PL_na>, though this is rather less
2131 efficient than using a local variable. Remember though, that hash
2132 keys in perl are free to contain embedded nulls, so using C<strlen()>
2133 or similar is not a good way to find the length of hash keys.
2134 This is very similar to the C<SvPV()> macro described elsewhere in
2137 char* HePV(HE* he, STRLEN len)
2141 Returns the key as an C<SV*>, or C<Nullsv> if the hash entry
2142 does not contain an C<SV*> key.
2148 Returns the key as an C<SV*>. Will create and return a temporary
2149 mortal C<SV*> if the hash entry contains only a C<char*> key.
2151 HeSVKEY_force(HE* he)
2155 Sets the key to a given C<SV*>, taking care to set the appropriate flags
2156 to indicate the presence of an C<SV*> key, and returns the same C<SV*>.
2158 HeSVKEY_set(HE* he, SV* sv)
2162 Returns the value slot (type C<SV*>) stored in the hash entry.
2168 Clears a hash, making it empty.
2170 void hv_clear (HV* tb)
2174 Deletes a key/value pair in the hash. The value SV is removed from the hash
2175 and returned to the caller. The C<klen> is the length of the key. The
2176 C<flags> value will normally be zero; if set to G_DISCARD then NULL will be
2179 SV* hv_delete (HV* tb, const char* key, U32 klen, I32 flags)
2183 Deletes a key/value pair in the hash. The value SV is removed from the hash
2184 and returned to the caller. The C<flags> value will normally be zero; if set
2185 to G_DISCARD then NULL will be returned. C<hash> can be a valid precomputed
2186 hash value, or 0 to ask for it to be computed.
2188 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash)
2192 Returns a boolean indicating whether the specified hash key exists. The
2193 C<klen> is the length of the key.
2195 bool hv_exists (HV* tb, const char* key, U32 klen)
2199 Returns a boolean indicating whether the specified hash key exists. C<hash>
2200 can be a valid precomputed hash value, or 0 to ask for it to be computed.
2202 bool hv_exists_ent (HV* tb, SV* key, U32 hash)
2206 Returns the SV which corresponds to the specified key in the hash. The
2207 C<klen> is the length of the key. If C<lval> is set then the fetch will be
2208 part of a store. Check that the return value is non-null before
2209 dereferencing it to a C<SV*>.
2211 See L<Understanding the Magic of Tied Hashes and Arrays> for more
2212 information on how to use this function on tied hashes.
2214 SV** hv_fetch (HV* tb, const char* key, U32 klen, I32 lval)
2218 Returns the hash entry which corresponds to the specified key in the hash.
2219 C<hash> must be a valid precomputed hash number for the given C<key>, or
2220 0 if you want the function to compute it. IF C<lval> is set then the
2221 fetch will be part of a store. Make sure the return value is non-null
2222 before accessing it. The return value when C<tb> is a tied hash
2223 is a pointer to a static location, so be sure to make a copy of the
2224 structure if you need to store it somewhere.
2226 See L<Understanding the Magic of Tied Hashes and Arrays> for more
2227 information on how to use this function on tied hashes.
2229 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash)
2233 Prepares a starting point to traverse a hash table.
2235 I32 hv_iterinit (HV* tb)
2237 Returns the number of keys in the hash (i.e. the same as C<HvKEYS(tb)>).
2238 The return value is currently only meaningful for hashes without tie
2241 NOTE: Before version 5.004_65, C<hv_iterinit> used to return the number
2242 of hash buckets that happen to be in use. If you still need that
2243 esoteric value, you can get it through the macro C<HvFILL(tb)>.
2247 Returns the key from the current position of the hash iterator. See
2250 char* hv_iterkey (HE* entry, I32* retlen)
2254 Returns the key as an C<SV*> from the current position of the hash
2255 iterator. The return value will always be a mortal copy of the
2256 key. Also see C<hv_iterinit>.
2258 SV* hv_iterkeysv (HE* entry)
2262 Returns entries from a hash iterator. See C<hv_iterinit>.
2264 HE* hv_iternext (HV* tb)
2268 Performs an C<hv_iternext>, C<hv_iterkey>, and C<hv_iterval> in one
2271 SV* hv_iternextsv (HV* hv, char** key, I32* retlen)
2275 Returns the value from the current position of the hash iterator. See
2278 SV* hv_iterval (HV* tb, HE* entry)
2282 Adds magic to a hash. See C<sv_magic>.
2284 void hv_magic (HV* hv, GV* gv, int how)
2288 Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
2290 char* HvNAME (HV* stash)
2294 Stores an SV in a hash. The hash key is specified as C<key> and C<klen> is
2295 the length of the key. The C<hash> parameter is the precomputed hash
2296 value; if it is zero then Perl will compute it. The return value will be
2297 NULL if the operation failed or if the value did not need to be actually
2298 stored within the hash (as in the case of tied hashes). Otherwise it can
2299 be dereferenced to get the original C<SV*>. Note that the caller is
2300 responsible for suitably incrementing the reference count of C<val>
2301 before the call, and decrementing it if the function returned NULL.
