1PERLGUTS(1)            Perl Programmers Reference Guide            PERLGUTS(1)
2
3
4

NAME

6       perlguts - Introduction to the Perl API
7

DESCRIPTION

9       This document attempts to describe how to use the Perl API, as well as
10       to provide some info on the basic workings of the Perl core.  It is far
11       from complete and probably contains many errors.  Please refer any
12       questions or comments to the author below.
13

Variables

15   Datatypes
16       Perl has three typedefs that handle Perl's three main data types:
17
18           SV  Scalar Value
19           AV  Array Value
20           HV  Hash Value
21
22       Each typedef has specific routines that manipulate the various data
23       types.
24
25   What is an "IV"?
26       Perl uses a special typedef IV which is a simple signed integer type
27       that is guaranteed to be large enough to hold a pointer (as well as an
28       integer).  Additionally, there is the UV, which is simply an unsigned
29       IV.
30
31       Perl also uses two special typedefs, I32 and I16, which will always be
32       at least 32-bits and 16-bits long, respectively.  (Again, there are U32
33       and U16, as well.)  They will usually be exactly 32 and 16 bits long,
34       but on Crays they will both be 64 bits.
35
36   Working with SVs
37       An SV can be created and loaded with one command.  There are five types
38       of values that can be loaded: an integer value (IV), an unsigned
39       integer value (UV), a double (NV), a string (PV), and another scalar
40       (SV).  ("PV" stands for "Pointer Value".  You might think that it is
41       misnamed because it is described as pointing only to strings.  However,
42       it is possible to have it point to other things.  For example, it could
43       point to an array of UVs.  But, using it for non-strings requires care,
44       as the underlying assumption of much of the internals is that PVs are
45       just for strings.  Often, for example, a trailing "NUL" is tacked on
46       automatically.  The non-string use is documented only in this
47       paragraph.)
48
49       The seven routines are:
50
51           SV*  newSViv(IV);
52           SV*  newSVuv(UV);
53           SV*  newSVnv(double);
54           SV*  newSVpv(const char*, STRLEN);
55           SV*  newSVpvn(const char*, STRLEN);
56           SV*  newSVpvf(const char*, ...);
57           SV*  newSVsv(SV*);
58
59       "STRLEN" is an integer type ("Size_t", usually defined as "size_t" in
60       config.h) guaranteed to be large enough to represent the size of any
61       string that perl can handle.
62
63       In the unlikely case of a SV requiring more complex initialization, you
64       can create an empty SV with newSV(len).  If "len" is 0 an empty SV of
65       type NULL is returned, else an SV of type PV is returned with len + 1
66       (for the "NUL") bytes of storage allocated, accessible via SvPVX.  In
67       both cases the SV has the undef value.
68
69           SV *sv = newSV(0);   /* no storage allocated  */
70           SV *sv = newSV(10);  /* 10 (+1) bytes of uninitialised storage
71                                 * allocated */
72
73       To change the value of an already-existing SV, there are eight
74       routines:
75
76           void  sv_setiv(SV*, IV);
77           void  sv_setuv(SV*, UV);
78           void  sv_setnv(SV*, double);
79           void  sv_setpv(SV*, const char*);
80           void  sv_setpvn(SV*, const char*, STRLEN)
81           void  sv_setpvf(SV*, const char*, ...);
82           void  sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
83                                               SV **, Size_t, bool *);
84           void  sv_setsv(SV*, SV*);
85
86       Notice that you can choose to specify the length of the string to be
87       assigned by using "sv_setpvn", "newSVpvn", or "newSVpv", or you may
88       allow Perl to calculate the length by using "sv_setpv" or by specifying
89       0 as the second argument to "newSVpv".  Be warned, though, that Perl
90       will determine the string's length by using "strlen", which depends on
91       the string terminating with a "NUL" character, and not otherwise
92       containing NULs.
93
94       The arguments of "sv_setpvf" are processed like "sprintf", and the
95       formatted output becomes the value.
96
97       "sv_vsetpvfn" is an analogue of "vsprintf", but it allows you to
98       specify either a pointer to a variable argument list or the address and
99       length of an array of SVs.  The last argument points to a boolean; on
100       return, if that boolean is true, then locale-specific information has
101       been used to format the string, and the string's contents are therefore
102       untrustworthy (see perlsec).  This pointer may be NULL if that
103       information is not important.  Note that this function requires you to
104       specify the length of the format.
105
106       The "sv_set*()" functions are not generic enough to operate on values
107       that have "magic".  See "Magic Virtual Tables" later in this document.
108
109       All SVs that contain strings should be terminated with a "NUL"
110       character.  If it is not "NUL"-terminated there is a risk of core dumps
111       and corruptions from code which passes the string to C functions or
112       system calls which expect a "NUL"-terminated string.  Perl's own
113       functions typically add a trailing "NUL" for this reason.
114       Nevertheless, you should be very careful when you pass a string stored
115       in an SV to a C function or system call.
116
117       To access the actual value that an SV points to, you can use the
118       macros:
119
120           SvIV(SV*)
121           SvUV(SV*)
122           SvNV(SV*)
123           SvPV(SV*, STRLEN len)
124           SvPV_nolen(SV*)
125
126       which will automatically coerce the actual scalar type into an IV, UV,
127       double, or string.
128
129       In the "SvPV" macro, the length of the string returned is placed into
130       the variable "len" (this is a macro, so you do not use &len).  If you
131       do not care what the length of the data is, use the "SvPV_nolen" macro.
132       Historically the "SvPV" macro with the global variable "PL_na" has been
133       used in this case.  But that can be quite inefficient because "PL_na"
134       must be accessed in thread-local storage in threaded Perl.  In any
135       case, remember that Perl allows arbitrary strings of data that may both
136       contain NULs and might not be terminated by a "NUL".
137
138       Also remember that C doesn't allow you to safely say "foo(SvPV(s, len),
139       len);".  It might work with your compiler, but it won't work for
140       everyone.  Break this sort of statement up into separate assignments:
141
142           SV *s;
143           STRLEN len;
144           char *ptr;
145           ptr = SvPV(s, len);
146           foo(ptr, len);
147
148       If you want to know if the scalar value is TRUE, you can use:
149
150           SvTRUE(SV*)
151
152       Although Perl will automatically grow strings for you, if you need to
153       force Perl to allocate more memory for your SV, you can use the macro
154
155           SvGROW(SV*, STRLEN newlen)
156
157       which will determine if more memory needs to be allocated.  If so, it
158       will call the function "sv_grow".  Note that "SvGROW" can only
159       increase, not decrease, the allocated memory of an SV and that it does
160       not automatically add space for the trailing "NUL" byte (perl's own
161       string functions typically do "SvGROW(sv, len + 1)").
162
163       If you want to write to an existing SV's buffer and set its value to a
164       string, use SvPV_force() or one of its variants to force the SV to be a
165       PV.  This will remove any of various types of non-stringness from the
166       SV while preserving the content of the SV in the PV.  This can be used,
167       for example, to append data from an API function to a buffer without
168       extra copying:
169
170           (void)SvPVbyte_force(sv, len);
171           s = SvGROW(sv, len + needlen + 1);
172           /* something that modifies up to needlen bytes at s+len, but
173              modifies newlen bytes
174                eg. newlen = read(fd, s + len, needlen);
175              ignoring errors for these examples
176            */
177           s[len + newlen] = '\0';
178           SvCUR_set(sv, len + newlen);
179           SvUTF8_off(sv);
180           SvSETMAGIC(sv);
181
182       If you already have the data in memory or if you want to keep your code
183       simple, you can use one of the sv_cat*() variants, such as sv_catpvn().
184       If you want to insert anywhere in the string you can use sv_insert() or
185       sv_insert_flags().
186
187       If you don't need the existing content of the SV, you can avoid some
188       copying with:
189
190           SvPVCLEAR(sv);
191           s = SvGROW(sv, needlen + 1);
192           /* something that modifies up to needlen bytes at s, but modifies
193              newlen bytes
194                eg. newlen = read(fd, s. needlen);
195            */
196           s[newlen] = '\0';
197           SvCUR_set(sv, newlen);
198           SvPOK_only(sv); /* also clears SVf_UTF8 */
199           SvSETMAGIC(sv);
200
201       Again, if you already have the data in memory or want to avoid the
202       complexity of the above, you can use sv_setpvn().
203
204       If you have a buffer allocated with Newx() and want to set that as the
205       SV's value, you can use sv_usepvn_flags().  That has some requirements
206       if you want to avoid perl re-allocating the buffer to fit the trailing
207       NUL:
208
209          Newx(buf, somesize+1, char);
210          /* ... fill in buf ... */
211          buf[somesize] = '\0';
212          sv_usepvn_flags(sv, buf, somesize, SV_SMAGIC | SV_HAS_TRAILING_NUL);
213          /* buf now belongs to perl, don't release it */
214
215       If you have an SV and want to know what kind of data Perl thinks is
216       stored in it, you can use the following macros to check the type of SV
217       you have.
218
219           SvIOK(SV*)
220           SvNOK(SV*)
221           SvPOK(SV*)
222
223       You can get and set the current length of the string stored in an SV
224       with the following macros:
225
226           SvCUR(SV*)
227           SvCUR_set(SV*, I32 val)
228
229       You can also get a pointer to the end of the string stored in the SV
230       with the macro:
231
232           SvEND(SV*)
233
234       But note that these last three macros are valid only if "SvPOK()" is
235       true.
236
237       If you want to append something to the end of string stored in an
238       "SV*", you can use the following functions:
239
240           void  sv_catpv(SV*, const char*);
241           void  sv_catpvn(SV*, const char*, STRLEN);
242           void  sv_catpvf(SV*, const char*, ...);
243           void  sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
244                                                                    I32, bool);
245           void  sv_catsv(SV*, SV*);
246
247       The first function calculates the length of the string to be appended
248       by using "strlen".  In the second, you specify the length of the string
249       yourself.  The third function processes its arguments like "sprintf"
250       and appends the formatted output.  The fourth function works like
251       "vsprintf".  You can specify the address and length of an array of SVs
252       instead of the va_list argument.  The fifth function extends the string
253       stored in the first SV with the string stored in the second SV.  It
254       also forces the second SV to be interpreted as a string.
255
256       The "sv_cat*()" functions are not generic enough to operate on values
257       that have "magic".  See "Magic Virtual Tables" later in this document.
258
259       If you know the name of a scalar variable, you can get a pointer to its
260       SV by using the following:
261
262           SV*  get_sv("package::varname", 0);
263
264       This returns NULL if the variable does not exist.
265
266       If you want to know if this variable (or any other SV) is actually
267       "defined", you can call:
268
269           SvOK(SV*)
270
271       The scalar "undef" value is stored in an SV instance called
272       "PL_sv_undef".
273
274       Its address can be used whenever an "SV*" is needed.  Make sure that
275       you don't try to compare a random sv with &PL_sv_undef.  For example
276       when interfacing Perl code, it'll work correctly for:
277
278         foo(undef);
279
280       But won't work when called as:
281
282         $x = undef;
283         foo($x);
284
285       So to repeat always use SvOK() to check whether an sv is defined.
286
287       Also you have to be careful when using &PL_sv_undef as a value in AVs
288       or HVs (see "AVs, HVs and undefined values").
289
290       There are also the two values "PL_sv_yes" and "PL_sv_no", which contain
291       boolean TRUE and FALSE values, respectively.  Like "PL_sv_undef", their
