3 perlguts - Introduction to the Perl API
7 This document attempts to describe how to use the Perl API, as well as
8 to provide some info on the basic workings of the Perl core. It is far
9 from complete and probably contains many errors. Please refer any
10 questions or comments to the author below.
16 Perl has three typedefs that handle Perl's three main data types:
22 Each typedef has specific routines that manipulate the various data types.
24 =head2 What is an "IV"?
26 Perl uses a special typedef IV which is a simple signed integer type that is
27 guaranteed to be large enough to hold a pointer (as well as an integer).
28 Additionally, there is the UV, which is simply an unsigned IV.
30 Perl also uses two special typedefs, I32 and I16, which will always be at
31 least 32-bits and 16-bits long, respectively. (Again, there are U32 and U16,
32 as well.) They will usually be exactly 32 and 16 bits long, but on Crays
33 they will both be 64 bits.
35 =head2 Working with SVs
37 An SV can be created and loaded with one command. There are five types of
38 values that can be loaded: an integer value (IV), an unsigned integer
39 value (UV), a double (NV), a string (PV), and another scalar (SV).
41 The seven routines are:
46 SV* newSVpv(const char*, STRLEN);
47 SV* newSVpvn(const char*, STRLEN);
48 SV* newSVpvf(const char*, ...);
51 C<STRLEN> is an integer type (Size_t, usually defined as size_t in
52 F<config.h>) guaranteed to be large enough to represent the size of
53 any string that perl can handle.
55 In the unlikely case of a SV requiring more complex initialisation, you
56 can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
57 type NULL is returned, else an SV of type PV is returned with len + 1 (for
58 the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
59 the SV has value undef.
61 SV *sv = newSV(0); /* no storage allocated */
62 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage allocated */
64 To change the value of an I<already-existing> SV, there are eight routines:
66 void sv_setiv(SV*, IV);
67 void sv_setuv(SV*, UV);
68 void sv_setnv(SV*, double);
69 void sv_setpv(SV*, const char*);
70 void sv_setpvn(SV*, const char*, STRLEN)
71 void sv_setpvf(SV*, const char*, ...);
72 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool *);
73 void sv_setsv(SV*, SV*);
75 Notice that you can choose to specify the length of the string to be
76 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
77 allow Perl to calculate the length by using C<sv_setpv> or by specifying
78 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
79 determine the string's length by using C<strlen>, which depends on the
80 string terminating with a NUL character.
82 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
83 formatted output becomes the value.
85 C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
86 either a pointer to a variable argument list or the address and length of
87 an array of SVs. The last argument points to a boolean; on return, if that
88 boolean is true, then locale-specific information has been used to format
89 the string, and the string's contents are therefore untrustworthy (see
90 L<perlsec>). This pointer may be NULL if that information is not
91 important. Note that this function requires you to specify the length of
94 The C<sv_set*()> functions are not generic enough to operate on values
95 that have "magic". See L<Magic Virtual Tables> later in this document.
97 All SVs that contain strings should be terminated with a NUL character.
98 If it is not NUL-terminated there is a risk of
99 core dumps and corruptions from code which passes the string to C
100 functions or system calls which expect a NUL-terminated string.
101 Perl's own functions typically add a trailing NUL for this reason.
102 Nevertheless, you should be very careful when you pass a string stored
103 in an SV to a C function or system call.
105 To access the actual value that an SV points to, you can use the macros:
110 SvPV(SV*, STRLEN len)
113 which will automatically coerce the actual scalar type into an IV, UV, double,
116 In the C<SvPV> macro, the length of the string returned is placed into the
117 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
118 not care what the length of the data is, use the C<SvPV_nolen> macro.
119 Historically the C<SvPV> macro with the global variable C<PL_na> has been
120 used in this case. But that can be quite inefficient because C<PL_na> must
121 be accessed in thread-local storage in threaded Perl. In any case, remember
122 that Perl allows arbitrary strings of data that may both contain NULs and
123 might not be terminated by a NUL.
125 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
126 len);>. It might work with your compiler, but it won't work for everyone.
127 Break this sort of statement up into separate assignments:
135 If you want to know if the scalar value is TRUE, you can use:
139 Although Perl will automatically grow strings for you, if you need to force
140 Perl to allocate more memory for your SV, you can use the macro
142 SvGROW(SV*, STRLEN newlen)
144 which will determine if more memory needs to be allocated. If so, it will
145 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
146 decrease, the allocated memory of an SV and that it does not automatically
147 add a byte for the a trailing NUL (perl's own string functions typically do
148 C<SvGROW(sv, len + 1)>).
150 If you have an SV and want to know what kind of data Perl thinks is stored
151 in it, you can use the following macros to check the type of SV you have.
157 You can get and set the current length of the string stored in an SV with
158 the following macros:
161 SvCUR_set(SV*, I32 val)
163 You can also get a pointer to the end of the string stored in the SV
168 But note that these last three macros are valid only if C<SvPOK()> is true.
170 If you want to append something to the end of string stored in an C<SV*>,
171 you can use the following functions:
173 void sv_catpv(SV*, const char*);
174 void sv_catpvn(SV*, const char*, STRLEN);
175 void sv_catpvf(SV*, const char*, ...);
176 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
177 void sv_catsv(SV*, SV*);
179 The first function calculates the length of the string to be appended by
180 using C<strlen>. In the second, you specify the length of the string
181 yourself. The third function processes its arguments like C<sprintf> and
182 appends the formatted output. The fourth function works like C<vsprintf>.
183 You can specify the address and length of an array of SVs instead of the
184 va_list argument. The fifth function extends the string stored in the first
185 SV with the string stored in the second SV. It also forces the second SV
186 to be interpreted as a string.
188 The C<sv_cat*()> functions are not generic enough to operate on values that
189 have "magic". See L<Magic Virtual Tables> later in this document.
191 If you know the name of a scalar variable, you can get a pointer to its SV
192 by using the following:
194 SV* get_sv("package::varname", FALSE);
196 This returns NULL if the variable does not exist.
198 If you want to know if this variable (or any other SV) is actually C<defined>,
203 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
204 Its address can be used whenever an C<SV*> is needed.
206 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
207 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
208 addresses can be used whenever an C<SV*> is needed.
210 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
214 if (I-am-to-return-a-real-value) {
215 sv = sv_2mortal(newSViv(42));
219 This code tries to return a new SV (which contains the value 42) if it should
220 return a real value, or undef otherwise. Instead it has returned a NULL
221 pointer which, somewhere down the line, will cause a segmentation violation,
222 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
223 first line and all will be well.
225 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
226 call is not necessary (see L<Reference Counts and Mortality>).
230 Perl provides the function C<sv_chop> to efficiently remove characters
231 from the beginning of a string; you give it an SV and a pointer to
232 somewhere inside the PV, and it discards everything before the
233 pointer. The efficiency comes by means of a little hack: instead of
234 actually removing the characters, C<sv_chop> sets the flag C<OOK>
235 (offset OK) to signal to other functions that the offset hack is in
236 effect, and it puts the number of bytes chopped off into the IV field
237 of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
238 many bytes, and adjusts C<SvCUR> and C<SvLEN>.
240 Hence, at this point, the start of the buffer that we allocated lives
241 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
242 into the middle of this allocated storage.
244 This is best demonstrated by example:
246 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
247 SV = PVIV(0x8128450) at 0x81340f0
249 FLAGS = (POK,OOK,pPOK)
251 PV = 0x8135781 ( "1" . ) "2345"\0
255 Here the number of bytes chopped off (1) is put into IV, and
256 C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
257 portion of the string between the "real" and the "fake" beginnings is
258 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
259 the fake beginning, not the real one.
261 Something similar to the offset hack is performed on AVs to enable
262 efficient shifting and splicing off the beginning of the array; while
263 C<AvARRAY> points to the first element in the array that is visible from
264 Perl, C<AvALLOC> points to the real start of the C array. These are
265 usually the same, but a C<shift> operation can be carried out by
266 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvLEN>.
267 Again, the location of the real start of the C array only comes into
268 play when freeing the array. See C<av_shift> in F<av.c>.
270 =head2 What's Really Stored in an SV?
272 Recall that the usual method of determining the type of scalar you have is
273 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
274 usually these macros will always return TRUE and calling the C<Sv*V>
275 macros will do the appropriate conversion of string to integer/double or
276 integer/double to string.
278 If you I<really> need to know if you have an integer, double, or string
279 pointer in an SV, you can use the following three macros instead:
285 These will tell you if you truly have an integer, double, or string pointer
286 stored in your SV. The "p" stands for private.
288 The are various ways in which the private and public flags may differ.
289 For example, a tied SV may have a valid underlying value in the IV slot
290 (so SvIOKp is true), but the data should be accessed via the FETCH
291 routine rather than directly, so SvIOK is false. Another is when
292 numeric conversion has occured and precision has been lost: only the
293 private flag is set on 'lossy' values. So when an NV is converted to an
294 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
296 In general, though, it's best to use the C<Sv*V> macros.
298 =head2 Working with AVs
300 There are two ways to create and load an AV. The first method creates an
305 The second method both creates the AV and initially populates it with SVs:
307 AV* av_make(I32 num, SV **ptr);
309 The second argument points to an array containing C<num> C<SV*>'s. Once the
310 AV has been created, the SVs can be destroyed, if so desired.
312 Once the AV has been created, the following operations are possible on AVs:
314 void av_push(AV*, SV*);
317 void av_unshift(AV*, I32 num);
319 These should be familiar operations, with the exception of C<av_unshift>.
320 This routine adds C<num> elements at the front of the array with the C<undef>
321 value. You must then use C<av_store> (described below) to assign values
322 to these new elements.