2303 See L<Understanding the Magic of Tied Hashes and Arrays> for more
2304 information on how to use this function on tied hashes.
2306 SV** hv_store (HV* tb, const char* key, U32 klen, SV* val, U32 hash)
2310 Stores C<val> in a hash. The hash key is specified as C<key>. The C<hash>
2311 parameter is the precomputed hash value; if it is zero then Perl will
2312 compute it. The return value is the new hash entry so created. It will be
2313 NULL if the operation failed or if the value did not need to be actually
2314 stored within the hash (as in the case of tied hashes). Otherwise the
2315 contents of the return value can be accessed using the C<He???> macros
2316 described here. Note that the caller is responsible for suitably
2317 incrementing the reference count of C<val> before the call, and decrementing
2318 it if the function returned NULL.
2320 See L<Understanding the Magic of Tied Hashes and Arrays> for more
2321 information on how to use this function on tied hashes.
2323 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash)
2329 void hv_undef (HV* tb)
2333 Returns a boolean indicating whether the C C<char> is an ascii alphanumeric
2336 int isALNUM (char c)
2340 Returns a boolean indicating whether the C C<char> is an ascii alphabetic
2343 int isALPHA (char c)
2347 Returns a boolean indicating whether the C C<char> is an ascii digit.
2349 int isDIGIT (char c)
2353 Returns a boolean indicating whether the C C<char> is a lowercase character.
2355 int isLOWER (char c)
2359 Returns a boolean indicating whether the C C<char> is whitespace.
2361 int isSPACE (char c)
2365 Returns a boolean indicating whether the C C<char> is an uppercase character.
2367 int isUPPER (char c)
2371 Variable which is setup by C<xsubpp> to indicate the number of items on the
2372 stack. See L<perlxs/"Variable-length Parameter Lists">.
2376 Variable which is setup by C<xsubpp> to indicate which of an XSUB's aliases
2377 was used to invoke it. See L<perlxs/"The ALIAS: Keyword">.
2381 Closing bracket on a callback. See C<ENTER> and L<perlcall>.
2385 =item looks_like_number
2387 Test if an the content of an SV looks like a number (or is a number).
2389 int looks_like_number(SV*)
2394 Stack marker variable for the XSUB. See C<dMARK>.
2398 Clear something magical that the SV represents. See C<sv_magic>.
2400 int mg_clear (SV* sv)
2404 Copies the magic from one SV to another. See C<sv_magic>.
2406 int mg_copy (SV *, SV *, const char *, STRLEN)
2410 Finds the magic pointer for type matching the SV. See C<sv_magic>.
2412 MAGIC* mg_find (SV* sv, int type)
2416 Free any magic storage used by the SV. See C<sv_magic>.
2418 int mg_free (SV* sv)
2422 Do magic after a value is retrieved from the SV. See C<sv_magic>.
2428 Report on the SV's length. See C<sv_magic>.
2434 Turns on the magical status of an SV. See C<sv_magic>.
2436 void mg_magical (SV* sv)
2440 Do magic after a value is assigned to the SV. See C<sv_magic>.
2446 C<modglobal> is a general purpose, interpreter global HV for use by
2447 extensions that need to keep information on a per-interpreter basis.
2448 In a pinch, it can also be used as a symbol table for extensions
2449 to share data among each other. It is a good idea to use keys
2450 prefixed by the package name of the extension that owns the data.
2454 The XSUB-writer's interface to the C C<memmove> function. The C<s> is the
2455 source, C<d> is the destination, C<n> is the number of items, and C<t> is
2456 the type. Can do overlapping moves. See also C<Copy>.
2458 void Move( s, d, n, t )
2462 A convenience variable which is typically used with C<SvPV> when one doesn't
2463 care about the length of the string. It is usually more efficient to
2464 either declare a local variable and use that instead or to use the C<SvPV_nolen>
2469 The XSUB-writer's interface to the C C<malloc> function.
2471 void* New( x, void *ptr, int size, type )
2475 Creates a new AV. The reference count is set to 1.
2481 The XSUB-writer's interface to the C C<malloc> function, with cast.
2483 void* Newc( x, void *ptr, int size, type, cast )
2487 Creates a constant sub equivalent to Perl C<sub FOO () { 123 }>
2488 which is eligible for inlining at compile-time.
2490 void newCONSTSUB(HV* stash, char* name, SV* sv)
2494 Creates a new HV. The reference count is set to 1.
2500 Creates an RV wrapper for an SV. The reference count for the original SV is
2503 SV* newRV_inc (SV* ref)
2505 For historical reasons, "newRV" is a synonym for "newRV_inc".
2509 Creates an RV wrapper for an SV. The reference count for the original
2510 SV is B<not> incremented.
2512 SV* newRV_noinc (SV* ref)
2516 Creates a new SV. A non-zero C<len> parameter indicates the number of
2517 bytes of preallocated string space the SV should have. An extra byte
2518 for a tailing NUL is also reserved. (SvPOK is not set for the SV even
2519 if string space is allocated.) The reference count for the new SV is
2520 set to 1. C<id> is an integer id between 0 and 1299 (used to identify
2523 SV* NEWSV (int id, STRLEN len)
2527 Creates a new SV and copies an integer into it. The reference count for the
2534 Creates a new SV and copies a double into it. The reference count for the
2541 Creates a new SV and copies a string into it. The reference count for the
2542 SV is set to 1. If C<len> is zero, Perl will compute the length using
2543 strlen(). For efficiency, consider using C<newSVpvn> instead.