292       addresses can be used whenever an "SV*" is needed.
293
294       Do not be fooled into thinking that "(SV *) 0" is the same as
295       &PL_sv_undef.  Take this code:
296
297           SV* sv = (SV*) 0;
298           if (I-am-to-return-a-real-value) {
299                   sv = sv_2mortal(newSViv(42));
300           }
301           sv_setsv(ST(0), sv);
302
303       This code tries to return a new SV (which contains the value 42) if it
304       should return a real value, or undef otherwise.  Instead it has
305       returned a NULL pointer which, somewhere down the line, will cause a
306       segmentation violation, bus error, or just weird results.  Change the
307       zero to &PL_sv_undef in the first line and all will be well.
308
309       To free an SV that you've created, call "SvREFCNT_dec(SV*)".  Normally
310       this call is not necessary (see "Reference Counts and Mortality").
311
312   Offsets
313       Perl provides the function "sv_chop" to efficiently remove characters
314       from the beginning of a string; you give it an SV and a pointer to
315       somewhere inside the PV, and it discards everything before the pointer.
316       The efficiency comes by means of a little hack: instead of actually
317       removing the characters, "sv_chop" sets the flag "OOK" (offset OK) to
318       signal to other functions that the offset hack is in effect, and it
319       moves the PV pointer (called "SvPVX") forward by the number of bytes
320       chopped off, and adjusts "SvCUR" and "SvLEN" accordingly.  (A portion
321       of the space between the old and new PV pointers is used to store the
322       count of chopped bytes.)
323
324       Hence, at this point, the start of the buffer that we allocated lives
325       at "SvPVX(sv) - SvIV(sv)" in memory and the PV pointer is pointing into
326       the middle of this allocated storage.
327
328       This is best demonstrated by example.  Normally copy-on-write will
329       prevent the substitution from operator from using this hack, but if you
330       can craft a string for which copy-on-write is not possible, you can see
331       it in play.  In the current implementation, the final byte of a string
332       buffer is used as a copy-on-write reference count.  If the buffer is
333       not big enough, then copy-on-write is skipped.  First have a look at an
334       empty string:
335
336         % ./perl -Ilib -MDevel::Peek -le '$a=""; $a .= ""; Dump $a'
337         SV = PV(0x7ffb7c008a70) at 0x7ffb7c030390
338           REFCNT = 1
339           FLAGS = (POK,pPOK)
340           PV = 0x7ffb7bc05b50 ""\0
341           CUR = 0
342           LEN = 10
343
344       Notice here the LEN is 10.  (It may differ on your platform.)  Extend
345       the length of the string to one less than 10, and do a substitution:
346
347        % ./perl -Ilib -MDevel::Peek -le '$a=""; $a.="123456789"; $a=~s/.//; \
348                                                                   Dump($a)'
349        SV = PV(0x7ffa04008a70) at 0x7ffa04030390
350          REFCNT = 1
351          FLAGS = (POK,OOK,pPOK)
352          OFFSET = 1
353          PV = 0x7ffa03c05b61 ( "\1" . ) "23456789"\0
354          CUR = 8
355          LEN = 9
356
357       Here the number of bytes chopped off (1) is shown next as the OFFSET.
358       The portion of the string between the "real" and the "fake" beginnings
359       is shown in parentheses, and the values of "SvCUR" and "SvLEN" reflect
360       the fake beginning, not the real one.  (The first character of the
361       string buffer happens to have changed to "\1" here, not "1", because
362       the current implementation stores the offset count in the string
363       buffer.  This is subject to change.)
364
365       Something similar to the offset hack is performed on AVs to enable
366       efficient shifting and splicing off the beginning of the array; while
367       "AvARRAY" points to the first element in the array that is visible from
368       Perl, "AvALLOC" points to the real start of the C array.  These are
369       usually the same, but a "shift" operation can be carried out by
370       increasing "AvARRAY" by one and decreasing "AvFILL" and "AvMAX".
371       Again, the location of the real start of the C array only comes into
372       play when freeing the array.  See "av_shift" in av.c.
373
374   What's Really Stored in an SV?
375       Recall that the usual method of determining the type of scalar you have
376       is to use "Sv*OK" macros.  Because a scalar can be both a number and a
377       string, usually these macros will always return TRUE and calling the
378       "Sv*V" macros will do the appropriate conversion of string to
379       integer/double or integer/double to string.
380
381       If you really need to know if you have an integer, double, or string
382       pointer in an SV, you can use the following three macros instead:
383
384           SvIOKp(SV*)
385           SvNOKp(SV*)
386           SvPOKp(SV*)
387
388       These will tell you if you truly have an integer, double, or string
389       pointer stored in your SV.  The "p" stands for private.
390
391       There are various ways in which the private and public flags may
392       differ.  For example, in perl 5.16 and earlier a tied SV may have a
393       valid underlying value in the IV slot (so SvIOKp is true), but the data
394       should be accessed via the FETCH routine rather than directly, so SvIOK
395       is false.  (In perl 5.18 onwards, tied scalars use the flags the same
396       way as untied scalars.)  Another is when numeric conversion has
397       occurred and precision has been lost: only the private flag is set on
398       'lossy' values.  So when an NV is converted to an IV with loss, SvIOKp,
399       SvNOKp and SvNOK will be set, while SvIOK wont be.
400
401       In general, though, it's best to use the "Sv*V" macros.
402
403   Working with AVs
404       There are two ways to create and load an AV.  The first method creates
405       an empty AV:
406
407           AV*  newAV();
408
409       The second method both creates the AV and initially populates it with
410       SVs:
411
412           AV*  av_make(SSize_t num, SV **ptr);
413
414       The second argument points to an array containing "num" "SV*"'s.  Once
415       the AV has been created, the SVs can be destroyed, if so desired.
416
417       Once the AV has been created, the following operations are possible on
418       it:
419
420           void  av_push(AV*, SV*);
421           SV*   av_pop(AV*);
422           SV*   av_shift(AV*);
423           void  av_unshift(AV*, SSize_t num);
424
425       These should be familiar operations, with the exception of
426       "av_unshift".  This routine adds "num" elements at the front of the
427       array with the "undef" value.  You must then use "av_store" (described
428       below) to assign values to these new elements.
429
430       Here are some other functions:
431
432           SSize_t av_top_index(AV*);
433           SV**    av_fetch(AV*, SSize_t key, I32 lval);
434           SV**    av_store(AV*, SSize_t key, SV* val);
435
436       The "av_top_index" function returns the highest index value in an array
437       (just like $#array in Perl).  If the array is empty, -1 is returned.
438       The "av_fetch" function returns the value at index "key", but if "lval"
439       is non-zero, then "av_fetch" will store an undef value at that index.
440       The "av_store" function stores the value "val" at index "key", and does
441       not increment the reference count of "val".  Thus the caller is
442       responsible for taking care of that, and if "av_store" returns NULL,
443       the caller will have to decrement the reference count to avoid a memory
444       leak.  Note that "av_fetch" and "av_store" both return "SV**"'s, not
445       "SV*"'s as their return value.
446
447       A few more:
448
449           void  av_clear(AV*);
450           void  av_undef(AV*);
451           void  av_extend(AV*, SSize_t key);
452
453       The "av_clear" function deletes all the elements in the AV* array, but
454       does not actually delete the array itself.  The "av_undef" function
455       will delete all the elements in the array plus the array itself.  The
456       "av_extend" function extends the array so that it contains at least
457       "key+1" elements.  If "key+1" is less than the currently allocated
458       length of the array, then nothing is done.
459
460       If you know the name of an array variable, you can get a pointer to its
461       AV by using the following:
462
463           AV*  get_av("package::varname", 0);
464
465       This returns NULL if the variable does not exist.
466
467       See "Understanding the Magic of Tied Hashes and Arrays" for more
468       information on how to use the array access functions on tied arrays.
469
470   Working with HVs
471       To create an HV, you use the following routine:
472
473           HV*  newHV();
474
475       Once the HV has been created, the following operations are possible on
476       it:
477
478           SV**  hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
479           SV**  hv_fetch(HV*, const char* key, U32 klen, I32 lval);
480
481       The "klen" parameter is the length of the key being passed in (Note
482       that you cannot pass 0 in as a value of "klen" to tell Perl to measure
483       the length of the key).  The "val" argument contains the SV pointer to
484       the scalar being stored, and "hash" is the precomputed hash value (zero
485       if you want "hv_store" to calculate it for you).  The "lval" parameter
486       indicates whether this fetch is actually a part of a store operation,
487       in which case a new undefined value will be added to the HV with the
488       supplied key and "hv_fetch" will return as if the value had already
489       existed.
490
491       Remember that "hv_store" and "hv_fetch" return "SV**"'s and not just
492       "SV*".  To access the scalar value, you must first dereference the
493       return value.  However, you should check to make sure that the return
494       value is not NULL before dereferencing it.
495
496       The first of these two functions checks if a hash table entry exists,
497       and the second deletes it.
498
499           bool  hv_exists(HV*, const char* key, U32 klen);
500           SV*   hv_delete(HV*, const char* key, U32 klen, I32 flags);
501
502       If "flags" does not include the "G_DISCARD" flag then "hv_delete" will
503       create and return a mortal copy of the deleted value.
504
505       And more miscellaneous functions:
506
507           void   hv_clear(HV*);
508           void   hv_undef(HV*);
509
510       Like their AV counterparts, "hv_clear" deletes all the entries in the
511       hash table but does not actually delete the hash table.  The "hv_undef"
512       deletes both the entries and the hash table itself.
513
514       Perl keeps the actual data in a linked list of structures with a
515       typedef of HE.  These contain the actual key and value pointers (plus
516       extra administrative overhead).  The key is a string pointer; the value
517       is an "SV*".  However, once you have an "HE*", to get the actual key
518       and value, use the routines specified below.
519
520           I32    hv_iterinit(HV*);
521                   /* Prepares starting point to traverse hash table */
522           HE*    hv_iternext(HV*);
523                   /* Get the next entry, and return a pointer to a
524                      structure that has both the key and value */
525           char*  hv_iterkey(HE* entry, I32* retlen);
526                   /* Get the key from an HE structure and also return
527                      the length of the key string */
528           SV*    hv_iterval(HV*, HE* entry);
529                   /* Return an SV pointer to the value of the HE
530                      structure */
531           SV*    hv_iternextsv(HV*, char** key, I32* retlen);
532                   /* This convenience routine combines hv_iternext,
533                      hv_iterkey, and hv_iterval.  The key and retlen
534                      arguments are return values for the key and its
535                      length.  The value is returned in the SV* argument */
536
537       If you know the name of a hash variable, you can get a pointer to its
538       HV by using the following:
539
540           HV*  get_hv("package::varname", 0);
541
542       This returns NULL if the variable does not exist.
543
544       The hash algorithm is defined in the "PERL_HASH" macro:
545
546           PERL_HASH(hash, key, klen)
547
548       The exact implementation of this macro varies by architecture and
549       version of perl, and the return value may change per invocation, so the
550       value is only valid for the duration of a single perl process.
551
552       See "Understanding the Magic of Tied Hashes and Arrays" for more
553       information on how to use the hash access functions on tied hashes.
554
555   Hash API Extensions
556       Beginning with version 5.004, the following functions are also
557       supported:
558
559           HE*     hv_fetch_ent  (HV* tb, SV* key, I32 lval, U32 hash);