324 Here are some other functions:
327 SV** av_fetch(AV*, I32 key, I32 lval);
328 SV** av_store(AV*, I32 key, SV* val);
330 The C<av_len> function returns the highest index value in array (just
331 like $#array in Perl). If the array is empty, -1 is returned. The
332 C<av_fetch> function returns the value at index C<key>, but if C<lval>
333 is non-zero, then C<av_fetch> will store an undef value at that index.
334 The C<av_store> function stores the value C<val> at index C<key>, and does
335 not increment the reference count of C<val>. Thus the caller is responsible
336 for taking care of that, and if C<av_store> returns NULL, the caller will
337 have to decrement the reference count to avoid a memory leak. Note that
338 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
343 void av_extend(AV*, I32 key);
345 The C<av_clear> function deletes all the elements in the AV* array, but
346 does not actually delete the array itself. The C<av_undef> function will
347 delete all the elements in the array plus the array itself. The
348 C<av_extend> function extends the array so that it contains at least C<key+1>
349 elements. If C<key+1> is less than the currently allocated length of the array,
350 then nothing is done.
352 If you know the name of an array variable, you can get a pointer to its AV
353 by using the following:
355 AV* get_av("package::varname", FALSE);
357 This returns NULL if the variable does not exist.
359 See L<Understanding the Magic of Tied Hashes and Arrays> for more
360 information on how to use the array access functions on tied arrays.
362 =head2 Working with HVs
364 To create an HV, you use the following routine:
368 Once the HV has been created, the following operations are possible on HVs:
370 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
371 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
373 The C<klen> parameter is the length of the key being passed in (Note that
374 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
375 length of the key). The C<val> argument contains the SV pointer to the
376 scalar being stored, and C<hash> is the precomputed hash value (zero if
377 you want C<hv_store> to calculate it for you). The C<lval> parameter
378 indicates whether this fetch is actually a part of a store operation, in
379 which case a new undefined value will be added to the HV with the supplied
380 key and C<hv_fetch> will return as if the value had already existed.
382 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
383 C<SV*>. To access the scalar value, you must first dereference the return
384 value. However, you should check to make sure that the return value is
385 not NULL before dereferencing it.
387 These two functions check if a hash table entry exists, and deletes it.
389 bool hv_exists(HV*, const char* key, U32 klen);
390 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
392 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
393 create and return a mortal copy of the deleted value.
395 And more miscellaneous functions:
400 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
401 table but does not actually delete the hash table. The C<hv_undef> deletes
402 both the entries and the hash table itself.
404 Perl keeps the actual data in linked list of structures with a typedef of HE.
405 These contain the actual key and value pointers (plus extra administrative
406 overhead). The key is a string pointer; the value is an C<SV*>. However,
407 once you have an C<HE*>, to get the actual key and value, use the routines
410 I32 hv_iterinit(HV*);
411 /* Prepares starting point to traverse hash table */
412 HE* hv_iternext(HV*);
413 /* Get the next entry, and return a pointer to a
414 structure that has both the key and value */
415 char* hv_iterkey(HE* entry, I32* retlen);
416 /* Get the key from an HE structure and also return
417 the length of the key string */
418 SV* hv_iterval(HV*, HE* entry);
419 /* Return an SV pointer to the value of the HE
421 SV* hv_iternextsv(HV*, char** key, I32* retlen);
422 /* This convenience routine combines hv_iternext,
423 hv_iterkey, and hv_iterval. The key and retlen
424 arguments are return values for the key and its
425 length. The value is returned in the SV* argument */
427 If you know the name of a hash variable, you can get a pointer to its HV
428 by using the following:
430 HV* get_hv("package::varname", FALSE);
432 This returns NULL if the variable does not exist.
434 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
438 hash = (hash * 33) + *key++;
439 hash = hash + (hash >> 5); /* after 5.6 */
441 The last step was added in version 5.6 to improve distribution of
442 lower bits in the resulting hash value.
444 See L<Understanding the Magic of Tied Hashes and Arrays> for more
445 information on how to use the hash access functions on tied hashes.
447 =head2 Hash API Extensions
449 Beginning with version 5.004, the following functions are also supported:
451 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
452 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
454 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
455 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
457 SV* hv_iterkeysv (HE* entry);
459 Note that these functions take C<SV*> keys, which simplifies writing
460 of extension code that deals with hash structures. These functions
461 also allow passing of C<SV*> keys to C<tie> functions without forcing
462 you to stringify the keys (unlike the previous set of functions).
464 They also return and accept whole hash entries (C<HE*>), making their
465 use more efficient (since the hash number for a particular string
466 doesn't have to be recomputed every time). See L<perlapi> for detailed
469 The following macros must always be used to access the contents of hash
470 entries. Note that the arguments to these macros must be simple
471 variables, since they may get evaluated more than once. See
472 L<perlapi> for detailed descriptions of these macros.
474 HePV(HE* he, STRLEN len)
478 HeSVKEY_force(HE* he)
479 HeSVKEY_set(HE* he, SV* sv)
481 These two lower level macros are defined, but must only be used when
482 dealing with keys that are not C<SV*>s:
487 Note that both C<hv_store> and C<hv_store_ent> do not increment the
488 reference count of the stored C<val>, which is the caller's responsibility.
489 If these functions return a NULL value, the caller will usually have to
490 decrement the reference count of C<val> to avoid a memory leak.
494 References are a special type of scalar that point to other data types
495 (including references).
497 To create a reference, use either of the following functions:
499 SV* newRV_inc((SV*) thing);
500 SV* newRV_noinc((SV*) thing);
502 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
503 functions are identical except that C<newRV_inc> increments the reference
504 count of the C<thing>, while C<newRV_noinc> does not. For historical
505 reasons, C<newRV> is a synonym for C<newRV_inc>.
507 Once you have a reference, you can use the following macro to dereference
512 then call the appropriate routines, casting the returned C<SV*> to either an
513 C<AV*> or C<HV*>, if required.
515 To determine if an SV is a reference, you can use the following macro:
519 To discover what type of value the reference refers to, use the following
520 macro and then check the return value.
524 The most useful types that will be returned are:
533 SVt_PVGV Glob (possible a file handle)
534 SVt_PVMG Blessed or Magical Scalar
536 See the sv.h header file for more details.
538 =head2 Blessed References and Class Objects
540 References are also used to support object-oriented programming. In perl's
541 OO lexicon, an object is simply a reference that has been blessed into a
542 package (or class). Once blessed, the programmer may now use the reference
543 to access the various methods in the class.
545 A reference can be blessed into a package with the following function:
547 SV* sv_bless(SV* sv, HV* stash);
549 The C<sv> argument must be a reference value. The C<stash> argument
550 specifies which class the reference will belong to. See
551 L<Stashes and Globs> for information on converting class names into stashes.
553 /* Still under construction */
555 Upgrades rv to reference if not already one. Creates new SV for rv to
556 point to. If C<classname> is non-null, the SV is blessed into the specified
557 class. SV is returned.
559 SV* newSVrv(SV* rv, const char* classname);
561 Copies integer, unsigned integer or double into an SV whose reference is C<rv>. SV is blessed
562 if C<classname> is non-null.
564 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
565 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
566 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
568 Copies the pointer value (I<the address, not the string!>) into an SV whose
569 reference is rv. SV is blessed if C<classname> is non-null.
571 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
573 Copies string into an SV whose reference is C<rv>. Set length to 0 to let
574 Perl calculate the string length. SV is blessed if C<classname> is non-null.
576 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
578 Tests whether the SV is blessed into the specified class. It does not
579 check inheritance relationships.
581 int sv_isa(SV* sv, const char* name);
583 Tests whether the SV is a reference to a blessed object.
585 int sv_isobject(SV* sv);
587 Tests whether the SV is derived from the specified class. SV can be either
588 a reference to a blessed object or a string containing a class name. This
589 is the function implementing the C<UNIVERSAL::isa> functionality.
591 bool sv_derived_from(SV* sv, const char* name);
593 To check if you've got an object derived from a specific class you have
596 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
598 =head2 Creating New Variables
600 To create a new Perl variable with an undef value which can be accessed from
601 your Perl script, use the following routines, depending on the variable type.
603 SV* get_sv("package::varname", TRUE);
604 AV* get_av("package::varname", TRUE);
605 HV* get_hv("package::varname", TRUE);
607 Notice the use of TRUE as the second parameter. The new variable can now
608 be set, using the routines appropriate to the data type.
610 There are additional macros whose values may be bitwise OR'ed with the
611 C<TRUE> argument to enable certain extra features. Those bits are:
617 Marks the variable as multiply defined, thus preventing the:
619 Name <varname> used only once: possible typo
627 Had to create <varname> unexpectedly
629 if the variable did not exist before the function was called.
633 If you do not specify a package name, the variable is created in the current
636 =head2 Reference Counts and Mortality
638 Perl uses a reference count-driven garbage collection mechanism. SVs,
639 AVs, or HVs (xV for short in the following) start their life with a
640 reference count of 1. If the reference count of an xV ever drops to 0,
641 then it will be destroyed and its memory made available for reuse.