2545 SV* newSVpv (const char* s, STRLEN len)
2549 Creates a new SV an initialize it with the string formatted like
2552 SV* newSVpvf(const char* pat, ...)
2556 Creates a new SV and copies a string into it. The reference count for the
2557 SV is set to 1. Note that if C<len> is zero, Perl will create a zero length
2558 string. You are responsible for ensuring that the source string is at least
2561 SV* newSVpvn (const char* s, STRLEN len)
2565 Creates a new SV for the RV, C<rv>, to point to. If C<rv> is not an RV then
2566 it will be upgraded to one. If C<classname> is non-null then the new SV will
2567 be blessed in the specified package. The new SV is returned and its
2568 reference count is 1.
2570 SV* newSVrv (SV* rv, const char* classname)
2574 Creates a new SV which is an exact duplicate of the original SV.
2576 SV* newSVsv (SV* old)
2580 Used by C<xsubpp> to hook up XSUBs as Perl subs.
2584 Used by C<xsubpp> to hook up XSUBs as Perl subs. Adds Perl prototypes to
2589 The XSUB-writer's interface to the C C<malloc> function. The allocated
2590 memory is zeroed with C<memzero>.
2592 void* Newz( x, void *ptr, int size, type )
2600 Null character pointer.
2616 The original stack mark for the XSUB. See C<dORIGMARK>.
2620 Allocates a new Perl interpreter. See L<perlembed>.
2622 =item perl_call_argv
2624 Performs a callback to the specified Perl sub. See L<perlcall>.
2626 I32 perl_call_argv (const char* subname, I32 flags, char** argv)
2628 =item perl_call_method
2630 Performs a callback to the specified Perl method. The blessed object must
2631 be on the stack. See L<perlcall>.
2633 I32 perl_call_method (const char* methname, I32 flags)
2637 Performs a callback to the specified Perl sub. See L<perlcall>.
2639 I32 perl_call_pv (const char* subname, I32 flags)
2643 Performs a callback to the Perl sub whose name is in the SV. See
2646 I32 perl_call_sv (SV* sv, I32 flags)
2648 =item perl_construct
2650 Initializes a new Perl interpreter. See L<perlembed>.
2654 Shuts down a Perl interpreter. See L<perlembed>.
2658 Tells Perl to C<eval> the string in the SV.
2660 I32 perl_eval_sv (SV* sv, I32 flags)
2664 Tells Perl to C<eval> the given string and return an SV* result.
2666 SV* perl_eval_pv (const char* p, I32 croak_on_error)
2670 Releases a Perl interpreter. See L<perlembed>.
2674 Returns the AV of the specified Perl array. If C<create> is set and the
2675 Perl variable does not exist then it will be created. If C<create> is not
2676 set and the variable does not exist then NULL is returned.
2678 AV* perl_get_av (const char* name, I32 create)
2682 Returns the CV of the specified Perl subroutine. If C<create> is set and
2683 the Perl subroutine does not exist then it will be declared (which has
2684 the same effect as saying C<sub name;>). If C<create> is not
2685 set and the subroutine does not exist then NULL is returned.
2687 CV* perl_get_cv (const char* name, I32 create)
2691 Returns the HV of the specified Perl hash. If C<create> is set and the Perl
2692 variable does not exist then it will be created. If C<create> is not
2693 set and the variable does not exist then NULL is returned.
2695 HV* perl_get_hv (const char* name, I32 create)
2699 Returns the SV of the specified Perl scalar. If C<create> is set and the
2700 Perl variable does not exist then it will be created. If C<create> is not
2701 set and the variable does not exist then NULL is returned.
2703 SV* perl_get_sv (const char* name, I32 create)
2707 Tells a Perl interpreter to parse a Perl script. See L<perlembed>.
2709 =item perl_require_pv
2711 Tells Perl to C<require> a module.
2713 void perl_require_pv (const char* pv)
2717 Tells a Perl interpreter to run. See L<perlembed>.
2721 Pops an integer off the stack.
2727 Pops a long off the stack.
2733 Pops a string off the stack.
2739 Pops a double off the stack.
2745 Pops an SV off the stack.
2751 Opening bracket for arguments on a callback. See C<PUTBACK> and L<perlcall>.
2757 Push an integer onto the stack. The stack must have room for this element.
2758 Handles 'set' magic. See C<XPUSHi>.
2764 Push a double onto the stack. The stack must have room for this element.
2765 Handles 'set' magic. See C<XPUSHn>.
2767 void PUSHn(double d)
2771 Push a string onto the stack. The stack must have room for this element.
2772 The C<len> indicates the length of the string. Handles 'set' magic. See
2775 void PUSHp(char *c, int len )
2779 Push an SV onto the stack. The stack must have room for this element. Does
2780 not handle 'set' magic. See C<XPUSHs>.