560           HE*     hv_store_ent  (HV* tb, SV* key, SV* val, U32 hash);
561
562           bool    hv_exists_ent (HV* tb, SV* key, U32 hash);
563           SV*     hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
564
565           SV*     hv_iterkeysv  (HE* entry);
566
567       Note that these functions take "SV*" keys, which simplifies writing of
568       extension code that deals with hash structures.  These functions also
569       allow passing of "SV*" keys to "tie" functions without forcing you to
570       stringify the keys (unlike the previous set of functions).
571
572       They also return and accept whole hash entries ("HE*"), making their
573       use more efficient (since the hash number for a particular string
574       doesn't have to be recomputed every time).  See perlapi for detailed
575       descriptions.
576
577       The following macros must always be used to access the contents of hash
578       entries.  Note that the arguments to these macros must be simple
579       variables, since they may get evaluated more than once.  See perlapi
580       for detailed descriptions of these macros.
581
582           HePV(HE* he, STRLEN len)
583           HeVAL(HE* he)
584           HeHASH(HE* he)
585           HeSVKEY(HE* he)
586           HeSVKEY_force(HE* he)
587           HeSVKEY_set(HE* he, SV* sv)
588
589       These two lower level macros are defined, but must only be used when
590       dealing with keys that are not "SV*"s:
591
592           HeKEY(HE* he)
593           HeKLEN(HE* he)
594
595       Note that both "hv_store" and "hv_store_ent" do not increment the
596       reference count of the stored "val", which is the caller's
597       responsibility.  If these functions return a NULL value, the caller
598       will usually have to decrement the reference count of "val" to avoid a
599       memory leak.
600
601   AVs, HVs and undefined values
602       Sometimes you have to store undefined values in AVs or HVs.  Although
603       this may be a rare case, it can be tricky.  That's because you're used
604       to using &PL_sv_undef if you need an undefined SV.
605
606       For example, intuition tells you that this XS code:
607
608           AV *av = newAV();
609           av_store( av, 0, &PL_sv_undef );
610
611       is equivalent to this Perl code:
612
613           my @av;
614           $av[0] = undef;
615
616       Unfortunately, this isn't true.  In perl 5.18 and earlier, AVs use
617       &PL_sv_undef as a marker for indicating that an array element has not
618       yet been initialized.  Thus, "exists $av[0]" would be true for the
619       above Perl code, but false for the array generated by the XS code.  In
620       perl 5.20, storing &PL_sv_undef will create a read-only element,
621       because the scalar &PL_sv_undef itself is stored, not a copy.
622
623       Similar problems can occur when storing &PL_sv_undef in HVs:
624
625           hv_store( hv, "key", 3, &PL_sv_undef, 0 );
626
627       This will indeed make the value "undef", but if you try to modify the
628       value of "key", you'll get the following error:
629
630           Modification of non-creatable hash value attempted
631
632       In perl 5.8.0, &PL_sv_undef was also used to mark placeholders in
633       restricted hashes.  This caused such hash entries not to appear when
634       iterating over the hash or when checking for the keys with the
635       "hv_exists" function.
636
637       You can run into similar problems when you store &PL_sv_yes or
638       &PL_sv_no into AVs or HVs.  Trying to modify such elements will give
639       you the following error:
640
641           Modification of a read-only value attempted
642
643       To make a long story short, you can use the special variables
644       &PL_sv_undef, &PL_sv_yes and &PL_sv_no with AVs and HVs, but you have
645       to make sure you know what you're doing.
646
647       Generally, if you want to store an undefined value in an AV or HV, you
648       should not use &PL_sv_undef, but rather create a new undefined value
649       using the "newSV" function, for example:
650
651           av_store( av, 42, newSV(0) );
652           hv_store( hv, "foo", 3, newSV(0), 0 );
653
654   References
655       References are a special type of scalar that point to other data types
656       (including other references).
657
658       To create a reference, use either of the following functions:
659
660           SV* newRV_inc((SV*) thing);
661           SV* newRV_noinc((SV*) thing);
662
663       The "thing" argument can be any of an "SV*", "AV*", or "HV*".  The
664       functions are identical except that "newRV_inc" increments the
665       reference count of the "thing", while "newRV_noinc" does not.  For
666       historical reasons, "newRV" is a synonym for "newRV_inc".
667
668       Once you have a reference, you can use the following macro to
669       dereference the reference:
670
671           SvRV(SV*)
672
673       then call the appropriate routines, casting the returned "SV*" to
674       either an "AV*" or "HV*", if required.
675
676       To determine if an SV is a reference, you can use the following macro:
677
678           SvROK(SV*)
679
680       To discover what type of value the reference refers to, use the
681       following macro and then check the return value.
682
683           SvTYPE(SvRV(SV*))
684
685       The most useful types that will be returned are:
686
687           < SVt_PVAV  Scalar
688           SVt_PVAV    Array
689           SVt_PVHV    Hash
690           SVt_PVCV    Code
691           SVt_PVGV    Glob (possibly a file handle)
692
693       See "svtype" in perlapi for more details.
694
695   Blessed References and Class Objects
696       References are also used to support object-oriented programming.  In
697       perl's OO lexicon, an object is simply a reference that has been
698       blessed into a package (or class).  Once blessed, the programmer may
699       now use the reference to access the various methods in the class.
700
701       A reference can be blessed into a package with the following function:
702
703           SV* sv_bless(SV* sv, HV* stash);
704
705       The "sv" argument must be a reference value.  The "stash" argument
706       specifies which class the reference will belong to.  See "Stashes and
707       Globs" for information on converting class names into stashes.
708
709       /* Still under construction */
710
711       The following function upgrades rv to reference if not already one.
712       Creates a new SV for rv to point to.  If "classname" is non-null, the
713       SV is blessed into the specified class.  SV is returned.
714
715               SV* newSVrv(SV* rv, const char* classname);
716
717       The following three functions copy integer, unsigned integer or double
718       into an SV whose reference is "rv".  SV is blessed if "classname" is
719       non-null.
720
721               SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
722               SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
723               SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
724
725       The following function copies the pointer value (the address, not the
726       string!) into an SV whose reference is rv.  SV is blessed if
727       "classname" is non-null.
728
729               SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
730
731       The following function copies a string into an SV whose reference is
732       "rv".  Set length to 0 to let Perl calculate the string length.  SV is
733       blessed if "classname" is non-null.
734
735           SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
736                                                                STRLEN length);
737
738       The following function tests whether the SV is blessed into the
739       specified class.  It does not check inheritance relationships.
740
741               int  sv_isa(SV* sv, const char* name);
742
743       The following function tests whether the SV is a reference to a blessed
744       object.
745
746               int  sv_isobject(SV* sv);
747
748       The following function tests whether the SV is derived from the
749       specified class.  SV can be either a reference to a blessed object or a
750       string containing a class name.  This is the function implementing the
751       "UNIVERSAL::isa" functionality.
752
753               bool sv_derived_from(SV* sv, const char* name);
754
755       To check if you've got an object derived from a specific class you have
756       to write:
757
758               if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
759
760   Creating New Variables
761       To create a new Perl variable with an undef value which can be accessed
762       from your Perl script, use the following routines, depending on the
763       variable type.
764
765           SV*  get_sv("package::varname", GV_ADD);
766           AV*  get_av("package::varname", GV_ADD);
767           HV*  get_hv("package::varname", GV_ADD);
768
769       Notice the use of GV_ADD as the second parameter.  The new variable can
770       now be set, using the routines appropriate to the data type.
771
772       There are additional macros whose values may be bitwise OR'ed with the
773       "GV_ADD" argument to enable certain extra features.  Those bits are:
774
775       GV_ADDMULTI
776           Marks the variable as multiply defined, thus preventing the:
777
778             Name <varname> used only once: possible typo
779
780           warning.
781
782       GV_ADDWARN
783           Issues the warning:
784
785             Had to create <varname> unexpectedly
786
787           if the variable did not exist before the function was called.
788
789       If you do not specify a package name, the variable is created in the
790       current package.
791
792   Reference Counts and Mortality
793       Perl uses a reference count-driven garbage collection mechanism.  SVs,
794       AVs, or HVs (xV for short in the following) start their life with a
795       reference count of 1.  If the reference count of an xV ever drops to 0,
796       then it will be destroyed and its memory made available for reuse.  At
797       the most basic internal level, reference counts can be manipulated with
798       the following macros:
799
800           int SvREFCNT(SV* sv);
801           SV* SvREFCNT_inc(SV* sv);
802           void SvREFCNT_dec(SV* sv);
803
804       (There are also suffixed versions of the increment and decrement
805       macros, for situations where the full generality of these basic macros
806       can be exchanged for some performance.)
807
808       However, the way a programmer should think about references is not so
809       much in terms of the bare reference count, but in terms of ownership of
810       references.  A reference to an xV can be owned by any of a variety of
811       entities: another xV, the Perl interpreter, an XS data structure, a
812       piece of running code, or a dynamic scope.  An xV generally does not
813       know what entities own the references to it; it only knows how many
814       references there are, which is the reference count.
815
816       To correctly maintain reference counts, it is essential to keep track
817       of what references the XS code is manipulating.  The programmer should
818       always know where a reference has come from and who owns it, and be
819       aware of any creation or destruction of references, and any transfers
820       of ownership.  Because ownership isn't represented explicitly in the xV
821       data structures, only the reference count need be actually maintained
822       by the code, and that means that this understanding of ownership is not
823       actually evident in the code.  For example, transferring ownership of a
824       reference from one owner to another doesn't change the reference count
825       at all, so may be achieved with no actual code.  (The transferring code
826       doesn't touch the referenced object, but does need to ensure that the
827       former owner knows that it no longer owns the reference, and that the
828       new owner knows that it now does.)
829
830       An xV that is visible at the Perl level should not become unreferenced
831       and thus be destroyed.  Normally, an object will only become
832       unreferenced when it is no longer visible, often by the same means that
833       makes it invisible.  For example, a Perl reference value (RV) owns a
834       reference to its referent, so if the RV is overwritten that reference
835       gets destroyed, and the no-longer-reachable referent may be destroyed
836       as a result.
837
838       Many functions have some kind of reference manipulation as part of
839       their purpose.  Sometimes this is documented in terms of ownership of
840       references, and sometimes it is (less helpfully) documented in terms of
841       changes to reference counts.  For example, the newRV_inc() function is
842       documented to create a new RV (with reference count 1) and increment
843       the reference count of the referent that was supplied by the caller.