643 This normally doesn't happen at the Perl level unless a variable is
644 undef'ed or the last variable holding a reference to it is changed or
645 overwritten. At the internal level, however, reference counts can be
646 manipulated with the following macros:
648 int SvREFCNT(SV* sv);
649 SV* SvREFCNT_inc(SV* sv);
650 void SvREFCNT_dec(SV* sv);
652 However, there is one other function which manipulates the reference
653 count of its argument. The C<newRV_inc> function, you will recall,
654 creates a reference to the specified argument. As a side effect,
655 it increments the argument's reference count. If this is not what
656 you want, use C<newRV_noinc> instead.
658 For example, imagine you want to return a reference from an XSUB function.
659 Inside the XSUB routine, you create an SV which initially has a reference
660 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
661 This returns the reference as a new SV, but the reference count of the
662 SV you passed to C<newRV_inc> has been incremented to two. Now you
663 return the reference from the XSUB routine and forget about the SV.
664 But Perl hasn't! Whenever the returned reference is destroyed, the
665 reference count of the original SV is decreased to one and nothing happens.
666 The SV will hang around without any way to access it until Perl itself
667 terminates. This is a memory leak.
669 The correct procedure, then, is to use C<newRV_noinc> instead of
670 C<newRV_inc>. Then, if and when the last reference is destroyed,
671 the reference count of the SV will go to zero and it will be destroyed,
672 stopping any memory leak.
674 There are some convenience functions available that can help with the
675 destruction of xVs. These functions introduce the concept of "mortality".
676 An xV that is mortal has had its reference count marked to be decremented,
677 but not actually decremented, until "a short time later". Generally the
678 term "short time later" means a single Perl statement, such as a call to
679 an XSUB function. The actual determinant for when mortal xVs have their
680 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
681 See L<perlcall> and L<perlxs> for more details on these macros.
683 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
684 However, if you mortalize a variable twice, the reference count will
685 later be decremented twice.
687 "Mortal" SVs are mainly used for SVs that are placed on perl's stack.
688 For example an SV which is created just to pass a number to a called sub
689 is made mortal to have it cleaned up automatically when it's popped off
690 the stack. Similarly, results returned by XSUBs (which are pushed on the
691 stack) are often made mortal.
693 To create a mortal variable, use the functions:
697 SV* sv_mortalcopy(SV*)
699 The first call creates a mortal SV (with no value), the second converts an existing
700 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
701 third creates a mortal copy of an existing SV.
702 Because C<sv_newmortal> gives the new SV no value,it must normally be given one
703 via C<sv_setpv>, C<sv_setiv>, etc. :
705 SV *tmp = sv_newmortal();
706 sv_setiv(tmp, an_integer);
708 As that is multiple C statements it is quite common so see this idiom instead:
710 SV *tmp = sv_2mortal(newSViv(an_integer));
713 You should be careful about creating mortal variables. Strange things
714 can happen if you make the same value mortal within multiple contexts,
715 or if you make a variable mortal multiple times. Thinking of "Mortalization"
716 as deferred C<SvREFCNT_dec> should help to minimize such problems.
717 For example if you are passing an SV which you I<know> has high enough REFCNT
718 to survive its use on the stack you need not do any mortalization.
719 If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
720 making a C<sv_mortalcopy> is safer.
722 The mortal routines are not just for SVs -- AVs and HVs can be
723 made mortal by passing their address (type-casted to C<SV*>) to the
724 C<sv_2mortal> or C<sv_mortalcopy> routines.
726 =head2 Stashes and Globs
728 A B<stash> is a hash that contains all variables that are defined
729 within a package. Each key of the stash is a symbol
730 name (shared by all the different types of objects that have the same
731 name), and each value in the hash table is a GV (Glob Value). This GV
732 in turn contains references to the various objects of that name,
733 including (but not limited to) the following:
742 There is a single stash called C<PL_defstash> that holds the items that exist
743 in the C<main> package. To get at the items in other packages, append the
744 string "::" to the package name. The items in the C<Foo> package are in
745 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
746 in the stash C<Baz::> in C<Bar::>'s stash.
748 To get the stash pointer for a particular package, use the function:
750 HV* gv_stashpv(const char* name, I32 create)
751 HV* gv_stashsv(SV*, I32 create)
753 The first function takes a literal string, the second uses the string stored
754 in the SV. Remember that a stash is just a hash table, so you get back an
755 C<HV*>. The C<create> flag will create a new package if it is set.
757 The name that C<gv_stash*v> wants is the name of the package whose symbol table
758 you want. The default package is called C<main>. If you have multiply nested
759 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
762 Alternately, if you have an SV that is a blessed reference, you can find
763 out the stash pointer by using:
765 HV* SvSTASH(SvRV(SV*));
767 then use the following to get the package name itself:
769 char* HvNAME(HV* stash);
771 If you need to bless or re-bless an object you can use the following
774 SV* sv_bless(SV*, HV* stash)
776 where the first argument, an C<SV*>, must be a reference, and the second
777 argument is a stash. The returned C<SV*> can now be used in the same way
780 For more information on references and blessings, consult L<perlref>.
782 =head2 Double-Typed SVs
784 Scalar variables normally contain only one type of value, an integer,
785 double, pointer, or reference. Perl will automatically convert the
786 actual scalar data from the stored type into the requested type.
788 Some scalar variables contain more than one type of scalar data. For
789 example, the variable C<$!> contains either the numeric value of C<errno>
790 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
792 To force multiple data values into an SV, you must do two things: use the
793 C<sv_set*v> routines to add the additional scalar type, then set a flag
794 so that Perl will believe it contains more than one type of data. The
795 four macros to set the flags are:
802 The particular macro you must use depends on which C<sv_set*v> routine
803 you called first. This is because every C<sv_set*v> routine turns on
804 only the bit for the particular type of data being set, and turns off
807 For example, to create a new Perl variable called "dberror" that contains
808 both the numeric and descriptive string error values, you could use the
812 extern char *dberror_list;
814 SV* sv = get_sv("dberror", TRUE);
815 sv_setiv(sv, (IV) dberror);
816 sv_setpv(sv, dberror_list[dberror]);
819 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
820 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
822 =head2 Magic Variables
824 [This section still under construction. Ignore everything here. Post no
825 bills. Everything not permitted is forbidden.]
827 Any SV may be magical, that is, it has special features that a normal
828 SV does not have. These features are stored in the SV structure in a
829 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
842 Note this is current as of patchlevel 0, and could change at any time.
844 =head2 Assigning Magic
846 Perl adds magic to an SV using the sv_magic function:
848 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
850 The C<sv> argument is a pointer to the SV that is to acquire a new magical
853 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
854 convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
855 to the beginning of the linked list of magical features. Any prior entry
856 of the same type of magic is deleted. Note that this can be overridden,
857 and multiple instances of the same type of magic can be associated with an
860 The C<name> and C<namlen> arguments are used to associate a string with
861 the magic, typically the name of a variable. C<namlen> is stored in the
862 C<mg_len> field and if C<name> is non-null and C<namlen> E<gt>= 0 a malloc'd
863 copy of the name is stored in C<mg_ptr> field.
865 The sv_magic function uses C<how> to determine which, if any, predefined
866 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
867 See the L<Magic Virtual Tables> section below. The C<how> argument is also
868 stored in the C<mg_type> field. The value of C<how> should be chosen
869 from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
870 these macros were added, Perl internals used to directly use character
871 literals, so you may occasionally come across old code or documentation
872 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
874 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
875 structure. If it is not the same as the C<sv> argument, the reference
876 count of the C<obj> object is incremented. If it is the same, or if
877 the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
878 then C<obj> is merely stored, without the reference count being incremented.
880 There is also a function to add magic to an C<HV>:
882 void hv_magic(HV *hv, GV *gv, int how);
884 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
886 To remove the magic from an SV, call the function sv_unmagic:
888 void sv_unmagic(SV *sv, int type);
890 The C<type> argument should be equal to the C<how> value when the C<SV>
891 was initially made magical.
893 =head2 Magic Virtual Tables
895 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
896 C<MGVTBL>, which is a structure of function pointers and stands for
897 "Magic Virtual Table" to handle the various operations that might be
898 applied to that variable.
900 The C<MGVTBL> has five pointers to the following routine types:
902 int (*svt_get)(SV* sv, MAGIC* mg);
903 int (*svt_set)(SV* sv, MAGIC* mg);
904 U32 (*svt_len)(SV* sv, MAGIC* mg);
905 int (*svt_clear)(SV* sv, MAGIC* mg);
906 int (*svt_free)(SV* sv, MAGIC* mg);
908 This MGVTBL structure is set at compile-time in F<perl.h> and there are
909 currently 19 types (or 21 with overloading turned on). These different
910 structures contain pointers to various routines that perform additional
911 actions depending on which function is being called.
913 Function pointer Action taken
914 ---------------- ------------
915 svt_get Do something before the value of the SV is retrieved.
916 svt_set Do something after the SV is assigned a value.
917 svt_len Report on the SV's length.
918 svt_clear Clear something the SV represents.
919 svt_free Free any extra storage associated with the SV.
921 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
922 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
924 { magic_get, magic_set, magic_len, 0, 0 }
926 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
927 if a get operation is being performed, the routine C<magic_get> is
928 called. All the various routines for the various magical types begin
929 with C<magic_>. NOTE: the magic routines are not considered part of
930 the Perl API, and may not be exported by the Perl library.