2786 Push an unsigned integer onto the stack. The stack must have room for
2787 this element. See C<XPUSHu>.
2789 void PUSHu(unsigned int d)
2794 Closing bracket for XSUB arguments. This is usually handled by C<xsubpp>.
2795 See C<PUSHMARK> and L<perlcall> for other uses.
2801 The XSUB-writer's interface to the C C<realloc> function.
2803 void* Renew( void *ptr, int size, type )
2807 The XSUB-writer's interface to the C C<realloc> function, with cast.
2809 void* Renewc( void *ptr, int size, type, cast )
2813 Variable which is setup by C<xsubpp> to hold the return value for an XSUB.
2814 This is always the proper type for the XSUB.
2815 See L<perlxs/"The RETVAL Variable">.
2819 The XSUB-writer's interface to the C C<free> function.
2823 The XSUB-writer's interface to the C C<malloc> function.
2827 The XSUB-writer's interface to the C C<realloc> function.
2831 Copy a string to a safe spot. This does not use an SV.
2833 char* savepv (const char* sv)
2837 Copy a string to a safe spot. The C<len> indicates number of bytes to
2838 copy. This does not use an SV.
2840 char* savepvn (const char* sv, I32 len)
2844 Opening bracket for temporaries on a callback. See C<FREETMPS> and
2851 Stack pointer. This is usually handled by C<xsubpp>. See C<dSP> and
2856 Refetch the stack pointer. Used after a callback. See L<perlcall>.
2862 Used to access elements on the XSUB's stack.
2868 Test two strings to see if they are equal. Returns true or false.
2870 int strEQ( char *s1, char *s2 )
2874 Test two strings to see if the first, C<s1>, is greater than or equal to the
2875 second, C<s2>. Returns true or false.
2877 int strGE( char *s1, char *s2 )
2881 Test two strings to see if the first, C<s1>, is greater than the second,
2882 C<s2>. Returns true or false.
2884 int strGT( char *s1, char *s2 )
2888 Test two strings to see if the first, C<s1>, is less than or equal to the
2889 second, C<s2>. Returns true or false.
2891 int strLE( char *s1, char *s2 )
2895 Test two strings to see if the first, C<s1>, is less than the second,
2896 C<s2>. Returns true or false.
2898 int strLT( char *s1, char *s2 )
2902 Test two strings to see if they are different. Returns true or false.
2904 int strNE( char *s1, char *s2 )
2908 Test two strings to see if they are equal. The C<len> parameter indicates
2909 the number of bytes to compare. Returns true or false.
2910 (A wrapper for C<strncmp>).
2912 int strnEQ( const char *s1, const char *s2, size_t len )
2916 Test two strings to see if they are different. The C<len> parameter
2917 indicates the number of bytes to compare. Returns true or false.
2918 (A wrapper for C<strncmp>).
2920 int strnNE( const char *s1, const char *s2, size_t len )
2924 Marks an SV as mortal. The SV will be destroyed when the current context
2927 SV* sv_2mortal (SV* sv)
2931 Blesses an SV into a specified package. The SV must be an RV. The package
2932 must be designated by its stash (see C<gv_stashpv()>). The reference count
2933 of the SV is unaffected.
2935 SV* sv_bless (SV* sv, HV* stash)
2939 Concatenates the string onto the end of the string which is in the SV.
2940 Handles 'get' magic, but not 'set' magic. See C<sv_catpv_mg>.
2942 void sv_catpv (SV* sv, const char* ptr)
2946 Like C<sv_catpv>, but also handles 'set' magic.
2948 void sv_catpvn (SV* sv, const char* ptr)
2952 Concatenates the string onto the end of the string which is in the SV. The
2953 C<len> indicates number of bytes to copy. Handles 'get' magic, but not
2954 'set' magic. See C<sv_catpvn_mg>.
2956 void sv_catpvn (SV* sv, const char* ptr, STRLEN len)
2960 Like C<sv_catpvn>, but also handles 'set' magic.
2962 void sv_catpvn_mg (SV* sv, const char* ptr, STRLEN len)
2966 Processes its arguments like C<sprintf> and appends the formatted output
2967 to an SV. Handles 'get' magic, but not 'set' magic. C<SvSETMAGIC()> must
2968 typically be called after calling this function to handle 'set' magic.
2970 void sv_catpvf (SV* sv, const char* pat, ...)
2974 Like C<sv_catpvf>, but also handles 'set' magic.
2976 void sv_catpvf_mg (SV* sv, const char* pat, ...)
2980 Concatenates the string from SV C<ssv> onto the end of the string in SV
2981 C<dsv>. Handles 'get' magic, but not 'set' magic. See C<sv_catsv_mg>.
2983 void sv_catsv (SV* dsv, SV* ssv)
2987 Like C<sv_catsv>, but also handles 'set' magic.
2989 void sv_catsv_mg (SV* dsv, SV* ssv)
2993 Efficient removal of characters from the beginning of the string
2994 buffer. SvPOK(sv) must be true and the C<ptr> must be a pointer to
2995 somewhere inside the string buffer. The C<ptr> becomes the first
2996 character of the adjusted string.