844       This is best understood as creating a new reference to the referent,
845       which is owned by the created RV, and returning to the caller ownership
846       of the sole reference to the RV.  The newRV_noinc() function instead
847       does not increment the reference count of the referent, but the RV
848       nevertheless ends up owning a reference to the referent.  It is
849       therefore implied that the caller of "newRV_noinc()" is relinquishing a
850       reference to the referent, making this conceptually a more complicated
851       operation even though it does less to the data structures.
852
853       For example, imagine you want to return a reference from an XSUB
854       function.  Inside the XSUB routine, you create an SV which initially
855       has just a single reference, owned by the XSUB routine.  This reference
856       needs to be disposed of before the routine is complete, otherwise it
857       will leak, preventing the SV from ever being destroyed.  So to create
858       an RV referencing the SV, it is most convenient to pass the SV to
859       "newRV_noinc()", which consumes that reference.  Now the XSUB routine
860       no longer owns a reference to the SV, but does own a reference to the
861       RV, which in turn owns a reference to the SV.  The ownership of the
862       reference to the RV is then transferred by the process of returning the
863       RV from the XSUB.
864
865       There are some convenience functions available that can help with the
866       destruction of xVs.  These functions introduce the concept of
867       "mortality".  Much documentation speaks of an xV itself being mortal,
868       but this is misleading.  It is really a reference to an xV that is
869       mortal, and it is possible for there to be more than one mortal
870       reference to a single xV.  For a reference to be mortal means that it
871       is owned by the temps stack, one of perl's many internal stacks, which
872       will destroy that reference "a short time later".  Usually the "short
873       time later" is the end of the current Perl statement.  However, it gets
874       more complicated around dynamic scopes: there can be multiple sets of
875       mortal references hanging around at the same time, with different death
876       dates.  Internally, the actual determinant for when mortal xV
877       references are destroyed depends on two macros, SAVETMPS and FREETMPS.
878       See perlcall and perlxs for more details on these macros.
879
880       Mortal references are mainly used for xVs that are placed on perl's
881       main stack.  The stack is problematic for reference tracking, because
882       it contains a lot of xV references, but doesn't own those references:
883       they are not counted.  Currently, there are many bugs resulting from
884       xVs being destroyed while referenced by the stack, because the stack's
885       uncounted references aren't enough to keep the xVs alive.  So when
886       putting an (uncounted) reference on the stack, it is vitally important
887       to ensure that there will be a counted reference to the same xV that
888       will last at least as long as the uncounted reference.  But it's also
889       important that that counted reference be cleaned up at an appropriate
890       time, and not unduly prolong the xV's life.  For there to be a mortal
891       reference is often the best way to satisfy this requirement, especially
892       if the xV was created especially to be put on the stack and would
893       otherwise be unreferenced.
894
895       To create a mortal reference, use the functions:
896
897           SV*  sv_newmortal()
898           SV*  sv_mortalcopy(SV*)
899           SV*  sv_2mortal(SV*)
900
901       "sv_newmortal()" creates an SV (with the undefined value) whose sole
902       reference is mortal.  "sv_mortalcopy()" creates an xV whose value is a
903       copy of a supplied xV and whose sole reference is mortal.
904       "sv_2mortal()" mortalises an existing xV reference: it transfers
905       ownership of a reference from the caller to the temps stack.  Because
906       "sv_newmortal" gives the new SV no value, it must normally be given one
907       via "sv_setpv", "sv_setiv", etc. :
908
909           SV *tmp = sv_newmortal();
910           sv_setiv(tmp, an_integer);
911
912       As that is multiple C statements it is quite common so see this idiom
913       instead:
914
915           SV *tmp = sv_2mortal(newSViv(an_integer));
916
917       The mortal routines are not just for SVs; AVs and HVs can be made
918       mortal by passing their address (type-casted to "SV*") to the
919       "sv_2mortal" or "sv_mortalcopy" routines.
920
921   Stashes and Globs
922       A stash is a hash that contains all variables that are defined within a
923       package.  Each key of the stash is a symbol name (shared by all the
924       different types of objects that have the same name), and each value in
925       the hash table is a GV (Glob Value).  This GV in turn contains
926       references to the various objects of that name, including (but not
927       limited to) the following:
928
929           Scalar Value
930           Array Value
931           Hash Value
932           I/O Handle
933           Format
934           Subroutine
935
936       There is a single stash called "PL_defstash" that holds the items that
937       exist in the "main" package.  To get at the items in other packages,
938       append the string "::" to the package name.  The items in the "Foo"
939       package are in the stash "Foo::" in PL_defstash.  The items in the
940       "Bar::Baz" package are in the stash "Baz::" in "Bar::"'s stash.
941
942       To get the stash pointer for a particular package, use the function:
943
944           HV*  gv_stashpv(const char* name, I32 flags)
945           HV*  gv_stashsv(SV*, I32 flags)
946
947       The first function takes a literal string, the second uses the string
948       stored in the SV.  Remember that a stash is just a hash table, so you
949       get back an "HV*".  The "flags" flag will create a new package if it is
950       set to GV_ADD.
951
952       The name that "gv_stash*v" wants is the name of the package whose
953       symbol table you want.  The default package is called "main".  If you
954       have multiply nested packages, pass their names to "gv_stash*v",
955       separated by "::" as in the Perl language itself.
956
957       Alternately, if you have an SV that is a blessed reference, you can
958       find out the stash pointer by using:
959
960           HV*  SvSTASH(SvRV(SV*));
961
962       then use the following to get the package name itself:
963
964           char*  HvNAME(HV* stash);
965
966       If you need to bless or re-bless an object you can use the following
967       function:
968
969           SV*  sv_bless(SV*, HV* stash)
970
971       where the first argument, an "SV*", must be a reference, and the second
972       argument is a stash.  The returned "SV*" can now be used in the same
973       way as any other SV.
974
975       For more information on references and blessings, consult perlref.
976
977   Double-Typed SVs
978       Scalar variables normally contain only one type of value, an integer,
979       double, pointer, or reference.  Perl will automatically convert the
980       actual scalar data from the stored type into the requested type.
981
982       Some scalar variables contain more than one type of scalar data.  For
983       example, the variable $! contains either the numeric value of "errno"
984       or its string equivalent from either "strerror" or "sys_errlist[]".
985
986       To force multiple data values into an SV, you must do two things: use
987       the "sv_set*v" routines to add the additional scalar type, then set a
988       flag so that Perl will believe it contains more than one type of data.
989       The four macros to set the flags are:
990
991               SvIOK_on
992               SvNOK_on
993               SvPOK_on
994               SvROK_on
995
996       The particular macro you must use depends on which "sv_set*v" routine
997       you called first.  This is because every "sv_set*v" routine turns on
998       only the bit for the particular type of data being set, and turns off
999       all the rest.
1000
1001       For example, to create a new Perl variable called "dberror" that
1002       contains both the numeric and descriptive string error values, you
1003       could use the following code:
1004
1005           extern int  dberror;
1006           extern char *dberror_list;
1007
1008           SV* sv = get_sv("dberror", GV_ADD);
1009           sv_setiv(sv, (IV) dberror);
1010           sv_setpv(sv, dberror_list[dberror]);
1011           SvIOK_on(sv);
1012
1013       If the order of "sv_setiv" and "sv_setpv" had been reversed, then the
1014       macro "SvPOK_on" would need to be called instead of "SvIOK_on".
1015
1016   Read-Only Values
1017       In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
1018       flag bit with read-only scalars.  So the only way to test whether
1019       "sv_setsv", etc., will raise a "Modification of a read-only value"
1020       error in those versions is:
1021
1022           SvREADONLY(sv) && !SvIsCOW(sv)
1023
1024       Under Perl 5.18 and later, SvREADONLY only applies to read-only
1025       variables, and, under 5.20, copy-on-write scalars can also be read-
1026       only, so the above check is incorrect.  You just want:
1027
1028           SvREADONLY(sv)
1029
1030       If you need to do this check often, define your own macro like this:
1031
1032           #if PERL_VERSION >= 18
1033           # define SvTRULYREADONLY(sv) SvREADONLY(sv)
1034           #else
1035           # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
1036           #endif
1037
1038   Copy on Write
1039       Perl implements a copy-on-write (COW) mechanism for scalars, in which
1040       string copies are not immediately made when requested, but are deferred
1041       until made necessary by one or the other scalar changing.  This is
1042       mostly transparent, but one must take care not to modify string buffers
1043       that are shared by multiple SVs.
1044
1045       You can test whether an SV is using copy-on-write with "SvIsCOW(sv)".
1046
1047       You can force an SV to make its own copy of its string buffer by
1048       calling "sv_force_normal(sv)" or SvPV_force_nolen(sv).
1049
1050       If you want to make the SV drop its string buffer, use
1051       "sv_force_normal_flags(sv, SV_COW_DROP_PV)" or simply "sv_setsv(sv,
1052       NULL)".
1053
1054       All of these functions will croak on read-only scalars (see the
1055       previous section for more on those).
1056
1057       To test that your code is behaving correctly and not modifying COW
1058       buffers, on systems that support mmap(2) (i.e., Unix) you can configure
1059       perl with "-Accflags=-DPERL_DEBUG_READONLY_COW" and it will turn buffer
1060       violations into crashes.  You will find it to be marvellously slow, so
1061       you may want to skip perl's own tests.
1062
1063   Magic Variables
1064       [This section still under construction.  Ignore everything here.  Post
1065       no bills.  Everything not permitted is forbidden.]
1066
1067       Any SV may be magical, that is, it has special features that a normal
1068       SV does not have.  These features are stored in the SV structure in a
1069       linked list of "struct magic"'s, typedef'ed to "MAGIC".
1070
1071           struct magic {
1072               MAGIC*      mg_moremagic;
1073               MGVTBL*     mg_virtual;
1074               U16         mg_private;
1075               char        mg_type;
1076               U8          mg_flags;
1077               I32         mg_len;
1078               SV*         mg_obj;
1079               char*       mg_ptr;
1080           };
1081
1082       Note this is current as of patchlevel 0, and could change at any time.
1083
1084   Assigning Magic
1085       Perl adds magic to an SV using the sv_magic function:
1086
1087         void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
1088
1089       The "sv" argument is a pointer to the SV that is to acquire a new
1090       magical feature.
1091
1092       If "sv" is not already magical, Perl uses the "SvUPGRADE" macro to
1093       convert "sv" to type "SVt_PVMG".  Perl then continues by adding new
1094       magic to the beginning of the linked list of magical features.  Any
1095       prior entry of the same type of magic is deleted.  Note that this can
1096       be overridden, and multiple instances of the same type of magic can be
1097       associated with an SV.
1098
1099       The "name" and "namlen" arguments are used to associate a string with
1100       the magic, typically the name of a variable.  "namlen" is stored in the
1101       "mg_len" field and if "name" is non-null then either a "savepvn" copy
1102       of "name" or "name" itself is stored in the "mg_ptr" field, depending
1103       on whether "namlen" is greater than zero or equal to zero respectively.
1104       As a special case, if "(name && namlen == HEf_SVKEY)" then "name" is