932 The current kinds of Magic Virtual Tables are:
935 (old-style char and macro) MGVTBL Type of magic
936 -------------------------- ------ ----------------------------
937 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
938 A PERL_MAGIC_overload vtbl_amagic %OVERLOAD hash
939 a PERL_MAGIC_overload_elem vtbl_amagicelem %OVERLOAD hash element
940 c PERL_MAGIC_overload_table (none) Holds overload table (AMT)
942 B PERL_MAGIC_bm vtbl_bm Boyer-Moore (fast string search)
943 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
945 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
947 E PERL_MAGIC_env vtbl_env %ENV hash
948 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
949 f PERL_MAGIC_fm vtbl_fm Formline ('compiled' format)
950 g PERL_MAGIC_regex_global vtbl_mglob m//g target / study()ed string
951 I PERL_MAGIC_isa vtbl_isa @ISA array
952 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
953 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
954 L PERL_MAGIC_dbfile (none) Debugger %_<filename
955 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename element
956 m PERL_MAGIC_mutex vtbl_mutex ???
957 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale collate transformation
958 P PERL_MAGIC_tied vtbl_pack Tied array or hash
959 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
960 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
961 r PERL_MAGIC_qr vtbl_qr precompiled qr// regex
962 S PERL_MAGIC_sig vtbl_sig %SIG hash
963 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
964 t PERL_MAGIC_taint vtbl_taint Taintedness
965 U PERL_MAGIC_uvar vtbl_uvar Available for use by extensions
966 v PERL_MAGIC_vec vtbl_vec vec() lvalue
967 V PERL_MAGIC_vstring (none) v-string scalars
968 w PERL_MAGIC_utf8 vtbl_utf8 UTF-8 length+offset cache
969 x PERL_MAGIC_substr vtbl_substr substr() lvalue
970 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
971 variable / smart parameter
973 * PERL_MAGIC_glob vtbl_glob GV (typeglob)
974 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
975 . PERL_MAGIC_pos vtbl_pos pos() lvalue
976 < PERL_MAGIC_backref vtbl_backref ???
977 ~ PERL_MAGIC_ext (none) Available for use by extensions
979 When an uppercase and lowercase letter both exist in the table, then the
980 uppercase letter is typically used to represent some kind of composite type
981 (a list or a hash), and the lowercase letter is used to represent an element
982 of that composite type. Some internals code makes use of this case
983 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
985 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
986 specifically for use by extensions and will not be used by perl itself.
987 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
988 to variables (typically objects). This is especially useful because
989 there is no way for normal perl code to corrupt this private information
990 (unlike using extra elements of a hash object).
992 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
993 C function any time a scalar's value is used or changed. The C<MAGIC>'s
994 C<mg_ptr> field points to a C<ufuncs> structure:
997 I32 (*uf_val)(pTHX_ IV, SV*);
998 I32 (*uf_set)(pTHX_ IV, SV*);
1002 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1003 function will be called with C<uf_index> as the first arg and a pointer to
1004 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1005 magic is shown below. Note that the ufuncs structure is copied by
1006 sv_magic, so you can safely allocate it on the stack.
1014 uf.uf_val = &my_get_fn;
1015 uf.uf_set = &my_set_fn;
1017 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1019 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1020 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1021 extra care to avoid conflict. Typically only using the magic on
1022 objects blessed into the same class as the extension is sufficient.
1023 For C<PERL_MAGIC_ext> magic, it may also be appropriate to add an I32
1024 'signature' at the top of the private data area and check that.
1026 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1027 earlier do B<not> invoke 'set' magic on their targets. This must
1028 be done by the user either by calling the C<SvSETMAGIC()> macro after
1029 calling these functions, or by using one of the C<sv_set*_mg()> or
1030 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1031 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1032 obtained from external sources in functions that don't handle magic.
1033 See L<perlapi> for a description of these functions.
1034 For example, calls to the C<sv_cat*()> functions typically need to be
1035 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1036 since their implementation handles 'get' magic.
1038 =head2 Finding Magic
1040 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
1042 This routine returns a pointer to the C<MAGIC> structure stored in the SV.
1043 If the SV does not have that magical feature, C<NULL> is returned. Also,
1044 if the SV is not of type SVt_PVMG, Perl may core dump.
1046 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1048 This routine checks to see what types of magic C<sv> has. If the mg_type
1049 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1050 the mg_type field is changed to be the lowercase letter.
1052 =head2 Understanding the Magic of Tied Hashes and Arrays
1054 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1057 WARNING: As of the 5.004 release, proper usage of the array and hash
1058 access functions requires understanding a few caveats. Some
1059 of these caveats are actually considered bugs in the API, to be fixed
1060 in later releases, and are bracketed with [MAYCHANGE] below. If
1061 you find yourself actually applying such information in this section, be
1062 aware that the behavior may change in the future, umm, without warning.
1064 The perl tie function associates a variable with an object that implements
1065 the various GET, SET, etc methods. To perform the equivalent of the perl
1066 tie function from an XSUB, you must mimic this behaviour. The code below
1067 carries out the necessary steps - firstly it creates a new hash, and then
1068 creates a second hash which it blesses into the class which will implement
1069 the tie methods. Lastly it ties the two hashes together, and returns a
1070 reference to the new tied hash. Note that the code below does NOT call the
1071 TIEHASH method in the MyTie class -
1072 see L<Calling Perl Routines from within C Programs> for details on how
1083 tie = newRV_noinc((SV*)newHV());
1084 stash = gv_stashpv("MyTie", TRUE);
1085 sv_bless(tie, stash);
1086 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1087 RETVAL = newRV_noinc(hash);
1091 The C<av_store> function, when given a tied array argument, merely
1092 copies the magic of the array onto the value to be "stored", using
1093 C<mg_copy>. It may also return NULL, indicating that the value did not
1094 actually need to be stored in the array. [MAYCHANGE] After a call to
1095 C<av_store> on a tied array, the caller will usually need to call
1096 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1097 TIEARRAY object. If C<av_store> did return NULL, a call to
1098 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1101 The previous paragraph is applicable verbatim to tied hash access using the
1102 C<hv_store> and C<hv_store_ent> functions as well.
1104 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1105 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1106 has been initialized using C<mg_copy>. Note the value so returned does not
1107 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1108 need to call C<mg_get()> on the returned value in order to actually invoke
1109 the perl level "FETCH" method on the underlying TIE object. Similarly,
1110 you may also call C<mg_set()> on the return value after possibly assigning
1111 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1112 method on the TIE object. [/MAYCHANGE]
1115 In other words, the array or hash fetch/store functions don't really
1116 fetch and store actual values in the case of tied arrays and hashes. They
1117 merely call C<mg_copy> to attach magic to the values that were meant to be
1118 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1119 do the job of invoking the TIE methods on the underlying objects. Thus
1120 the magic mechanism currently implements a kind of lazy access to arrays
1123 Currently (as of perl version 5.004), use of the hash and array access
1124 functions requires the user to be aware of whether they are operating on
1125 "normal" hashes and arrays, or on their tied variants. The API may be
1126 changed to provide more transparent access to both tied and normal data
1127 types in future versions.
1130 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1131 are mere sugar to invoke some perl method calls while using the uniform hash
1132 and array syntax. The use of this sugar imposes some overhead (typically
1133 about two to four extra opcodes per FETCH/STORE operation, in addition to
1134 the creation of all the mortal variables required to invoke the methods).
1135 This overhead will be comparatively small if the TIE methods are themselves
1136 substantial, but if they are only a few statements long, the overhead
1137 will not be insignificant.
1139 =head2 Localizing changes
1141 Perl has a very handy construction
1148 This construction is I<approximately> equivalent to
1157 The biggest difference is that the first construction would
1158 reinstate the initial value of $var, irrespective of how control exits
1159 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1160 more efficient as well.
1162 There is a way to achieve a similar task from C via Perl API: create a
1163 I<pseudo-block>, and arrange for some changes to be automatically
1164 undone at the end of it, either explicit, or via a non-local exit (via
1165 die()). A I<block>-like construct is created by a pair of
1166 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1167 Such a construct may be created specially for some important localized
1168 task, or an existing one (like boundaries of enclosing Perl
1169 subroutine/block, or an existing pair for freeing TMPs) may be
1170 used. (In the second case the overhead of additional localization must
1171 be almost negligible.) Note that any XSUB is automatically enclosed in
1172 an C<ENTER>/C<LEAVE> pair.
1174 Inside such a I<pseudo-block> the following service is available:
1178 =item C<SAVEINT(int i)>
1180 =item C<SAVEIV(IV i)>
1182 =item C<SAVEI32(I32 i)>
1184 =item C<SAVELONG(long i)>
1186 These macros arrange things to restore the value of integer variable
1187 C<i> at the end of enclosing I<pseudo-block>.
1189 =item C<SAVESPTR(s)>
1191 =item C<SAVEPPTR(p)>
1193 These macros arrange things to restore the value of pointers C<s> and
1194 C<p>. C<s> must be a pointer of a type which survives conversion to
1195 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1198 =item C<SAVEFREESV(SV *sv)>
1200 The refcount of C<sv> would be decremented at the end of
1201 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1202 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1203 extends the lifetime of C<sv> until the beginning of the next statement,
1204 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1205 lifetimes can be wildly different.
1207 Also compare C<SAVEMORTALIZESV>.
1209 =item C<SAVEMORTALIZESV(SV *sv)>
1211 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1212 scope instead of decrementing its reference count. This usually has the
1213 effect of keeping C<sv> alive until the statement that called the currently
1214 live scope has finished executing.