2998 void sv_chop(SV* sv, const char *ptr)
3003 Compares the strings in two SVs. Returns -1, 0, or 1 indicating whether the
3004 string in C<sv1> is less than, equal to, or greater than the string in
3007 I32 sv_cmp (SV* sv1, SV* sv2)
3011 Returns the length of the string which is in the SV. See C<SvLEN>.
3017 Set the length of the string which is in the SV. See C<SvCUR>.
3019 void SvCUR_set (SV* sv, int val)
3023 Auto-decrement of the value in the SV.
3025 void sv_dec (SV* sv)
3027 =item sv_derived_from
3029 Returns a boolean indicating whether the SV is derived from the specified
3030 class. This is the function that implements C<UNIVERSAL::isa>. It works
3031 for class names as well as for objects.
3033 bool sv_derived_from (SV* sv, const char* name);
3037 Returns a pointer to the last character in the string which is in the SV.
3038 See C<SvCUR>. Access the character as
3044 Returns a boolean indicating whether the strings in the two SVs are
3047 I32 sv_eq (SV* sv1, SV* sv2)
3051 Invokes C<mg_get> on an SV if it has 'get' magic. This macro evaluates
3052 its argument more than once.
3054 void SvGETMAGIC(SV *sv)
3058 Expands the character buffer in the SV so that it has room for the
3059 indicated number of bytes (remember to reserve space for an extra
3060 trailing NUL character). Calls C<sv_grow> to perform the expansion if
3061 necessary. Returns a pointer to the character buffer.
3063 char* SvGROW(SV* sv, STRLEN len)
3067 Expands the character buffer in the SV. This will use C<sv_unref> and will
3068 upgrade the SV to C<SVt_PV>. Returns a pointer to the character buffer.
3073 Auto-increment of the value in the SV.
3075 void sv_inc (SV* sv)
3079 Inserts a string at the specified offset/length within the SV.
3080 Similar to the Perl substr() function.
3082 void sv_insert(SV *sv, STRLEN offset, STRLEN len,
3083 char *str, STRLEN strlen)
3087 Returns a boolean indicating whether the SV contains an integer.
3093 Unsets the IV status of an SV.
3095 void SvIOK_off (SV* sv)
3099 Tells an SV that it is an integer.
3101 void SvIOK_on (SV* sv)
3105 Tells an SV that it is an integer and disables all other OK bits.
3107 void SvIOK_only (SV* sv)
3111 Returns a boolean indicating whether the SV contains an integer. Checks the
3112 B<private> setting. Use C<SvIOK>.
3118 Returns a boolean indicating whether the SV is blessed into the specified
3119 class. This does not check for subtypes; use C<sv_derived_from> to verify
3120 an inheritance relationship.
3122 int sv_isa (SV* sv, char* name)
3126 Returns a boolean indicating whether the SV is an RV pointing to a blessed
3127 object. If the SV is not an RV, or if the object is not blessed, then this
3130 int sv_isobject (SV* sv)
3134 Coerces the given SV to an integer and returns it.
3140 Returns the integer which is stored in the SV, assuming SvIOK is true.
3146 Returns the size of the string buffer in the SV. See C<SvCUR>.
3152 Returns the length of the string in the SV. Use C<SvCUR>.
3154 STRLEN sv_len (SV* sv)
3158 Adds magic to an SV.
3160 void sv_magic (SV* sv, SV* obj, int how, const char* name, I32 namlen)
3164 Creates a new SV which is a copy of the original SV. The new SV is marked
3167 SV* sv_mortalcopy (SV* oldsv)
3171 Creates a new SV which is mortal. The reference count of the SV is set to 1.
3173 SV* sv_newmortal (void)
3177 Returns a boolean indicating whether the SV contains a number, integer or
3184 Unsets the NV/IV status of an SV.
3186 void SvNIOK_off (SV* sv)
3190 Returns a boolean indicating whether the SV contains a number, integer or
3191 double. Checks the B<private> setting. Use C<SvNIOK>.
3193 int SvNIOKp (SV* SV)
3197 This is the C<false> SV. See C<PL_sv_yes>. Always refer to this as C<&PL_sv_no>.
3201 Returns a boolean indicating whether the SV contains a double.
3207 Unsets the NV status of an SV.
3209 void SvNOK_off (SV* sv)
3213 Tells an SV that it is a double.
3215 void SvNOK_on (SV* sv)
3219 Tells an SV that it is a double and disables all other OK bits.
3221 void SvNOK_only (SV* sv)
3225 Returns a boolean indicating whether the SV contains a double. Checks the
3226 B<private> setting. Use C<SvNOK>.
3232 Coerce the given SV to a double and return it.
3234 double SvNV (SV* sv)
3238 Returns the double which is stored in the SV, assuming SvNOK is true.
3240 double SvNVX (SV* sv)
3244 Returns a boolean indicating whether the value is an SV.
3250 Returns a boolean indicating whether the SvIVX is a valid offset value
3251 for the SvPVX. This hack is used internally to speed up removal of
3252 characters from the beginning of a SvPV. When SvOOK is true, then the
3253 start of the allocated string buffer is really (SvPVX - SvIVX).