1105       assumed to contain an "SV*" and is stored as-is with its REFCNT
1106       incremented.
1107
1108       The sv_magic function uses "how" to determine which, if any, predefined
1109       "Magic Virtual Table" should be assigned to the "mg_virtual" field.
1110       See the "Magic Virtual Tables" section below.  The "how" argument is
1111       also stored in the "mg_type" field.  The value of "how" should be
1112       chosen from the set of macros "PERL_MAGIC_foo" found in perl.h.  Note
1113       that before these macros were added, Perl internals used to directly
1114       use character literals, so you may occasionally come across old code or
1115       documentation referring to 'U' magic rather than "PERL_MAGIC_uvar" for
1116       example.
1117
1118       The "obj" argument is stored in the "mg_obj" field of the "MAGIC"
1119       structure.  If it is not the same as the "sv" argument, the reference
1120       count of the "obj" object is incremented.  If it is the same, or if the
1121       "how" argument is "PERL_MAGIC_arylen", "PERL_MAGIC_regdatum",
1122       "PERL_MAGIC_regdata", or if it is a NULL pointer, then "obj" is merely
1123       stored, without the reference count being incremented.
1124
1125       See also "sv_magicext" in perlapi for a more flexible way to add magic
1126       to an SV.
1127
1128       There is also a function to add magic to an "HV":
1129
1130           void hv_magic(HV *hv, GV *gv, int how);
1131
1132       This simply calls "sv_magic" and coerces the "gv" argument into an
1133       "SV".
1134
1135       To remove the magic from an SV, call the function sv_unmagic:
1136
1137           int sv_unmagic(SV *sv, int type);
1138
1139       The "type" argument should be equal to the "how" value when the "SV"
1140       was initially made magical.
1141
1142       However, note that "sv_unmagic" removes all magic of a certain "type"
1143       from the "SV".  If you want to remove only certain magic of a "type"
1144       based on the magic virtual table, use "sv_unmagicext" instead:
1145
1146           int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1147
1148   Magic Virtual Tables
1149       The "mg_virtual" field in the "MAGIC" structure is a pointer to an
1150       "MGVTBL", which is a structure of function pointers and stands for
1151       "Magic Virtual Table" to handle the various operations that might be
1152       applied to that variable.
1153
1154       The "MGVTBL" has five (or sometimes eight) pointers to the following
1155       routine types:
1156
1157           int  (*svt_get)  (pTHX_ SV* sv, MAGIC* mg);
1158           int  (*svt_set)  (pTHX_ SV* sv, MAGIC* mg);
1159           U32  (*svt_len)  (pTHX_ SV* sv, MAGIC* mg);
1160           int  (*svt_clear)(pTHX_ SV* sv, MAGIC* mg);
1161           int  (*svt_free) (pTHX_ SV* sv, MAGIC* mg);
1162
1163           int  (*svt_copy) (pTHX_ SV *sv, MAGIC* mg, SV *nsv,
1164                                                 const char *name, I32 namlen);
1165           int  (*svt_dup)  (pTHX_ MAGIC *mg, CLONE_PARAMS *param);
1166           int  (*svt_local)(pTHX_ SV *nsv, MAGIC *mg);
1167
1168       This MGVTBL structure is set at compile-time in perl.h and there are
1169       currently 32 types.  These different structures contain pointers to
1170       various routines that perform additional actions depending on which
1171       function is being called.
1172
1173          Function pointer    Action taken
1174          ----------------    ------------
1175          svt_get             Do something before the value of the SV is
1176                              retrieved.
1177          svt_set             Do something after the SV is assigned a value.
1178          svt_len             Report on the SV's length.
1179          svt_clear           Clear something the SV represents.
1180          svt_free            Free any extra storage associated with the SV.
1181
1182          svt_copy            copy tied variable magic to a tied element
1183          svt_dup             duplicate a magic structure during thread cloning
1184          svt_local           copy magic to local value during 'local'
1185
1186       For instance, the MGVTBL structure called "vtbl_sv" (which corresponds
1187       to an "mg_type" of "PERL_MAGIC_sv") contains:
1188
1189           { magic_get, magic_set, magic_len, 0, 0 }
1190
1191       Thus, when an SV is determined to be magical and of type
1192       "PERL_MAGIC_sv", if a get operation is being performed, the routine
1193       "magic_get" is called.  All the various routines for the various
1194       magical types begin with "magic_".  NOTE: the magic routines are not
1195       considered part of the Perl API, and may not be exported by the Perl
1196       library.
1197
1198       The last three slots are a recent addition, and for source code
1199       compatibility they are only checked for if one of the three flags
1200       MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.  This means that
1201       most code can continue declaring a vtable as a 5-element value.  These
1202       three are currently used exclusively by the threading code, and are
1203       highly subject to change.
1204
1205       The current kinds of Magic Virtual Tables are:
1206
1207        mg_type
1208        (old-style char and macro)   MGVTBL         Type of magic
1209        --------------------------   ------         -------------
1210        \0 PERL_MAGIC_sv             vtbl_sv        Special scalar variable
1211        #  PERL_MAGIC_arylen         vtbl_arylen    Array length ($#ary)
1212        %  PERL_MAGIC_rhash          (none)         Extra data for restricted
1213                                                    hashes
1214        *  PERL_MAGIC_debugvar       vtbl_debugvar  $DB::single, signal, trace
1215                                                    vars
1216        .  PERL_MAGIC_pos            vtbl_pos       pos() lvalue
1217        :  PERL_MAGIC_symtab         (none)         Extra data for symbol
1218                                                    tables
1219        <  PERL_MAGIC_backref        vtbl_backref   For weak ref data
1220        @  PERL_MAGIC_arylen_p       (none)         To move arylen out of XPVAV
1221        B  PERL_MAGIC_bm             vtbl_regexp    Boyer-Moore
1222                                                    (fast string search)
1223        c  PERL_MAGIC_overload_table vtbl_ovrld     Holds overload table
1224                                                    (AMT) on stash
1225        D  PERL_MAGIC_regdata        vtbl_regdata   Regex match position data
1226                                                    (@+ and @- vars)
1227        d  PERL_MAGIC_regdatum       vtbl_regdatum  Regex match position data
1228                                                    element
1229        E  PERL_MAGIC_env            vtbl_env       %ENV hash
1230        e  PERL_MAGIC_envelem        vtbl_envelem   %ENV hash element
1231        f  PERL_MAGIC_fm             vtbl_regexp    Formline
1232                                                    ('compiled' format)
1233        g  PERL_MAGIC_regex_global   vtbl_mglob     m//g target
1234        H  PERL_MAGIC_hints          vtbl_hints     %^H hash
1235        h  PERL_MAGIC_hintselem      vtbl_hintselem %^H hash element
1236        I  PERL_MAGIC_isa            vtbl_isa       @ISA array
1237        i  PERL_MAGIC_isaelem        vtbl_isaelem   @ISA array element
1238        k  PERL_MAGIC_nkeys          vtbl_nkeys     scalar(keys()) lvalue
1239        L  PERL_MAGIC_dbfile         (none)         Debugger %_<filename
1240        l  PERL_MAGIC_dbline         vtbl_dbline    Debugger %_<filename
1241                                                    element
1242        N  PERL_MAGIC_shared         (none)         Shared between threads
1243        n  PERL_MAGIC_shared_scalar  (none)         Shared between threads
1244        o  PERL_MAGIC_collxfrm       vtbl_collxfrm  Locale transformation
1245        P  PERL_MAGIC_tied           vtbl_pack      Tied array or hash
1246        p  PERL_MAGIC_tiedelem       vtbl_packelem  Tied array or hash element
1247        q  PERL_MAGIC_tiedscalar     vtbl_packelem  Tied scalar or handle
1248        r  PERL_MAGIC_qr             vtbl_regexp    Precompiled qr// regex
1249        S  PERL_MAGIC_sig            (none)         %SIG hash
1250        s  PERL_MAGIC_sigelem        vtbl_sigelem   %SIG hash element
1251        t  PERL_MAGIC_taint          vtbl_taint     Taintedness
1252        U  PERL_MAGIC_uvar           vtbl_uvar      Available for use by
1253                                                    extensions
1254        u  PERL_MAGIC_uvar_elem      (none)         Reserved for use by
1255                                                    extensions
1256        V  PERL_MAGIC_vstring        (none)         SV was vstring literal
1257        v  PERL_MAGIC_vec            vtbl_vec       vec() lvalue
1258        w  PERL_MAGIC_utf8           vtbl_utf8      Cached UTF-8 information
1259        x  PERL_MAGIC_substr         vtbl_substr    substr() lvalue
1260        Y  PERL_MAGIC_nonelem        vtbl_nonelem   Array element that does not
1261                                                    exist
1262        y  PERL_MAGIC_defelem        vtbl_defelem   Shadow "foreach" iterator
1263                                                    variable / smart parameter
1264                                                    vivification
1265        \  PERL_MAGIC_lvref          vtbl_lvref     Lvalue reference
1266                                                    constructor
1267        ]  PERL_MAGIC_checkcall      vtbl_checkcall Inlining/mutation of call
1268                                                    to this CV
1269        ~  PERL_MAGIC_ext            (none)         Available for use by
1270                                                    extensions
1271
1272       When an uppercase and lowercase letter both exist in the table, then
1273       the uppercase letter is typically used to represent some kind of
1274       composite type (a list or a hash), and the lowercase letter is used to
1275       represent an element of that composite type.  Some internals code makes
1276       use of this case relationship.  However, 'v' and 'V' (vec and v-string)
1277       are in no way related.
1278
1279       The "PERL_MAGIC_ext" and "PERL_MAGIC_uvar" magic types are defined
1280       specifically for use by extensions and will not be used by perl itself.
1281       Extensions can use "PERL_MAGIC_ext" magic to 'attach' private
1282       information to variables (typically objects).  This is especially
1283       useful because there is no way for normal perl code to corrupt this
1284       private information (unlike using extra elements of a hash object).
1285
1286       Similarly, "PERL_MAGIC_uvar" magic can be used much like tie() to call
1287       a C function any time a scalar's value is used or changed.  The
1288       "MAGIC"'s "mg_ptr" field points to a "ufuncs" structure:
1289
1290           struct ufuncs {
1291               I32 (*uf_val)(pTHX_ IV, SV*);
1292               I32 (*uf_set)(pTHX_ IV, SV*);
1293               IV uf_index;
1294           };
1295
1296       When the SV is read from or written to, the "uf_val" or "uf_set"
1297       function will be called with "uf_index" as the first arg and a pointer
1298       to the SV as the second.  A simple example of how to add
1299       "PERL_MAGIC_uvar" magic is shown below.  Note that the ufuncs structure
1300       is copied by sv_magic, so you can safely allocate it on the stack.
1301
1302           void
1303           Umagic(sv)
1304               SV *sv;
1305           PREINIT:
1306               struct ufuncs uf;
1307           CODE:
1308               uf.uf_val   = &my_get_fn;
1309               uf.uf_set   = &my_set_fn;
1310               uf.uf_index = 0;
1311               sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1312
1313       Attaching "PERL_MAGIC_uvar" to arrays is permissible but has no effect.
1314
1315       For hashes there is a specialized hook that gives control over hash
1316       keys (but not values).  This hook calls "PERL_MAGIC_uvar" 'get' magic
1317       if the "set" function in the "ufuncs" structure is NULL.  The hook is
1318       activated whenever the hash is accessed with a key specified as an "SV"
1319       through the functions "hv_store_ent", "hv_fetch_ent", "hv_delete_ent",
1320       and "hv_exists_ent".  Accessing the key as a string through the
1321       functions without the "..._ent" suffix circumvents the hook.  See
1322       "GUTS" in Hash::Util::FieldHash for a detailed description.
1323
1324       Note that because multiple extensions may be using "PERL_MAGIC_ext" or
1325       "PERL_MAGIC_uvar" magic, it is important for extensions to take extra
1326       care to avoid conflict.  Typically only using the magic on objects
1327       blessed into the same class as the extension is sufficient.  For
1328       "PERL_MAGIC_ext" magic, it is usually a good idea to define an
1329       "MGVTBL", even if all its fields will be 0, so that individual "MAGIC"
1330       pointers can be identified as a particular kind of magic using their