1216 =item C<SAVEFREEOP(OP *op)>
1218 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1220 =item C<SAVEFREEPV(p)>
1222 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1223 end of I<pseudo-block>.
1225 =item C<SAVECLEARSV(SV *sv)>
1227 Clears a slot in the current scratchpad which corresponds to C<sv> at
1228 the end of I<pseudo-block>.
1230 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1232 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1233 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1234 short-lived storage, the corresponding string may be reallocated like
1237 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1239 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1241 At the end of I<pseudo-block> the function C<f> is called with the
1244 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1246 At the end of I<pseudo-block> the function C<f> is called with the
1247 implicit context argument (if any), and C<p>.
1249 =item C<SAVESTACK_POS()>
1251 The current offset on the Perl internal stack (cf. C<SP>) is restored
1252 at the end of I<pseudo-block>.
1256 The following API list contains functions, thus one needs to
1257 provide pointers to the modifiable data explicitly (either C pointers,
1258 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1259 function takes C<int *>.
1263 =item C<SV* save_scalar(GV *gv)>
1265 Equivalent to Perl code C<local $gv>.
1267 =item C<AV* save_ary(GV *gv)>
1269 =item C<HV* save_hash(GV *gv)>
1271 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1273 =item C<void save_item(SV *item)>
1275 Duplicates the current value of C<SV>, on the exit from the current
1276 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1277 using the stored value.
1279 =item C<void save_list(SV **sarg, I32 maxsarg)>
1281 A variant of C<save_item> which takes multiple arguments via an array
1282 C<sarg> of C<SV*> of length C<maxsarg>.
1284 =item C<SV* save_svref(SV **sptr)>
1286 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1288 =item C<void save_aptr(AV **aptr)>
1290 =item C<void save_hptr(HV **hptr)>
1292 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1296 The C<Alias> module implements localization of the basic types within the
1297 I<caller's scope>. People who are interested in how to localize things in
1298 the containing scope should take a look there too.
1302 =head2 XSUBs and the Argument Stack
1304 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1305 An XSUB routine will have a stack that contains the arguments from the Perl
1306 program, and a way to map from the Perl data structures to a C equivalent.
1308 The stack arguments are accessible through the C<ST(n)> macro, which returns
1309 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1310 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1313 Most of the time, output from the C routine can be handled through use of
1314 the RETVAL and OUTPUT directives. However, there are some cases where the
1315 argument stack is not already long enough to handle all the return values.
1316 An example is the POSIX tzname() call, which takes no arguments, but returns
1317 two, the local time zone's standard and summer time abbreviations.
1319 To handle this situation, the PPCODE directive is used and the stack is
1320 extended using the macro:
1324 where C<SP> is the macro that represents the local copy of the stack pointer,
1325 and C<num> is the number of elements the stack should be extended by.
1327 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1328 macro. The pushed values will often need to be "mortal" (See
1329 L</Reference Counts and Mortality>).
1331 PUSHs(sv_2mortal(newSViv(an_integer)))
1332 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1333 PUSHs(sv_2mortal(newSVnv(3.141592)))
1335 And now the Perl program calling C<tzname>, the two values will be assigned
1338 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1340 An alternate (and possibly simpler) method to pushing values on the stack is
1345 This macro automatically adjust the stack for you, if needed. Thus, you
1346 do not need to call C<EXTEND> to extend the stack.
1348 Despite their suggestions in earlier versions of this document the macros
1349 C<PUSHi>, C<PUSHn> and C<PUSHp> are I<not> suited to XSUBs which return
1350 multiple results, see L</Putting a C value on Perl stack>.
1352 For more information, consult L<perlxs> and L<perlxstut>.
1354 =head2 Calling Perl Routines from within C Programs
1356 There are four routines that can be used to call a Perl subroutine from
1357 within a C program. These four are:
1359 I32 call_sv(SV*, I32);
1360 I32 call_pv(const char*, I32);
1361 I32 call_method(const char*, I32);
1362 I32 call_argv(const char*, I32, register char**);
1364 The routine most often used is C<call_sv>. The C<SV*> argument
1365 contains either the name of the Perl subroutine to be called, or a
1366 reference to the subroutine. The second argument consists of flags
1367 that control the context in which the subroutine is called, whether
1368 or not the subroutine is being passed arguments, how errors should be
1369 trapped, and how to treat return values.
1371 All four routines return the number of arguments that the subroutine returned
1374 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1375 but those names are now deprecated; macros of the same name are provided for
1378 When using any of these routines (except C<call_argv>), the programmer
1379 must manipulate the Perl stack. These include the following macros and
1394 For a detailed description of calling conventions from C to Perl,
1395 consult L<perlcall>.
1397 =head2 Memory Allocation
1401 All memory meant to be used with the Perl API functions should be manipulated
1402 using the macros described in this section. The macros provide the necessary
1403 transparency between differences in the actual malloc implementation that is
1406 It is suggested that you enable the version of malloc that is distributed
1407 with Perl. It keeps pools of various sizes of unallocated memory in
1408 order to satisfy allocation requests more quickly. However, on some
1409 platforms, it may cause spurious malloc or free errors.
1411 The following three macros are used to initially allocate memory :
1413 New(x, pointer, number, type);
1414 Newc(x, pointer, number, type, cast);
1415 Newz(x, pointer, number, type);
1417 The first argument C<x> was a "magic cookie" that was used to keep track
1418 of who called the macro, to help when debugging memory problems. However,
1419 the current code makes no use of this feature (most Perl developers now
1420 use run-time memory checkers), so this argument can be any number.
1422 The second argument C<pointer> should be the name of a variable that will
1423 point to the newly allocated memory.
1425 The third and fourth arguments C<number> and C<type> specify how many of
1426 the specified type of data structure should be allocated. The argument
1427 C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1428 should be used if the C<pointer> argument is different from the C<type>
1431 Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1432 to zero out all the newly allocated memory.
1436 Renew(pointer, number, type);
1437 Renewc(pointer, number, type, cast);
1440 These three macros are used to change a memory buffer size or to free a
1441 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1442 match those of C<New> and C<Newc> with the exception of not needing the
1443 "magic cookie" argument.
1447 Move(source, dest, number, type);
1448 Copy(source, dest, number, type);
1449 Zero(dest, number, type);
1451 These three macros are used to move, copy, or zero out previously allocated
1452 memory. The C<source> and C<dest> arguments point to the source and
1453 destination starting points. Perl will move, copy, or zero out C<number>
1454 instances of the size of the C<type> data structure (using the C<sizeof>
1459 The most recent development releases of Perl has been experimenting with
1460 removing Perl's dependency on the "normal" standard I/O suite and allowing
1461 other stdio implementations to be used. This involves creating a new
1462 abstraction layer that then calls whichever implementation of stdio Perl
1463 was compiled with. All XSUBs should now use the functions in the PerlIO
1464 abstraction layer and not make any assumptions about what kind of stdio
1467 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1469 =head2 Putting a C value on Perl stack
1471 A lot of opcodes (this is an elementary operation in the internal perl
1472 stack machine) put an SV* on the stack. However, as an optimization
1473 the corresponding SV is (usually) not recreated each time. The opcodes
1474 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1475 not constantly freed/created.
1477 Each of the targets is created only once (but see
1478 L<Scratchpads and recursion> below), and when an opcode needs to put
1479 an integer, a double, or a string on stack, it just sets the
1480 corresponding parts of its I<target> and puts the I<target> on stack.
1482 The macro to put this target on stack is C<PUSHTARG>, and it is
1483 directly used in some opcodes, as well as indirectly in zillions of
1484 others, which use it via C<(X)PUSH[pni]>.
1486 Because the target is reused, you must be careful when pushing multiple
1487 values on the stack. The following code will not do what you think:
1492 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1493 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1494 At the end of the operation, the stack does not contain the values 10
1495 and 20, but actually contains two pointers to C<TARG>, which we have set
1496 to 20. If you need to push multiple different values, use C<XPUSHs>,
1497 which bypasses C<TARG>.
1499 On a related note, if you do use C<(X)PUSH[npi]>, then you're going to
1500 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
1501 macros can make use of the local variable C<TARG>.
1505 The question remains on when the SVs which are I<target>s for opcodes
1506 are created. The answer is that they are created when the current unit --
1507 a subroutine or a file (for opcodes for statements outside of
1508 subroutines) -- is compiled. During this time a special anonymous Perl
1509 array is created, which is called a scratchpad for the current
1512 A scratchpad keeps SVs which are lexicals for the current unit and are
1513 targets for opcodes. One can deduce that an SV lives on a scratchpad
1514 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1515 I<target>s have C<SVs_PADTMP> set.
1517 The correspondence between OPs and I<target>s is not 1-to-1. Different
1518 OPs in the compile tree of the unit can use the same target, if this
1519 would not conflict with the expected life of the temporary.
1521 =head2 Scratchpads and recursion
1523 In fact it is not 100% true that a compiled unit contains a pointer to
1524 the scratchpad AV. In fact it contains a pointer to an AV of
1525 (initially) one element, and this element is the scratchpad AV. Why do
1526 we need an extra level of indirection?
1528 The answer is B<recursion>, and maybe B<threads>. Both
1529 these can create several execution pointers going into the same
1530 subroutine. For the subroutine-child not write over the temporaries
1531 for the subroutine-parent (lifespan of which covers the call to the
1532 child), the parent and the child should have different
1533 scratchpads. (I<And> the lexicals should be separate anyway!)