3259 Returns a boolean indicating whether the SV contains a character string.
3265 Unsets the PV status of an SV.
3267 void SvPOK_off (SV* sv)
3271 Tells an SV that it is a string.
3273 void SvPOK_on (SV* sv)
3277 Tells an SV that it is a string and disables all other OK bits.
3279 void SvPOK_only (SV* sv)
3283 Returns a boolean indicating whether the SV contains a character string.
3284 Checks the B<private> setting. Use C<SvPOK>.
3290 Returns a pointer to the string in the SV, or a stringified form of the SV
3291 if the SV does not contain a string. Handles 'get' magic.
3293 char* SvPV (SV* sv, STRLEN len)
3297 Like <SvPV> but will force the SV into becoming a string (SvPOK). You
3298 want force if you are going to update the SvPVX directly.
3300 char* SvPV_force(SV* sv, STRLEN len)
3304 Returns a pointer to the string in the SV, or a stringified form of the SV
3305 if the SV does not contain a string. Handles 'get' magic.
3307 char* SvPV_nolen (SV* sv)
3311 Returns a pointer to the string in the SV. The SV must contain a string.
3313 char* SvPVX (SV* sv)
3317 Returns the value of the object's reference count.
3319 int SvREFCNT (SV* sv)
3323 Decrements the reference count of the given SV.
3325 void SvREFCNT_dec (SV* sv)
3329 Increments the reference count of the given SV.
3331 void SvREFCNT_inc (SV* sv)
3335 Tests if the SV is an RV.
3341 Unsets the RV status of an SV.
3343 void SvROK_off (SV* sv)
3347 Tells an SV that it is an RV.
3349 void SvROK_on (SV* sv)
3353 Dereferences an RV to return the SV.
3359 Invokes C<mg_set> on an SV if it has 'set' magic. This macro evaluates
3360 its argument more than once.
3362 void SvSETMAGIC( SV *sv )
3366 Copies an integer into the given SV. Does not handle 'set' magic.
3369 void sv_setiv (SV* sv, IV num)
3373 Like C<sv_setiv>, but also handles 'set' magic.
3375 void sv_setiv_mg (SV* sv, IV num)
3379 Copies a double into the given SV. Does not handle 'set' magic.
3382 void sv_setnv (SV* sv, double num)
3386 Like C<sv_setnv>, but also handles 'set' magic.
3388 void sv_setnv_mg (SV* sv, double num)
3392 Copies a string into an SV. The string must be null-terminated.
3393 Does not handle 'set' magic. See C<sv_setpv_mg>.
3395 void sv_setpv (SV* sv, const char* ptr)
3399 Like C<sv_setpv>, but also handles 'set' magic.
3401 void sv_setpv_mg (SV* sv, const char* ptr)
3405 Copies an integer into the given SV, also updating its string value.
3406 Does not handle 'set' magic. See C<sv_setpviv_mg>.
3408 void sv_setpviv (SV* sv, IV num)
3412 Like C<sv_setpviv>, but also handles 'set' magic.
3414 void sv_setpviv_mg (SV* sv, IV num)
3418 Copies a string into an SV. The C<len> parameter indicates the number of
3419 bytes to be copied. Does not handle 'set' magic. See C<sv_setpvn_mg>.
3421 void sv_setpvn (SV* sv, const char* ptr, STRLEN len)
3425 Like C<sv_setpvn>, but also handles 'set' magic.
3427 void sv_setpvn_mg (SV* sv, const char* ptr, STRLEN len)
3431 Processes its arguments like C<sprintf> and sets an SV to the formatted
3432 output. Does not handle 'set' magic. See C<sv_setpvf_mg>.
3434 void sv_setpvf (SV* sv, const char* pat, ...)
3438 Like C<sv_setpvf>, but also handles 'set' magic.
3440 void sv_setpvf_mg (SV* sv, const char* pat, ...)
3444 Copies an integer into a new SV, optionally blessing the SV. The C<rv>
3445 argument will be upgraded to an RV. That RV will be modified to point to
3446 the new SV. The C<classname> argument indicates the package for the
3447 blessing. Set C<classname> to C<Nullch> to avoid the blessing. The new SV
3448 will be returned and will have a reference count of 1.
3450 SV* sv_setref_iv (SV *rv, char *classname, IV iv)
3454 Copies a double into a new SV, optionally blessing the SV. The C<rv>
3455 argument will be upgraded to an RV. That RV will be modified to point to
3456 the new SV. The C<classname> argument indicates the package for the
3457 blessing. Set C<classname> to C<Nullch> to avoid the blessing. The new SV
3458 will be returned and will have a reference count of 1.