1331       magic virtual table.  "mg_findext" provides an easy way to do that:
1332
1333           STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1334
1335           MAGIC *mg;
1336           if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1337               /* this is really ours, not another module's PERL_MAGIC_ext */
1338               my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1339               ...
1340           }
1341
1342       Also note that the "sv_set*()" and "sv_cat*()" functions described
1343       earlier do not invoke 'set' magic on their targets.  This must be done
1344       by the user either by calling the "SvSETMAGIC()" macro after calling
1345       these functions, or by using one of the "sv_set*_mg()" or
1346       "sv_cat*_mg()" functions.  Similarly, generic C code must call the
1347       "SvGETMAGIC()" macro to invoke any 'get' magic if they use an SV
1348       obtained from external sources in functions that don't handle magic.
1349       See perlapi for a description of these functions.  For example, calls
1350       to the "sv_cat*()" functions typically need to be followed by
1351       "SvSETMAGIC()", but they don't need a prior "SvGETMAGIC()" since their
1352       implementation handles 'get' magic.
1353
1354   Finding Magic
1355           MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1356                                              * type */
1357
1358       This routine returns a pointer to a "MAGIC" structure stored in the SV.
1359       If the SV does not have that magical feature, "NULL" is returned.  If
1360       the SV has multiple instances of that magical feature, the first one
1361       will be returned.  "mg_findext" can be used to find a "MAGIC" structure
1362       of an SV based on both its magic type and its magic virtual table:
1363
1364           MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
1365
1366       Also, if the SV passed to "mg_find" or "mg_findext" is not of type
1367       SVt_PVMG, Perl may core dump.
1368
1369           int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1370
1371       This routine checks to see what types of magic "sv" has.  If the
1372       mg_type field is an uppercase letter, then the mg_obj is copied to
1373       "nsv", but the mg_type field is changed to be the lowercase letter.
1374
1375   Understanding the Magic of Tied Hashes and Arrays
1376       Tied hashes and arrays are magical beasts of the "PERL_MAGIC_tied"
1377       magic type.
1378
1379       WARNING: As of the 5.004 release, proper usage of the array and hash
1380       access functions requires understanding a few caveats.  Some of these
1381       caveats are actually considered bugs in the API, to be fixed in later
1382       releases, and are bracketed with [MAYCHANGE] below.  If you find
1383       yourself actually applying such information in this section, be aware
1384       that the behavior may change in the future, umm, without warning.
1385
1386       The perl tie function associates a variable with an object that
1387       implements the various GET, SET, etc methods.  To perform the
1388       equivalent of the perl tie function from an XSUB, you must mimic this
1389       behaviour.  The code below carries out the necessary steps -- firstly
1390       it creates a new hash, and then creates a second hash which it blesses
1391       into the class which will implement the tie methods.  Lastly it ties
1392       the two hashes together, and returns a reference to the new tied hash.
1393       Note that the code below does NOT call the TIEHASH method in the MyTie
1394       class - see "Calling Perl Routines from within C Programs" for details
1395       on how to do this.
1396
1397           SV*
1398           mytie()
1399           PREINIT:
1400               HV *hash;
1401               HV *stash;
1402               SV *tie;
1403           CODE:
1404               hash = newHV();
1405               tie = newRV_noinc((SV*)newHV());
1406               stash = gv_stashpv("MyTie", GV_ADD);
1407               sv_bless(tie, stash);
1408               hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1409               RETVAL = newRV_noinc(hash);
1410           OUTPUT:
1411               RETVAL
1412
1413       The "av_store" function, when given a tied array argument, merely
1414       copies the magic of the array onto the value to be "stored", using
1415       "mg_copy".  It may also return NULL, indicating that the value did not
1416       actually need to be stored in the array.  [MAYCHANGE] After a call to
1417       "av_store" on a tied array, the caller will usually need to call
1418       "mg_set(val)" to actually invoke the perl level "STORE" method on the
1419       TIEARRAY object.  If "av_store" did return NULL, a call to
1420       "SvREFCNT_dec(val)" will also be usually necessary to avoid a memory
1421       leak. [/MAYCHANGE]
1422
1423       The previous paragraph is applicable verbatim to tied hash access using
1424       the "hv_store" and "hv_store_ent" functions as well.
1425
1426       "av_fetch" and the corresponding hash functions "hv_fetch" and
1427       "hv_fetch_ent" actually return an undefined mortal value whose magic
1428       has been initialized using "mg_copy".  Note the value so returned does
1429       not need to be deallocated, as it is already mortal.  [MAYCHANGE] But
1430       you will need to call "mg_get()" on the returned value in order to
1431       actually invoke the perl level "FETCH" method on the underlying TIE
1432       object.  Similarly, you may also call "mg_set()" on the return value
1433       after possibly assigning a suitable value to it using "sv_setsv",
1434       which will invoke the "STORE" method on the TIE object. [/MAYCHANGE]
1435
1436       [MAYCHANGE] In other words, the array or hash fetch/store functions
1437       don't really fetch and store actual values in the case of tied arrays
1438       and hashes.  They merely call "mg_copy" to attach magic to the values
1439       that were meant to be "stored" or "fetched".  Later calls to "mg_get"
1440       and "mg_set" actually do the job of invoking the TIE methods on the
1441       underlying objects.  Thus the magic mechanism currently implements a
1442       kind of lazy access to arrays and hashes.
1443
1444       Currently (as of perl version 5.004), use of the hash and array access
1445       functions requires the user to be aware of whether they are operating
1446       on "normal" hashes and arrays, or on their tied variants.  The API may
1447       be changed to provide more transparent access to both tied and normal
1448       data types in future versions.  [/MAYCHANGE]
1449
1450       You would do well to understand that the TIEARRAY and TIEHASH
1451       interfaces are mere sugar to invoke some perl method calls while using
1452       the uniform hash and array syntax.  The use of this sugar imposes some
1453       overhead (typically about two to four extra opcodes per FETCH/STORE
1454       operation, in addition to the creation of all the mortal variables
1455       required to invoke the methods).  This overhead will be comparatively
1456       small if the TIE methods are themselves substantial, but if they are
1457       only a few statements long, the overhead will not be insignificant.
1458
1459   Localizing changes
1460       Perl has a very handy construction
1461
1462         {
1463           local $var = 2;
1464           ...
1465         }
1466
1467       This construction is approximately equivalent to
1468
1469         {
1470           my $oldvar = $var;
1471           $var = 2;
1472           ...
1473           $var = $oldvar;
1474         }
1475
1476       The biggest difference is that the first construction would reinstate
1477       the initial value of $var, irrespective of how control exits the block:
1478       "goto", "return", "die"/"eval", etc.  It is a little bit more efficient
1479       as well.
1480
1481       There is a way to achieve a similar task from C via Perl API: create a
1482       pseudo-block, and arrange for some changes to be automatically undone
1483       at the end of it, either explicit, or via a non-local exit (via die()).
1484       A block-like construct is created by a pair of "ENTER"/"LEAVE" macros
1485       (see "Returning a Scalar" in perlcall).  Such a construct may be
1486       created specially for some important localized task, or an existing one
1487       (like boundaries of enclosing Perl subroutine/block, or an existing
1488       pair for freeing TMPs) may be used.  (In the second case the overhead
1489       of additional localization must be almost negligible.)  Note that any
1490       XSUB is automatically enclosed in an "ENTER"/"LEAVE" pair.
1491
1492       Inside such a pseudo-block the following service is available:
1493
1494       "SAVEINT(int i)"
1495       "SAVEIV(IV i)"
1496       "SAVEI32(I32 i)"
1497       "SAVELONG(long i)"
1498           These macros arrange things to restore the value of integer
1499           variable "i" at the end of enclosing pseudo-block.
1500
1501       SAVESPTR(s)
1502       SAVEPPTR(p)
1503           These macros arrange things to restore the value of pointers "s"
1504           and "p".  "s" must be a pointer of a type which survives conversion
1505           to "SV*" and back, "p" should be able to survive conversion to
1506           "char*" and back.
1507
1508       "SAVEFREESV(SV *sv)"
1509           The refcount of "sv" will be decremented at the end of pseudo-
1510           block.  This is similar to "sv_2mortal" in that it is also a
1511           mechanism for doing a delayed "SvREFCNT_dec".  However, while
1512           "sv_2mortal" extends the lifetime of "sv" until the beginning of
1513           the next statement, "SAVEFREESV" extends it until the end of the
1514           enclosing scope.  These lifetimes can be wildly different.
1515
1516           Also compare "SAVEMORTALIZESV".
1517
1518       "SAVEMORTALIZESV(SV *sv)"
1519           Just like "SAVEFREESV", but mortalizes "sv" at the end of the
1520           current scope instead of decrementing its reference count.  This
1521           usually has the effect of keeping "sv" alive until the statement
1522           that called the currently live scope has finished executing.
1523
1524       "SAVEFREEOP(OP *op)"
1525           The "OP *" is op_free()ed at the end of pseudo-block.
1526
1527       SAVEFREEPV(p)
1528           The chunk of memory which is pointed to by "p" is Safefree()ed at
1529           the end of pseudo-block.
1530
1531       "SAVECLEARSV(SV *sv)"
1532           Clears a slot in the current scratchpad which corresponds to "sv"
1533           at the end of pseudo-block.
1534
1535       "SAVEDELETE(HV *hv, char *key, I32 length)"
1536           The key "key" of "hv" is deleted at the end of pseudo-block.  The
1537           string pointed to by "key" is Safefree()ed.  If one has a key in
1538           short-lived storage, the corresponding string may be reallocated
1539           like this:
1540
1541             SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1542
1543       "SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)"
1544           At the end of pseudo-block the function "f" is called with the only
1545           argument "p".
1546
1547       "SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)"
1548           At the end of pseudo-block the function "f" is called with the
1549           implicit context argument (if any), and "p".
1550
1551       "SAVESTACK_POS()"
1552           The current offset on the Perl internal stack (cf. "SP") is
1553           restored at the end of pseudo-block.
1554
1555       The following API list contains functions, thus one needs to provide
1556       pointers to the modifiable data explicitly (either C pointers, or
1557       Perlish "GV *"s).  Where the above macros take "int", a similar
1558       function takes "int *".
1559
1560       "SV* save_scalar(GV *gv)"
1561           Equivalent to Perl code "local $gv".
1562
1563       "AV* save_ary(GV *gv)"
1564       "HV* save_hash(GV *gv)"
1565           Similar to "save_scalar", but localize @gv and %gv.
1566
1567       "void save_item(SV *item)"
1568           Duplicates the current value of "SV", on the exit from the current
1569           "ENTER"/"LEAVE" pseudo-block will restore the value of "SV" using
1570           the stored value.  It doesn't handle magic.  Use "save_scalar" if
1571           magic is affected.
1572
1573       "void save_list(SV **sarg, I32 maxsarg)"
1574           A variant of "save_item" which takes multiple arguments via an
1575           array "sarg" of "SV*" of length "maxsarg".
1576
1577       "SV* save_svref(SV **sptr)"
1578           Similar to "save_scalar", but will reinstate an "SV *".
1579
1580       "void save_aptr(AV **aptr)"
1581       "void save_hptr(HV **hptr)"
1582           Similar to "save_svref", but localize "AV *" and "HV *".
1583
1584       The "Alias" module implements localization of the basic types within
1585       the caller's scope.  People who are interested in how to localize
1586       things in the containing scope should take a look there too.
1587