1535 So each subroutine is born with an array of scratchpads (of length 1).
1536 On each entry to the subroutine it is checked that the current
1537 depth of the recursion is not more than the length of this array, and
1538 if it is, new scratchpad is created and pushed into the array.
1540 The I<target>s on this scratchpad are C<undef>s, but they are already
1541 marked with correct flags.
1543 =head1 Compiled code
1547 Here we describe the internal form your code is converted to by
1548 Perl. Start with a simple example:
1552 This is converted to a tree similar to this one:
1560 (but slightly more complicated). This tree reflects the way Perl
1561 parsed your code, but has nothing to do with the execution order.
1562 There is an additional "thread" going through the nodes of the tree
1563 which shows the order of execution of the nodes. In our simplified
1564 example above it looks like:
1566 $b ---> $c ---> + ---> $a ---> assign-to
1568 But with the actual compile tree for C<$a = $b + $c> it is different:
1569 some nodes I<optimized away>. As a corollary, though the actual tree
1570 contains more nodes than our simplified example, the execution order
1571 is the same as in our example.
1573 =head2 Examining the tree
1575 If you have your perl compiled for debugging (usually done with
1576 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
1577 compiled tree by specifying C<-Dx> on the Perl command line. The
1578 output takes several lines per node, and for C<$b+$c> it looks like
1583 FLAGS = (SCALAR,KIDS)
1585 TYPE = null ===> (4)
1587 FLAGS = (SCALAR,KIDS)
1589 3 TYPE = gvsv ===> 4
1595 TYPE = null ===> (5)
1597 FLAGS = (SCALAR,KIDS)
1599 4 TYPE = gvsv ===> 5
1605 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1606 not optimized away (one per number in the left column). The immediate
1607 children of the given node correspond to C<{}> pairs on the same level
1608 of indentation, thus this listing corresponds to the tree:
1616 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1617 4 5 6> (node C<6> is not included into above listing), i.e.,
1618 C<gvsv gvsv add whatever>.
1620 Each of these nodes represents an op, a fundamental operation inside the
1621 Perl core. The code which implements each operation can be found in the
1622 F<pp*.c> files; the function which implements the op with type C<gvsv>
1623 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1624 different numbers of children: C<add> is a binary operator, as one would
1625 expect, and so has two children. To accommodate the various different
1626 numbers of children, there are various types of op data structure, and
1627 they link together in different ways.
1629 The simplest type of op structure is C<OP>: this has no children. Unary
1630 operators, C<UNOP>s, have one child, and this is pointed to by the
1631 C<op_first> field. Binary operators (C<BINOP>s) have not only an
1632 C<op_first> field but also an C<op_last> field. The most complex type of
1633 op is a C<LISTOP>, which has any number of children. In this case, the
1634 first child is pointed to by C<op_first> and the last child by
1635 C<op_last>. The children in between can be found by iteratively
1636 following the C<op_sibling> pointer from the first child to the last.
1638 There are also two other op types: a C<PMOP> holds a regular expression,
1639 and has no children, and a C<LOOP> may or may not have children. If the
1640 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1641 complicate matters, if a C<UNOP> is actually a C<null> op after
1642 optimization (see L</Compile pass 2: context propagation>) it will still
1643 have children in accordance with its former type.
1645 Another way to examine the tree is to use a compiler back-end module, such
1648 =head2 Compile pass 1: check routines
1650 The tree is created by the compiler while I<yacc> code feeds it
1651 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1652 the first pass of perl compilation.
1654 What makes this pass interesting for perl developers is that some
1655 optimization may be performed on this pass. This is optimization by
1656 so-called "check routines". The correspondence between node names
1657 and corresponding check routines is described in F<opcode.pl> (do not
1658 forget to run C<make regen_headers> if you modify this file).
1660 A check routine is called when the node is fully constructed except
1661 for the execution-order thread. Since at this time there are no
1662 back-links to the currently constructed node, one can do most any
1663 operation to the top-level node, including freeing it and/or creating
1664 new nodes above/below it.
1666 The check routine returns the node which should be inserted into the
1667 tree (if the top-level node was not modified, check routine returns
1670 By convention, check routines have names C<ck_*>. They are usually
1671 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1672 called from F<perly.y>).
1674 =head2 Compile pass 1a: constant folding
1676 Immediately after the check routine is called the returned node is
1677 checked for being compile-time executable. If it is (the value is
1678 judged to be constant) it is immediately executed, and a I<constant>
1679 node with the "return value" of the corresponding subtree is
1680 substituted instead. The subtree is deleted.
1682 If constant folding was not performed, the execution-order thread is
1685 =head2 Compile pass 2: context propagation
1687 When a context for a part of compile tree is known, it is propagated
1688 down through the tree. At this time the context can have 5 values
1689 (instead of 2 for runtime context): void, boolean, scalar, list, and
1690 lvalue. In contrast with the pass 1 this pass is processed from top
1691 to bottom: a node's context determines the context for its children.
1693 Additional context-dependent optimizations are performed at this time.
1694 Since at this moment the compile tree contains back-references (via
1695 "thread" pointers), nodes cannot be free()d now. To allow
1696 optimized-away nodes at this stage, such nodes are null()ified instead
1697 of free()ing (i.e. their type is changed to OP_NULL).
1699 =head2 Compile pass 3: peephole optimization
1701 After the compile tree for a subroutine (or for an C<eval> or a file)
1702 is created, an additional pass over the code is performed. This pass
1703 is neither top-down or bottom-up, but in the execution order (with
1704 additional complications for conditionals). These optimizations are
1705 done in the subroutine peep(). Optimizations performed at this stage
1706 are subject to the same restrictions as in the pass 2.
1708 =head2 Pluggable runops
1710 The compile tree is executed in a runops function. There are two runops
1711 functions in F<run.c>. C<Perl_runops_debug> is used with DEBUGGING and
1712 C<Perl_runops_standard> is used otherwise. For fine control over the
1713 execution of the compile tree it is possible to provide your own runops
1716 It's probably best to copy one of the existing runops functions and
1717 change it to suit your needs. Then, in the BOOT section of your XS
1720 PL_runops = my_runops;
1722 This function should be as efficient as possible to keep your programs
1723 running as fast as possible.
1725 =head1 Examining internal data structures with the C<dump> functions
1727 To aid debugging, the source file F<dump.c> contains a number of
1728 functions which produce formatted output of internal data structures.
1730 The most commonly used of these functions is C<Perl_sv_dump>; it's used
1731 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
1732 C<sv_dump> to produce debugging output from Perl-space, so users of that
1733 module should already be familiar with its format.
1735 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
1736 derivatives, and produces output similar to C<perl -Dx>; in fact,
1737 C<Perl_dump_eval> will dump the main root of the code being evaluated,
1738 exactly like C<-Dx>.
1740 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
1741 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
1742 subroutines in a package like so: (Thankfully, these are all xsubs, so
1743 there is no op tree)
1745 (gdb) print Perl_dump_packsubs(PL_defstash)
1747 SUB attributes::bootstrap = (xsub 0x811fedc 0)
1749 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
1751 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
1753 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
1755 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
1757 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
1758 the op tree of the main root.
1760 =head1 How multiple interpreters and concurrency are supported
1762 =head2 Background and PERL_IMPLICIT_CONTEXT
1764 The Perl interpreter can be regarded as a closed box: it has an API
1765 for feeding it code or otherwise making it do things, but it also has
1766 functions for its own use. This smells a lot like an object, and
1767 there are ways for you to build Perl so that you can have multiple
1768 interpreters, with one interpreter represented either as a C structure,
1769 or inside a thread-specific structure. These structures contain all
1770 the context, the state of that interpreter.
1772 Two macros control the major Perl build flavors: MULTIPLICITY and
1773 USE_5005THREADS. The MULTIPLICITY build has a C structure
1774 that packages all the interpreter state, and there is a similar thread-specific
1775 data structure under USE_5005THREADS. In both cases,
1776 PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1777 support for passing in a "hidden" first argument that represents all three
1780 All this obviously requires a way for the Perl internal functions to be
1781 either subroutines taking some kind of structure as the first
1782 argument, or subroutines taking nothing as the first argument. To
1783 enable these two very different ways of building the interpreter,
1784 the Perl source (as it does in so many other situations) makes heavy
1785 use of macros and subroutine naming conventions.
1787 First problem: deciding which functions will be public API functions and
1788 which will be private. All functions whose names begin C<S_> are private
1789 (think "S" for "secret" or "static"). All other functions begin with
1790 "Perl_", but just because a function begins with "Perl_" does not mean it is
1791 part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
1792 function is part of the API is to find its entry in L<perlapi>.
1793 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
1794 think it should be (i.e., you need it for your extension), send mail via
1795 L<perlbug> explaining why you think it should be.
1797 Second problem: there must be a syntax so that the same subroutine
1798 declarations and calls can pass a structure as their first argument,
1799 or pass nothing. To solve this, the subroutines are named and
1800 declared in a particular way. Here's a typical start of a static
1801 function used within the Perl guts:
1804 S_incline(pTHX_ char *s)
1806 STATIC becomes "static" in C, and may be #define'd to nothing in some
1807 configurations in future.