3460 SV* sv_setref_nv (SV *rv, char *classname, double nv)
3464 Copies a pointer into a new SV, optionally blessing the SV. The C<rv>
3465 argument will be upgraded to an RV. That RV will be modified to point to
3466 the new SV. If the C<pv> argument is NULL then C<PL_sv_undef> will be placed
3467 into the SV. The C<classname> argument indicates the package for the
3468 blessing. Set C<classname> to C<Nullch> to avoid the blessing. The new SV
3469 will be returned and will have a reference count of 1.
3471 SV* sv_setref_pv (SV *rv, char *classname, void* pv)
3473 Do not use with integral Perl types such as HV, AV, SV, CV, because those
3474 objects will become corrupted by the pointer copy process.
3476 Note that C<sv_setref_pvn> copies the string while this copies the pointer.
3480 Copies a string into a new SV, optionally blessing the SV. The length of the
3481 string must be specified with C<n>. The C<rv> argument will be upgraded to
3482 an RV. That RV will be modified to point to the new SV. The C<classname>
3483 argument indicates the package for the blessing. Set C<classname> to
3484 C<Nullch> to avoid the blessing. The new SV will be returned and will have
3485 a reference count of 1.
3487 SV* sv_setref_pvn (SV *rv, char *classname, char* pv, I32 n)
3489 Note that C<sv_setref_pv> copies the pointer while this copies the string.
3493 Calls C<sv_setsv> if dsv is not the same as ssv. May evaluate arguments
3496 void SvSetSV (SV* dsv, SV* ssv)
3498 =item SvSetSV_nosteal
3500 Calls a non-destructive version of C<sv_setsv> if dsv is not the same as ssv.
3501 May evaluate arguments more than once.
3503 void SvSetSV_nosteal (SV* dsv, SV* ssv)
3507 Copies the contents of the source SV C<ssv> into the destination SV C<dsv>.
3508 The source SV may be destroyed if it is mortal. Does not handle 'set' magic.
3509 See the macro forms C<SvSetSV>, C<SvSetSV_nosteal> and C<sv_setsv_mg>.
3511 void sv_setsv (SV* dsv, SV* ssv)
3515 Like C<sv_setsv>, but also handles 'set' magic.
3517 void sv_setsv_mg (SV* dsv, SV* ssv)
3521 Copies an unsigned integer into the given SV. Does not handle 'set' magic.
3524 void sv_setuv (SV* sv, UV num)
3528 Like C<sv_setuv>, but also handles 'set' magic.
3530 void sv_setuv_mg (SV* sv, UV num)
3534 Returns the stash of the SV.
3536 HV* SvSTASH (SV* sv)
3540 Taints an SV if tainting is enabled
3542 void SvTAINT (SV* sv)
3546 Checks to see if an SV is tainted. Returns TRUE if it is, FALSE if not.
3548 int SvTAINTED (SV* sv)
3552 Untaints an SV. Be I<very> careful with this routine, as it short-circuits
3553 some of Perl's fundamental security features. XS module authors should
3554 not use this function unless they fully understand all the implications
3555 of unconditionally untainting the value. Untainting should be done in
3556 the standard perl fashion, via a carefully crafted regexp, rather than
3557 directly untainting variables.
3559 void SvTAINTED_off (SV* sv)
3563 Marks an SV as tainted.
3565 void SvTAINTED_on (SV* sv)
3569 Integer type flag for scalars. See C<svtype>.
3573 Pointer type flag for scalars. See C<svtype>.
3577 Type flag for arrays. See C<svtype>.
3581 Type flag for code refs. See C<svtype>.
3585 Type flag for hashes. See C<svtype>.
3589 Type flag for blessed scalars. See C<svtype>.
3593 Double type flag for scalars. See C<svtype>.
3597 Returns a boolean indicating whether Perl would evaluate the SV as true or
3598 false, defined or undefined. Does not handle 'get' magic.
3604 Returns the type of the SV. See C<svtype>.
3606 svtype SvTYPE (SV* sv)
3610 An enum of flags for Perl types. These are found in the file B<sv.h> in the
3611 C<svtype> enum. Test these flags with the C<SvTYPE> macro.
3615 This is the C<undef> SV. Always refer to this as C<&PL_sv_undef>.
3619 Unsets the RV status of the SV, and decrements the reference count of
3620 whatever was being referenced by the RV. This can almost be thought of
3621 as a reversal of C<newSVrv>. See C<SvROK_off>.
3623 void sv_unref (SV* sv)
3627 Used to upgrade an SV to a more complex form. Uses C<sv_upgrade> to perform
3628 the upgrade if necessary. See C<svtype>.
3630 bool SvUPGRADE (SV* sv, svtype mt)
3634 Upgrade an SV to a more complex form. Use C<SvUPGRADE>. See C<svtype>.
3638 Tells an SV to use C<ptr> to find its string value. Normally the string is
3639 stored inside the SV but sv_usepvn allows the SV to use an outside string.
3640 The C<ptr> should point to memory that was allocated by C<malloc>. The
3641 string length, C<len>, must be supplied. This function will realloc the
3642 memory pointed to by C<ptr>, so that pointer should not be freed or used by
3643 the programmer after giving it to sv_usepvn. Does not handle 'set' magic.