Subroutines

1589   XSUBs and the Argument Stack
1590       The XSUB mechanism is a simple way for Perl programs to access C
1591       subroutines.  An XSUB routine will have a stack that contains the
1592       arguments from the Perl program, and a way to map from the Perl data
1593       structures to a C equivalent.
1594
1595       The stack arguments are accessible through the ST(n) macro, which
1596       returns the "n"'th stack argument.  Argument 0 is the first argument
1597       passed in the Perl subroutine call.  These arguments are "SV*", and can
1598       be used anywhere an "SV*" is used.
1599
1600       Most of the time, output from the C routine can be handled through use
1601       of the RETVAL and OUTPUT directives.  However, there are some cases
1602       where the argument stack is not already long enough to handle all the
1603       return values.  An example is the POSIX tzname() call, which takes no
1604       arguments, but returns two, the local time zone's standard and summer
1605       time abbreviations.
1606
1607       To handle this situation, the PPCODE directive is used and the stack is
1608       extended using the macro:
1609
1610           EXTEND(SP, num);
1611
1612       where "SP" is the macro that represents the local copy of the stack
1613       pointer, and "num" is the number of elements the stack should be
1614       extended by.
1615
1616       Now that there is room on the stack, values can be pushed on it using
1617       "PUSHs" macro.  The pushed values will often need to be "mortal" (See
1618       "Reference Counts and Mortality"):
1619
1620           PUSHs(sv_2mortal(newSViv(an_integer)))
1621           PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1622           PUSHs(sv_2mortal(newSVnv(a_double)))
1623           PUSHs(sv_2mortal(newSVpv("Some String",0)))
1624           /* Although the last example is better written as the more
1625            * efficient: */
1626           PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
1627
1628       And now the Perl program calling "tzname", the two values will be
1629       assigned as in:
1630
1631           ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1632
1633       An alternate (and possibly simpler) method to pushing values on the
1634       stack is to use the macro:
1635
1636           XPUSHs(SV*)
1637
1638       This macro automatically adjusts the stack for you, if needed.  Thus,
1639       you do not need to call "EXTEND" to extend the stack.
1640
1641       Despite their suggestions in earlier versions of this document the
1642       macros "(X)PUSH[iunp]" are not suited to XSUBs which return multiple
1643       results.  For that, either stick to the "(X)PUSHs" macros shown above,
1644       or use the new "m(X)PUSH[iunp]" macros instead; see "Putting a C value
1645       on Perl stack".
1646
1647       For more information, consult perlxs and perlxstut.
1648
1649   Autoloading with XSUBs
1650       If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts
1651       the fully-qualified name of the autoloaded subroutine in the $AUTOLOAD
1652       variable of the XSUB's package.
1653
1654       But it also puts the same information in certain fields of the XSUB
1655       itself:
1656
1657           HV *stash           = CvSTASH(cv);
1658           const char *subname = SvPVX(cv);
1659           STRLEN name_length  = SvCUR(cv); /* in bytes */
1660           U32 is_utf8         = SvUTF8(cv);
1661
1662       "SvPVX(cv)" contains just the sub name itself, not including the
1663       package.  For an AUTOLOAD routine in UNIVERSAL or one of its
1664       superclasses, "CvSTASH(cv)" returns NULL during a method call on a
1665       nonexistent package.
1666
1667       Note: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1668       XS AUTOLOAD subs at all.  Perl 5.8.0 introduced the use of fields in
1669       the XSUB itself.  Perl 5.16.0 restored the setting of $AUTOLOAD.  If
1670       you need to support 5.8-5.14, use the XSUB's fields.
1671
1672   Calling Perl Routines from within C Programs
1673       There are four routines that can be used to call a Perl subroutine from
1674       within a C program.  These four are:
1675
1676           I32  call_sv(SV*, I32);
1677           I32  call_pv(const char*, I32);
1678           I32  call_method(const char*, I32);
1679           I32  call_argv(const char*, I32, char**);
1680
1681       The routine most often used is "call_sv".  The "SV*" argument contains
1682       either the name of the Perl subroutine to be called, or a reference to
1683       the subroutine.  The second argument consists of flags that control the
1684       context in which the subroutine is called, whether or not the
1685       subroutine is being passed arguments, how errors should be trapped, and
1686       how to treat return values.
1687
1688       All four routines return the number of arguments that the subroutine
1689       returned on the Perl stack.
1690
1691       These routines used to be called "perl_call_sv", etc., before Perl
1692       v5.6.0, but those names are now deprecated; macros of the same name are
1693       provided for compatibility.
1694
1695       When using any of these routines (except "call_argv"), the programmer
1696       must manipulate the Perl stack.  These include the following macros and
1697       functions:
1698
1699           dSP
1700           SP
1701           PUSHMARK()
1702           PUTBACK
1703           SPAGAIN
1704           ENTER
1705           SAVETMPS
1706           FREETMPS
1707           LEAVE
1708           XPUSH*()
1709           POP*()
1710
1711       For a detailed description of calling conventions from C to Perl,
1712       consult perlcall.
1713
1714   Putting a C value on Perl stack
1715       A lot of opcodes (this is an elementary operation in the internal perl
1716       stack machine) put an SV* on the stack.  However, as an optimization
1717       the corresponding SV is (usually) not recreated each time.  The opcodes
1718       reuse specially assigned SVs (targets) which are (as a corollary) not
1719       constantly freed/created.
1720
1721       Each of the targets is created only once (but see "Scratchpads and
1722       recursion" below), and when an opcode needs to put an integer, a
1723       double, or a string on stack, it just sets the corresponding parts of
1724       its target and puts the target on stack.
1725
1726       The macro to put this target on stack is "PUSHTARG", and it is directly
1727       used in some opcodes, as well as indirectly in zillions of others,
1728       which use it via "(X)PUSH[iunp]".
1729
1730       Because the target is reused, you must be careful when pushing multiple
1731       values on the stack.  The following code will not do what you think:
1732
1733           XPUSHi(10);
1734           XPUSHi(20);
1735
1736       This translates as "set "TARG" to 10, push a pointer to "TARG" onto the
1737       stack; set "TARG" to 20, push a pointer to "TARG" onto the stack".  At
1738       the end of the operation, the stack does not contain the values 10 and
1739       20, but actually contains two pointers to "TARG", which we have set to
1740       20.
1741
1742       If you need to push multiple different values then you should either
1743       use the "(X)PUSHs" macros, or else use the new "m(X)PUSH[iunp]" macros,
1744       none of which make use of "TARG".  The "(X)PUSHs" macros simply push an
1745       SV* on the stack, which, as noted under "XSUBs and the Argument Stack",
1746       will often need to be "mortal".  The new "m(X)PUSH[iunp]" macros make
1747       this a little easier to achieve by creating a new mortal for you (via
1748       "(X)PUSHmortal"), pushing that onto the stack (extending it if
1749       necessary in the case of the "mXPUSH[iunp]" macros), and then setting
1750       its value.  Thus, instead of writing this to "fix" the example above:
1751
1752           XPUSHs(sv_2mortal(newSViv(10)))
1753           XPUSHs(sv_2mortal(newSViv(20)))
1754
1755       you can simply write:
1756
1757           mXPUSHi(10)
1758           mXPUSHi(20)
1759
1760       On a related note, if you do use "(X)PUSH[iunp]", then you're going to
1761       need a "dTARG" in your variable declarations so that the "*PUSH*"
1762       macros can make use of the local variable "TARG".  See also "dTARGET"
1763       and "dXSTARG".
1764
1765   Scratchpads
1766       The question remains on when the SVs which are targets for opcodes are
1767       created.  The answer is that they are created when the current unit--a
1768       subroutine or a file (for opcodes for statements outside of
1769       subroutines)--is compiled.  During this time a special anonymous Perl
1770       array is created, which is called a scratchpad for the current unit.
1771
1772       A scratchpad keeps SVs which are lexicals for the current unit and are
1773       targets for opcodes.  A previous version of this document stated that
1774       one can deduce that an SV lives on a scratchpad by looking on its
1775       flags: lexicals have "SVs_PADMY" set, and targets have "SVs_PADTMP"
1776       set.  But this has never been fully true.  "SVs_PADMY" could be set on
1777       a variable that no longer resides in any pad.  While targets do have
1778       "SVs_PADTMP" set, it can also be set on variables that have never
1779       resided in a pad, but nonetheless act like targets.  As of perl 5.21.5,
1780       the "SVs_PADMY" flag is no longer used and is defined as 0.
1781       "SvPADMY()" now returns true for anything without "SVs_PADTMP".
1782
1783       The correspondence between OPs and targets is not 1-to-1.  Different
1784       OPs in the compile tree of the unit can use the same target, if this
1785       would not conflict with the expected life of the temporary.
1786
1787   Scratchpads and recursion
1788       In fact it is not 100% true that a compiled unit contains a pointer to
1789       the scratchpad AV.  In fact it contains a pointer to an AV of
1790       (initially) one element, and this element is the scratchpad AV.  Why do
1791       we need an extra level of indirection?
1792
1793       The answer is recursion, and maybe threads.  Both these can create
1794       several execution pointers going into the same subroutine.  For the
1795       subroutine-child not write over the temporaries for the subroutine-
1796       parent (lifespan of which covers the call to the child), the parent and
1797       the child should have different scratchpads.  (And the lexicals should
1798       be separate anyway!)
1799
1800       So each subroutine is born with an array of scratchpads (of length 1).
1801       On each entry to the subroutine it is checked that the current depth of
1802       the recursion is not more than the length of this array, and if it is,
1803       new scratchpad is created and pushed into the array.
1804
1805       The targets on this scratchpad are "undef"s, but they are already
1806       marked with correct flags.
1807

Memory Allocation

1809   Allocation
1810       All memory meant to be used with the Perl API functions should be
1811       manipulated using the macros described in this section.  The macros
1812       provide the necessary transparency between differences in the actual
1813       malloc implementation that is used within perl.
1814
1815       It is suggested that you enable the version of malloc that is
1816       distributed with Perl.  It keeps pools of various sizes of unallocated
1817       memory in order to satisfy allocation requests more quickly.  However,
1818       on some platforms, it may cause spurious malloc or free errors.
1819
1820       The following three macros are used to initially allocate memory :
1821
1822           Newx(pointer, number, type);
1823           Newxc(pointer, number, type, cast);
1824           Newxz(pointer, number, type);
1825
1826       The first argument "pointer" should be the name of a variable that will
1827       point to the newly allocated memory.
1828
1829       The second and third arguments "number" and "type" specify how many of
1830       the specified type of data structure should be allocated.  The argument
1831       "type" is passed to "sizeof".  The final argument to "Newxc", "cast",
1832       should be used if the "pointer" argument is different from the "type"
1833       argument.
1834
1835       Unlike the "Newx" and "Newxc" macros, the "Newxz" macro calls "memzero"
1836       to zero out all the newly allocated memory.
1837
1838   Reallocation
1839           Renew(pointer, number, type);
1840           Renewc(pointer, number, type, cast);
1841           Safefree(pointer)
1842
1843       These three macros are used to change a memory buffer size or to free a
1844       piece of memory no longer needed.  The arguments to "Renew" and
1845       "Renewc" match those of "New" and "Newc" with the exception of not
1846       needing the "magic cookie" argument.
1847
1848   Moving
1849           Move(source, dest, number, type);
1850           Copy(source, dest, number, type);
1851           Zero(dest, number, type);
1852
1853       These three macros are used to move, copy, or zero out previously
1854       allocated memory.  The "source" and "dest" arguments point to the
1855       source and destination starting points.  Perl will move, copy, or zero
1856       out "number" instances of the size of the "type" data structure (using
1857       the "sizeof" function).
1858

PerlIO

1860       The most recent development releases of Perl have been experimenting
1861       with removing Perl's dependency on the "normal" standard I/O suite and
1862       allowing other stdio implementations to be used.  This involves
1863       creating a new abstraction layer that then calls whichever
1864       implementation of stdio Perl was compiled with.  All XSUBs should now
1865       use the functions in the PerlIO abstraction layer and not make any
1866       assumptions about what kind of stdio is being used.
1867
1868       For a complete description of the PerlIO abstraction, consult perlapio.
1869