1809 A public function (i.e. part of the internal API, but not necessarily
1810 sanctioned for use in extensions) begins like this:
1813 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
1815 C<pTHX_> is one of a number of macros (in perl.h) that hide the
1816 details of the interpreter's context. THX stands for "thread", "this",
1817 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1818 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
1819 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
1822 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
1823 first argument containing the interpreter's context. The trailing underscore
1824 in the pTHX_ macro indicates that the macro expansion needs a comma
1825 after the context argument because other arguments follow it. If
1826 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
1827 subroutine is not prototyped to take the extra argument. The form of the
1828 macro without the trailing underscore is used when there are no additional
1831 When a core function calls another, it must pass the context. This
1832 is normally hidden via macros. Consider C<sv_setiv>. It expands into
1833 something like this:
1835 #ifdef PERL_IMPLICIT_CONTEXT
1836 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
1837 /* can't do this for vararg functions, see below */
1839 #define sv_setiv Perl_sv_setiv
1842 This works well, and means that XS authors can gleefully write:
1846 and still have it work under all the modes Perl could have been
1849 This doesn't work so cleanly for varargs functions, though, as macros
1850 imply that the number of arguments is known in advance. Instead we
1851 either need to spell them out fully, passing C<aTHX_> as the first
1852 argument (the Perl core tends to do this with functions like
1853 Perl_warner), or use a context-free version.
1855 The context-free version of Perl_warner is called
1856 Perl_warner_nocontext, and does not take the extra argument. Instead
1857 it does dTHX; to get the context from thread-local storage. We
1858 C<#define warner Perl_warner_nocontext> so that extensions get source
1859 compatibility at the expense of performance. (Passing an arg is
1860 cheaper than grabbing it from thread-local storage.)
1862 You can ignore [pad]THXx when browsing the Perl headers/sources.
1863 Those are strictly for use within the core. Extensions and embedders
1864 need only be aware of [pad]THX.
1866 =head2 So what happened to dTHR?
1868 C<dTHR> was introduced in perl 5.005 to support the older thread model.
1869 The older thread model now uses the C<THX> mechanism to pass context
1870 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
1871 later still have it for backward source compatibility, but it is defined
1874 =head2 How do I use all this in extensions?
1876 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1877 any functions in the Perl API will need to pass the initial context
1878 argument somehow. The kicker is that you will need to write it in
1879 such a way that the extension still compiles when Perl hasn't been
1880 built with PERL_IMPLICIT_CONTEXT enabled.
1882 There are three ways to do this. First, the easy but inefficient way,
1883 which is also the default, in order to maintain source compatibility
1884 with extensions: whenever XSUB.h is #included, it redefines the aTHX
1885 and aTHX_ macros to call a function that will return the context.
1886 Thus, something like:
1890 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
1893 Perl_sv_setiv(Perl_get_context(), sv, num);
1895 or to this otherwise:
1897 Perl_sv_setiv(sv, num);
1899 You have to do nothing new in your extension to get this; since
1900 the Perl library provides Perl_get_context(), it will all just
1903 The second, more efficient way is to use the following template for
1906 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1911 static my_private_function(int arg1, int arg2);
1914 my_private_function(int arg1, int arg2)
1916 dTHX; /* fetch context */
1917 ... call many Perl API functions ...
1922 MODULE = Foo PACKAGE = Foo
1930 my_private_function(arg, 10);
1932 Note that the only two changes from the normal way of writing an
1933 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
1934 including the Perl headers, followed by a C<dTHX;> declaration at
1935 the start of every function that will call the Perl API. (You'll
1936 know which functions need this, because the C compiler will complain
1937 that there's an undeclared identifier in those functions.) No changes
1938 are needed for the XSUBs themselves, because the XS() macro is
1939 correctly defined to pass in the implicit context if needed.
1941 The third, even more efficient way is to ape how it is done within
1945 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1950 /* pTHX_ only needed for functions that call Perl API */
1951 static my_private_function(pTHX_ int arg1, int arg2);
1954 my_private_function(pTHX_ int arg1, int arg2)
1956 /* dTHX; not needed here, because THX is an argument */
1957 ... call Perl API functions ...
1962 MODULE = Foo PACKAGE = Foo
1970 my_private_function(aTHX_ arg, 10);
1972 This implementation never has to fetch the context using a function
1973 call, since it is always passed as an extra argument. Depending on
1974 your needs for simplicity or efficiency, you may mix the previous
1975 two approaches freely.
1977 Never add a comma after C<pTHX> yourself--always use the form of the
1978 macro with the underscore for functions that take explicit arguments,
1979 or the form without the argument for functions with no explicit arguments.
1981 =head2 Should I do anything special if I call perl from multiple threads?
1983 If you create interpreters in one thread and then proceed to call them in
1984 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
1985 initialized correctly in each of those threads.
1987 The C<perl_alloc> and C<perl_clone> API functions will automatically set
1988 the TLS slot to the interpreter they created, so that there is no need to do
1989 anything special if the interpreter is always accessed in the same thread that
1990 created it, and that thread did not create or call any other interpreters
1991 afterwards. If that is not the case, you have to set the TLS slot of the
1992 thread before calling any functions in the Perl API on that particular
1993 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
1994 thread as the first thing you do:
1996 /* do this before doing anything else with some_perl */
1997 PERL_SET_CONTEXT(some_perl);
1999 ... other Perl API calls on some_perl go here ...
2001 =head2 Future Plans and PERL_IMPLICIT_SYS
2003 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2004 that the interpreter knows about itself and pass it around, so too are
2005 there plans to allow the interpreter to bundle up everything it knows
2006 about the environment it's running on. This is enabled with the
2007 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS
2008 and USE_5005THREADS on Windows (see inside iperlsys.h).
2010 This allows the ability to provide an extra pointer (called the "host"
2011 environment) for all the system calls. This makes it possible for
2012 all the system stuff to maintain their own state, broken down into
2013 seven C structures. These are thin wrappers around the usual system
2014 calls (see win32/perllib.c) for the default perl executable, but for a
2015 more ambitious host (like the one that would do fork() emulation) all
2016 the extra work needed to pretend that different interpreters are
2017 actually different "processes", would be done here.
2019 The Perl engine/interpreter and the host are orthogonal entities.
2020 There could be one or more interpreters in a process, and one or
2021 more "hosts", with free association between them.
2023 =head1 Internal Functions
2025 All of Perl's internal functions which will be exposed to the outside
2026 world are prefixed by C<Perl_> so that they will not conflict with XS
2027 functions or functions used in a program in which Perl is embedded.
2028 Similarly, all global variables begin with C<PL_>. (By convention,
2029 static functions start with C<S_>.)
2031 Inside the Perl core, you can get at the functions either with or
2032 without the C<Perl_> prefix, thanks to a bunch of defines that live in
2033 F<embed.h>. This header file is generated automatically from
2034 F<embed.pl>. F<embed.pl> also creates the prototyping header files for
2035 the internal functions, generates the documentation and a lot of other
2036 bits and pieces. It's important that when you add a new function to the
2037 core or change an existing one, you change the data in the table at the
2038 end of F<embed.pl> as well. Here's a sample entry from that table:
2040 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2042 The second column is the return type, the third column the name. Columns
2043 after that are the arguments. The first column is a set of flags:
2049 This function is a part of the public API.
2053 This function has a C<Perl_> prefix; ie, it is defined as C<Perl_av_fetch>
2057 This function has documentation using the C<apidoc> feature which we'll
2058 look at in a second.
2062 Other available flags are:
2068 This is a static function and is defined as C<S_whatever>, and usually
2069 called within the sources as C<whatever(...)>.
2073 This does not use C<aTHX_> and C<pTHX> to pass interpreter context. (See
2074 L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
2078 This function never returns; C<croak>, C<exit> and friends.
2082 This function takes a variable number of arguments, C<printf> style.
2083 The argument list should end with C<...>, like this:
2085 Afprd |void |croak |const char* pat|...
2089 This function is part of the experimental development API, and may change
2090 or disappear without notice.
2094 This function should not have a compatibility macro to define, say,
2095 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2099 This function is not a member of C<CPerlObj>. If you don't know
2100 what this means, don't use it.
2104 This function isn't exported out of the Perl core.
2108 If you edit F<embed.pl>, you will need to run C<make regen_headers> to
2109 force a rebuild of F<embed.h> and other auto-generated files.
2111 =head2 Formatted Printing of IVs, UVs, and NVs
2113 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2114 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2115 following macros for portability
2120 UVxf UV in hexadecimal
2125 These will take care of 64-bit integers and long doubles.
2128 printf("IV is %"IVdf"\n", iv);
2130 The IVdf will expand to whatever is the correct format for the IVs.
2132 If you are printing addresses of pointers, use UVxf combined
2133 with PTR2UV(), do not use %lx or %p.
2135 =head2 Pointer-To-Integer and Integer-To-Pointer
2137 Because pointer size does not necessarily equal integer size,
2138 use the follow macros to do it right.
2143 INT2PTR(pointertotype, integer)
2148 SV *sv = INT2PTR(SV*, iv);
2155 =head2 Source Documentation
2157 There's an effort going on to document the internal functions and
2158 automatically produce reference manuals from them - L<perlapi> is one
2159 such manual which details all the functions which are available to XS
2160 writers. L<perlintern> is the autogenerated manual for the functions
2161 which are not part of the API and are supposedly for internal use only.