3644 See C<sv_usepvn_mg>.
3646 void sv_usepvn (SV* sv, char* ptr, STRLEN len)
3650 Like C<sv_usepvn>, but also handles 'set' magic.
3652 void sv_usepvn_mg (SV* sv, char* ptr, STRLEN len)
3656 Processes its arguments like C<vsprintf> and appends the formatted output
3657 to an SV. Uses an array of SVs if the C style variable argument list is
3658 missing (NULL). When running with taint checks enabled, indicates via
3659 C<maybe_tainted> if results are untrustworthy (often due to the use of
3662 void sv_catpvfn (SV* sv, const char* pat, STRLEN patlen,
3663 va_list *args, SV **svargs, I32 svmax,
3664 bool *maybe_tainted);
3668 Works like C<vcatpvfn> but copies the text into the SV instead of
3671 void sv_setpvfn (SV* sv, const char* pat, STRLEN patlen,
3672 va_list *args, SV **svargs, I32 svmax,
3673 bool *maybe_tainted);
3677 Coerces the given SV to an unsigned integer and returns it.
3683 Returns the unsigned integer which is stored in the SV, assuming SvIOK is true.
3689 This is the C<true> SV. See C<PL_sv_no>. Always refer to this as C<&PL_sv_yes>.
3693 Variable which is setup by C<xsubpp> to designate the object in a C++ XSUB.
3694 This is always the proper type for the C++ object. See C<CLASS> and
3695 L<perlxs/"Using XS With C++">.
3699 Converts the specified character to lowercase.
3701 int toLOWER (char c)
3705 Converts the specified character to uppercase.
3707 int toUPPER (char c)
3711 This is the XSUB-writer's interface to Perl's C<warn> function. Use this
3712 function the same way you use the C C<printf> function. See C<croak()>.
3716 Push an integer onto the stack, extending the stack if necessary. Handles
3717 'set' magic. See C<PUSHi>.
3723 Push a double onto the stack, extending the stack if necessary. Handles 'set'
3724 magic. See C<PUSHn>.
3730 Push a string onto the stack, extending the stack if necessary. The C<len>
3731 indicates the length of the string. Handles 'set' magic. See C<PUSHp>.
3733 XPUSHp(char *c, int len)
3737 Push an SV onto the stack, extending the stack if necessary. Does not
3738 handle 'set' magic. See C<PUSHs>.
3744 Push an unsigned integer onto the stack, extending the stack if
3745 necessary. See C<PUSHu>.
3749 Macro to declare an XSUB and its C parameter list. This is handled by
3754 Return from XSUB, indicating number of items on the stack. This is usually
3755 handled by C<xsubpp>.
3759 =item XSRETURN_EMPTY
3761 Return an empty list from an XSUB immediately.
3767 Return an integer from an XSUB immediately. Uses C<XST_mIV>.
3773 Return C<&PL_sv_no> from an XSUB immediately. Uses C<XST_mNO>.
3779 Return an double from an XSUB immediately. Uses C<XST_mNV>.
3785 Return a copy of a string from an XSUB immediately. Uses C<XST_mPV>.
3787 XSRETURN_PV(char *v)
3789 =item XSRETURN_UNDEF
3791 Return C<&PL_sv_undef> from an XSUB immediately. Uses C<XST_mUNDEF>.
3797 Return C<&PL_sv_yes> from an XSUB immediately. Uses C<XST_mYES>.
3803 Place an integer into the specified position C<i> on the stack. The value is
3804 stored in a new mortal SV.
3806 XST_mIV( int i, IV v )
3810 Place a double into the specified position C<i> on the stack. The value is
3811 stored in a new mortal SV.
3813 XST_mNV( int i, NV v )
3817 Place C<&PL_sv_no> into the specified position C<i> on the stack.
3823 Place a copy of a string into the specified position C<i> on the stack. The
3824 value is stored in a new mortal SV.
3826 XST_mPV( int i, char *v )
3830 Place C<&PL_sv_undef> into the specified position C<i> on the stack.
3836 Place C<&PL_sv_yes> into the specified position C<i> on the stack.
3842 The version identifier for an XS module. This is usually handled
3843 automatically by C<ExtUtils::MakeMaker>. See C<XS_VERSION_BOOTCHECK>.
3845 =item XS_VERSION_BOOTCHECK
3847 Macro to verify that a PM module's $VERSION variable matches the XS module's
3848 C<XS_VERSION> variable. This is usually handled automatically by
3849 C<xsubpp>. See L<perlxs/"The VERSIONCHECK: Keyword">.
3853 The XSUB-writer's interface to the C C<memzero> function. The C<d> is the
3854 destination, C<n> is the number of items, and C<t> is the type.
3856 void Zero( d, n, t )
3862 Until May 1997, this document was maintained by Jeff Okamoto
3863 <okamoto@corp.hp.com>. It is now maintained as part of Perl itself.
3865 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
3866 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
3867 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
3868 Stephen McCamant, and Gurusamy Sarathy.
3870 API Listing originally by Dean Roehrich <roehrich@cray.com>.