2163 Source documentation is created by putting POD comments into the C
2167 =for apidoc sv_setiv
2169 Copies an integer into the given SV. Does not handle 'set' magic. See
2175 Please try and supply some documentation if you add functions to the
2178 =head1 Unicode Support
2180 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2181 writers to understand this support and make sure that the code they
2182 write does not corrupt Unicode data.
2184 =head2 What B<is> Unicode, anyway?
2186 In the olden, less enlightened times, we all used to use ASCII. Most of
2187 us did, anyway. The big problem with ASCII is that it's American. Well,
2188 no, that's not actually the problem; the problem is that it's not
2189 particularly useful for people who don't use the Roman alphabet. What
2190 used to happen was that particular languages would stick their own
2191 alphabet in the upper range of the sequence, between 128 and 255. Of
2192 course, we then ended up with plenty of variants that weren't quite
2193 ASCII, and the whole point of it being a standard was lost.
2195 Worse still, if you've got a language like Chinese or
2196 Japanese that has hundreds or thousands of characters, then you really
2197 can't fit them into a mere 256, so they had to forget about ASCII
2198 altogether, and build their own systems using pairs of numbers to refer
2201 To fix this, some people formed Unicode, Inc. and
2202 produced a new character set containing all the characters you can
2203 possibly think of and more. There are several ways of representing these
2204 characters, and the one Perl uses is called UTF8. UTF8 uses
2205 a variable number of bytes to represent a character, instead of just
2206 one. You can learn more about Unicode at http://www.unicode.org/
2208 =head2 How can I recognise a UTF8 string?
2210 You can't. This is because UTF8 data is stored in bytes just like
2211 non-UTF8 data. The Unicode character 200, (C<0xC8> for you hex types)
2212 capital E with a grave accent, is represented by the two bytes
2213 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2214 has that byte sequence as well. So you can't tell just by looking - this
2215 is what makes Unicode input an interesting problem.
2217 The API function C<is_utf8_string> can help; it'll tell you if a string
2218 contains only valid UTF8 characters. However, it can't do the work for
2219 you. On a character-by-character basis, C<is_utf8_char> will tell you
2220 whether the current character in a string is valid UTF8.
2222 =head2 How does UTF8 represent Unicode characters?
2224 As mentioned above, UTF8 uses a variable number of bytes to store a
2225 character. Characters with values 1...128 are stored in one byte, just
2226 like good ol' ASCII. Character 129 is stored as C<v194.129>; this
2227 continues up to character 191, which is C<v194.191>. Now we've run out of
2228 bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2229 so it goes on, moving to three bytes at character 2048.
2231 Assuming you know you're dealing with a UTF8 string, you can find out
2232 how long the first character in it is with the C<UTF8SKIP> macro:
2234 char *utf = "\305\233\340\240\201";
2237 len = UTF8SKIP(utf); /* len is 2 here */
2239 len = UTF8SKIP(utf); /* len is 3 here */
2241 Another way to skip over characters in a UTF8 string is to use
2242 C<utf8_hop>, which takes a string and a number of characters to skip
2243 over. You're on your own about bounds checking, though, so don't use it
2246 All bytes in a multi-byte UTF8 character will have the high bit set,
2247 so you can test if you need to do something special with this
2248 character like this (the UTF8_IS_INVARIANT() is a macro that tests
2249 whether the byte can be encoded as a single byte even in UTF-8):
2252 UV uv; /* Note: a UV, not a U8, not a char */
2254 if (!UTF8_IS_INVARIANT(*utf))
2255 /* Must treat this as UTF8 */
2256 uv = utf8_to_uv(utf);
2258 /* OK to treat this character as a byte */
2261 You can also see in that example that we use C<utf8_to_uv> to get the
2262 value of the character; the inverse function C<uv_to_utf8> is available
2263 for putting a UV into UTF8:
2265 if (!UTF8_IS_INVARIANT(uv))
2266 /* Must treat this as UTF8 */
2267 utf8 = uv_to_utf8(utf8, uv);
2269 /* OK to treat this character as a byte */
2272 You B<must> convert characters to UVs using the above functions if
2273 you're ever in a situation where you have to match UTF8 and non-UTF8
2274 characters. You may not skip over UTF8 characters in this case. If you
2275 do this, you'll lose the ability to match hi-bit non-UTF8 characters;
2276 for instance, if your UTF8 string contains C<v196.172>, and you skip
2277 that character, you can never match a C<chr(200)> in a non-UTF8 string.
2280 =head2 How does Perl store UTF8 strings?
2282 Currently, Perl deals with Unicode strings and non-Unicode strings
2283 slightly differently. If a string has been identified as being UTF-8
2284 encoded, Perl will set a flag in the SV, C<SVf_UTF8>. You can check and
2285 manipulate this flag with the following macros:
2291 This flag has an important effect on Perl's treatment of the string: if
2292 Unicode data is not properly distinguished, regular expressions,
2293 C<length>, C<substr> and other string handling operations will have
2294 undesirable results.
2296 The problem comes when you have, for instance, a string that isn't
2297 flagged is UTF8, and contains a byte sequence that could be UTF8 -
2298 especially when combining non-UTF8 and UTF8 strings.
2300 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2301 need be sure you don't accidentally knock it off while you're
2302 manipulating SVs. More specifically, you cannot expect to do this:
2311 nsv = newSVpvn(p, len);
2313 The C<char*> string does not tell you the whole story, and you can't
2314 copy or reconstruct an SV just by copying the string value. Check if the
2315 old SV has the UTF8 flag set, and act accordingly:
2319 nsv = newSVpvn(p, len);
2323 In fact, your C<frobnicate> function should be made aware of whether or
2324 not it's dealing with UTF8 data, so that it can handle the string
2327 Since just passing an SV to an XS function and copying the data of
2328 the SV is not enough to copy the UTF8 flags, even less right is just
2329 passing a C<char *> to an XS function.
2331 =head2 How do I convert a string to UTF8?
2333 If you're mixing UTF8 and non-UTF8 strings, you might find it necessary
2334 to upgrade one of the strings to UTF8. If you've got an SV, the easiest
2337 sv_utf8_upgrade(sv);
2339 However, you must not do this, for example:
2342 sv_utf8_upgrade(left);
2344 If you do this in a binary operator, you will actually change one of the
2345 strings that came into the operator, and, while it shouldn't be noticeable
2346 by the end user, it can cause problems.
2348 Instead, C<bytes_to_utf8> will give you a UTF8-encoded B<copy> of its
2349 string argument. This is useful for having the data available for
2350 comparisons and so on, without harming the original SV. There's also
2351 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2352 the string contains any characters above 255 that can't be represented
2355 =head2 Is there anything else I need to know?
2357 Not really. Just remember these things:
2363 There's no way to tell if a string is UTF8 or not. You can tell if an SV
2364 is UTF8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2365 something should be UTF8. Treat the flag as part of the PV, even though
2366 it's not - if you pass on the PV to somewhere, pass on the flag too.
2370 If a string is UTF8, B<always> use C<utf8_to_uv> to get at the value,
2371 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
2375 When writing a character C<uv> to a UTF8 string, B<always> use
2376 C<uv_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2377 you can use C<*s = uv>.
2381 Mixing UTF8 and non-UTF8 strings is tricky. Use C<bytes_to_utf8> to get
2382 a new string which is UTF8 encoded. There are tricks you can use to
2383 delay deciding whether you need to use a UTF8 string until you get to a
2384 high character - C<HALF_UPGRADE> is one of those.
2388 =head1 Custom Operators
2390 Custom operator support is a new experimental feature that allows you to
2391 define your own ops. This is primarily to allow the building of
2392 interpreters for other languages in the Perl core, but it also allows
2393 optimizations through the creation of "macro-ops" (ops which perform the
2394 functions of multiple ops which are usually executed together, such as
2395 C<gvsv, gvsv, add>.)
2397 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
2398 core does not "know" anything special about this op type, and so it will
2399 not be involved in any optimizations. This also means that you can
2400 define your custom ops to be any op structure - unary, binary, list and
2403 It's important to know what custom operators won't do for you. They
2404 won't let you add new syntax to Perl, directly. They won't even let you
2405 add new keywords, directly. In fact, they won't change the way Perl
2406 compiles a program at all. You have to do those changes yourself, after
2407 Perl has compiled the program. You do this either by manipulating the op
2408 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2409 a custom peephole optimizer with the C<optimize> module.
2411 When you do this, you replace ordinary Perl ops with custom ops by
2412 creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2413 PP function. This should be defined in XS code, and should look like
2414 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2415 takes the appropriate number of values from the stack, and you are
2416 responsible for adding stack marks if necessary.
2418 You should also "register" your op with the Perl interpreter so that it
2419 can produce sensible error and warning messages. Since it is possible to
2420 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2421 Perl uses the value of C<< o->op_ppaddr >> as a key into the
2422 C<PL_custom_op_descs> and C<PL_custom_op_names> hashes. This means you
2423 need to enter a name and description for your op at the appropriate
2424 place in the C<PL_custom_op_names> and C<PL_custom_op_descs> hashes.
2426 Forthcoming versions of C<B::Generate> (version 1.0 and above) should
2427 directly support the creation of custom ops by name; C<Opcodes::Custom>
2428 will provide functions which make it trivial to "register" custom ops to
2429 the Perl interpreter.
2433 Until May 1997, this document was maintained by Jeff Okamoto
2434 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2435 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
2437 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2438 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2439 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2440 Stephen McCamant, and Gurusamy Sarathy.
2444 perlapi(1), perlintern(1), perlxs(1), perlembed(1)