Compiled code

1871   Code tree
1872       Here we describe the internal form your code is converted to by Perl.
1873       Start with a simple example:
1874
1875         $a = $b + $c;
1876
1877       This is converted to a tree similar to this one:
1878
1879                    assign-to
1880                  /           \
1881                 +             $a
1882               /   \
1883             $b     $c
1884
1885       (but slightly more complicated).  This tree reflects the way Perl
1886       parsed your code, but has nothing to do with the execution order.
1887       There is an additional "thread" going through the nodes of the tree
1888       which shows the order of execution of the nodes.  In our simplified
1889       example above it looks like:
1890
1891            $b ---> $c ---> + ---> $a ---> assign-to
1892
1893       But with the actual compile tree for "$a = $b + $c" it is different:
1894       some nodes optimized away.  As a corollary, though the actual tree
1895       contains more nodes than our simplified example, the execution order is
1896       the same as in our example.
1897
1898   Examining the tree
1899       If you have your perl compiled for debugging (usually done with
1900       "-DDEBUGGING" on the "Configure" command line), you may examine the
1901       compiled tree by specifying "-Dx" on the Perl command line.  The output
1902       takes several lines per node, and for "$b+$c" it looks like this:
1903
1904           5           TYPE = add  ===> 6
1905                       TARG = 1
1906                       FLAGS = (SCALAR,KIDS)
1907                       {
1908                           TYPE = null  ===> (4)
1909                             (was rv2sv)
1910                           FLAGS = (SCALAR,KIDS)
1911                           {
1912           3                   TYPE = gvsv  ===> 4
1913                               FLAGS = (SCALAR)
1914                               GV = main::b
1915                           }
1916                       }
1917                       {
1918                           TYPE = null  ===> (5)
1919                             (was rv2sv)
1920                           FLAGS = (SCALAR,KIDS)
1921                           {
1922           4                   TYPE = gvsv  ===> 5
1923                               FLAGS = (SCALAR)
1924                               GV = main::c
1925                           }
1926                       }
1927
1928       This tree has 5 nodes (one per "TYPE" specifier), only 3 of them are
1929       not optimized away (one per number in the left column).  The immediate
1930       children of the given node correspond to "{}" pairs on the same level
1931       of indentation, thus this listing corresponds to the tree:
1932
1933                          add
1934                        /     \
1935                      null    null
1936                       |       |
1937                      gvsv    gvsv
1938
1939       The execution order is indicated by "===>" marks, thus it is "3 4 5 6"
1940       (node 6 is not included into above listing), i.e., "gvsv gvsv add
1941       whatever".
1942
1943       Each of these nodes represents an op, a fundamental operation inside
1944       the Perl core.  The code which implements each operation can be found
1945       in the pp*.c files; the function which implements the op with type
1946       "gvsv" is "pp_gvsv", and so on.  As the tree above shows, different ops
1947       have different numbers of children: "add" is a binary operator, as one
1948       would expect, and so has two children.  To accommodate the various
1949       different numbers of children, there are various types of op data
1950       structure, and they link together in different ways.
1951
1952       The simplest type of op structure is "OP": this has no children.  Unary
1953       operators, "UNOP"s, have one child, and this is pointed to by the
1954       "op_first" field.  Binary operators ("BINOP"s) have not only an
1955       "op_first" field but also an "op_last" field.  The most complex type of
1956       op is a "LISTOP", which has any number of children.  In this case, the
1957       first child is pointed to by "op_first" and the last child by
1958       "op_last".  The children in between can be found by iteratively
1959       following the "OpSIBLING" pointer from the first child to the last (but
1960       see below).
1961
1962       There are also some other op types: a "PMOP" holds a regular
1963       expression, and has no children, and a "LOOP" may or may not have
1964       children.  If the "op_children" field is non-zero, it behaves like a
1965       "LISTOP".  To complicate matters, if a "UNOP" is actually a "null" op
1966       after optimization (see "Compile pass 2: context propagation") it will
1967       still have children in accordance with its former type.
1968
1969       Finally, there is a "LOGOP", or logic op. Like a "LISTOP", this has one
1970       or more children, but it doesn't have an "op_last" field: so you have
1971       to follow "op_first" and then the "OpSIBLING" chain itself to find the
1972       last child. Instead it has an "op_other" field, which is comparable to
1973       the "op_next" field described below, and represents an alternate
1974       execution path. Operators like "and", "or" and "?" are "LOGOP"s. Note
1975       that in general, "op_other" may not point to any of the direct children
1976       of the "LOGOP".
1977
1978       Starting in version 5.21.2, perls built with the experimental define
1979       "-DPERL_OP_PARENT" add an extra boolean flag for each op, "op_moresib".
1980       When not set, this indicates that this is the last op in an "OpSIBLING"
1981       chain. This frees up the "op_sibling" field on the last sibling to
1982       point back to the parent op. Under this build, that field is also
1983       renamed "op_sibparent" to reflect its joint role. The macro
1984       OpSIBLING(o) wraps this special behaviour, and always returns NULL on
1985       the last sibling.  With this build the op_parent(o) function can be
1986       used to find the parent of any op. Thus for forward compatibility, you
1987       should always use the OpSIBLING(o) macro rather than accessing
1988       "op_sibling" directly.
1989
1990       Another way to examine the tree is to use a compiler back-end module,
1991       such as B::Concise.
1992
1993   Compile pass 1: check routines
1994       The tree is created by the compiler while yacc code feeds it the
1995       constructions it recognizes.  Since yacc works bottom-up, so does the
1996       first pass of perl compilation.
1997
1998       What makes this pass interesting for perl developers is that some
1999       optimization may be performed on this pass.  This is optimization by
2000       so-called "check routines".  The correspondence between node names and
2001       corresponding check routines is described in opcode.pl (do not forget
2002       to run "make regen_headers" if you modify this file).
2003
2004       A check routine is called when the node is fully constructed except for
2005       the execution-order thread.  Since at this time there are no back-links
2006       to the currently constructed node, one can do most any operation to the
2007       top-level node, including freeing it and/or creating new nodes
2008       above/below it.
2009
2010       The check routine returns the node which should be inserted into the
2011       tree (if the top-level node was not modified, check routine returns its
2012       argument).
2013
2014       By convention, check routines have names "ck_*".  They are usually
2015       called from "new*OP" subroutines (or "convert") (which in turn are
2016       called from perly.y).
2017
2018   Compile pass 1a: constant folding
2019       Immediately after the check routine is called the returned node is
2020       checked for being compile-time executable.  If it is (the value is
2021       judged to be constant) it is immediately executed, and a constant node
2022       with the "return value" of the corresponding subtree is substituted
2023       instead.  The subtree is deleted.
2024
2025       If constant folding was not performed, the execution-order thread is
2026       created.
2027
2028   Compile pass 2: context propagation
2029       When a context for a part of compile tree is known, it is propagated
2030       down through the tree.  At this time the context can have 5 values
2031       (instead of 2 for runtime context): void, boolean, scalar, list, and
2032       lvalue.  In contrast with the pass 1 this pass is processed from top to
2033       bottom: a node's context determines the context for its children.
2034
2035       Additional context-dependent optimizations are performed at this time.
2036       Since at this moment the compile tree contains back-references (via
2037       "thread" pointers), nodes cannot be free()d now.  To allow optimized-
2038       away nodes at this stage, such nodes are null()ified instead of
2039       free()ing (i.e. their type is changed to OP_NULL).
2040
2041   Compile pass 3: peephole optimization
2042       After the compile tree for a subroutine (or for an "eval" or a file) is
2043       created, an additional pass over the code is performed.  This pass is
2044       neither top-down or bottom-up, but in the execution order (with
2045       additional complications for conditionals).  Optimizations performed at
2046       this stage are subject to the same restrictions as in the pass 2.
2047
2048       Peephole optimizations are done by calling the function pointed to by
2049       the global variable "PL_peepp".  By default, "PL_peepp" just calls the
2050       function pointed to by the global variable "PL_rpeepp".  By default,
2051       that performs some basic op fixups and optimisations along the
2052       execution-order op chain, and recursively calls "PL_rpeepp" for each
2053       side chain of ops (resulting from conditionals).  Extensions may
2054       provide additional optimisations or fixups, hooking into either the
2055       per-subroutine or recursive stage, like this:
2056
2057           static peep_t prev_peepp;
2058           static void my_peep(pTHX_ OP *o)
2059           {
2060               /* custom per-subroutine optimisation goes here */
2061               prev_peepp(aTHX_ o);
2062               /* custom per-subroutine optimisation may also go here */
2063           }
2064           BOOT:
2065               prev_peepp = PL_peepp;
2066               PL_peepp = my_peep;
2067
2068           static peep_t prev_rpeepp;
2069           static void my_rpeep(pTHX_ OP *o)
2070           {
2071               OP *orig_o = o;
2072               for(; o; o = o->op_next) {
2073                   /* custom per-op optimisation goes here */
2074               }
2075               prev_rpeepp(aTHX_ orig_o);
2076           }
2077           BOOT:
2078               prev_rpeepp = PL_rpeepp;
2079               PL_rpeepp = my_rpeep;
2080
2081   Pluggable runops
2082       The compile tree is executed in a runops function.  There are two
2083       runops functions, in run.c and in dump.c.  "Perl_runops_debug" is used
2084       with DEBUGGING and "Perl_runops_standard" is used otherwise.  For fine
2085       control over the execution of the compile tree it is possible to
2086       provide your own runops function.
2087
2088       It's probably best to copy one of the existing runops functions and
2089       change it to suit your needs.  Then, in the BOOT section of your XS
2090       file, add the line:
2091
2092         PL_runops = my_runops;
2093
2094       This function should be as efficient as possible to keep your programs
2095       running as fast as possible.
2096
2097   Compile-time scope hooks
2098       As of perl 5.14 it is possible to hook into the compile-time lexical
2099       scope mechanism using "Perl_blockhook_register".  This is used like
2100       this:
2101
2102           STATIC void my_start_hook(pTHX_ int full);
2103           STATIC BHK my_hooks;
2104
2105           BOOT:
2106               BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
2107               Perl_blockhook_register(aTHX_ &my_hooks);
2108
2109       This will arrange to have "my_start_hook" called at the start of
2110       compiling every lexical scope.  The available hooks are:
2111
2112       "void bhk_start(pTHX_ int full)"
2113           This is called just after starting a new lexical scope.  Note that
2114           Perl code like
2115
2116               if ($x) { ... }
2117
2118           creates two scopes: the first starts at the "(" and has "full ==
2119           1", the second starts at the "{" and has "full == 0".  Both end at
2120           the "}", so calls to "start" and "pre"/"post_end" will match.
2121           Anything pushed onto the save stack by this hook will be popped
2122           just before the scope ends (between the "pre_" and "post_end"
2123           hooks, in fact).
2124
2125       "void bhk_pre_end(pTHX_ OP **o)"
2126           This is called at the end of a lexical scope, just before unwinding
2127           the stack.  o is the root of the optree representing the scope; it
2128           is a double pointer so you can replace the OP if you need to.
2129
2130       "void bhk_post_end(pTHX_ OP **o)"
2131           This is called at the end of a lexical scope, just after unwinding
2132           the stack.  o is as above.  Note that it is possible for calls to
2133           "pre_" and "post_end" to nest, if there is something on the save
2134           stack that calls string eval.
2135
2136       "void bhk_eval(pTHX_ OP *const o)"
2137           This is called just before starting to compile an "eval STRING",
2138           "do FILE", "require" or "use", after the eval has been set up.  o
2139           is the OP that requested the eval, and will normally be an
2140           "OP_ENTEREVAL", "OP_DOFILE" or "OP_REQUIRE".
2141
2142       Once you have your hook functions, you need a "BHK" structure to put
2143       them in.  It's best to allocate it statically, since there is no way to
2144       free it once it's registered.  The function pointers should be inserted
2145       into this structure using the "BhkENTRY_set" macro, which will also set
2146       flags indicating which entries are valid.  If you do need to allocate
2147       your "BHK" dynamically for some reason, be sure to zero it before you
2148       start.
2149
2150       Once registered, there is no mechanism to switch these hooks off, so if
2151       that is necessary you will need to do this yourself.  An entry in "%^H"
2152       is probably the best way, so the effect is lexically scoped; however it
2153       is also possible to use the "BhkDISABLE" and "BhkENABLE" macros to
2154       temporarily switch entries on and off.  You should also be aware that
2155       generally speaking at least one scope will have opened before your
2156       extension is loaded, so you will see some "pre"/"post_end" pairs that
2157       didn't have a matching "start".
2158

Examining internal data structures with the "dump" functions

2160       To aid debugging, the source file dump.c contains a number of functions
2161       which produce formatted output of internal data structures.
2162
2163       The most commonly used of these functions is "Perl_sv_dump"; it's used
2164       for dumping SVs, AVs, HVs, and CVs.  The "Devel::Peek" module calls
2165       "sv_dump" to produce debugging output from Perl-space, so users of that
2166       module should already be familiar with its format.
2167
2168       "Perl_op_dump" can be used to dump an "OP" structure or any of its
2169       derivatives, and produces output similar to "perl -Dx"; in fact,
2170       "Perl_dump_eval" will dump the main root of the code being evaluated,
2171       exactly like "-Dx".
2172
2173       Other useful functions are "Perl_dump_sub", which turns a "GV" into an
2174       op tree, "Perl_dump_packsubs" which calls "Perl_dump_sub" on all the
2175       subroutines in a package like so: (Thankfully, these are all xsubs, so
2176       there is no op tree)
2177
2178           (gdb) print Perl_dump_packsubs(PL_defstash)
2179
2180           SUB attributes::bootstrap = (xsub 0x811fedc 0)
2181
2182           SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2183
2184           SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2185
2186           SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2187
2188           SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2189
2190       and "Perl_dump_all", which dumps all the subroutines in the stash and
2191       the op tree of the main root.
2192

How multiple interpreters and concurrency are supported

2194   Background and PERL_IMPLICIT_CONTEXT
2195       The Perl interpreter can be regarded as a closed box: it has an API for
2196       feeding it code or otherwise making it do things, but it also has
2197       functions for its own use.  This smells a lot like an object, and there
2198       are ways for you to build Perl so that you can have multiple
2199       interpreters, with one interpreter represented either as a C structure,
2200       or inside a thread-specific structure.  These structures contain all
2201       the context, the state of that interpreter.
2202
2203       One macro controls the major Perl build flavor: MULTIPLICITY.  The
2204       MULTIPLICITY build has a C structure that packages all the interpreter
2205       state.  With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2206       normally defined, and enables the support for passing in a "hidden"
2207       first argument that represents all three data structures.  MULTIPLICITY
2208       makes multi-threaded perls possible (with the ithreads threading model,
2209       related to the macro USE_ITHREADS.)
2210
2211       Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2212       PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2213       former turns on MULTIPLICITY.)  The PERL_GLOBAL_STRUCT causes all the
2214       internal variables of Perl to be wrapped inside a single global struct,
2215       struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or the
2216       function  Perl_GetVars().  The PERL_GLOBAL_STRUCT_PRIVATE goes one step
2217       further, there is still a single struct (allocated in main() either
2218       from heap or from stack) but there are no global data symbols pointing
2219       to it.  In either case the global struct should be initialized as the
2220       very first thing in main() using Perl_init_global_struct() and
2221       correspondingly tear it down after perl_free() using
2222       Perl_free_global_struct(), please see miniperlmain.c for usage details.
2223       You may also need to use "dVAR" in your coding to "declare the global
2224       variables" when you are using them.  dTHX does this for you
2225       automatically.
2226
2227       To see whether you have non-const data you can use a BSD (or GNU)
2228       compatible "nm":
2229
2230         nm libperl.a | grep -v ' [TURtr] '
2231
2232       If this displays any "D" or "d" symbols (or possibly "C" or "c"), you
2233       have non-const data.  The symbols the "grep" removed are as follows:
2234       "Tt" are text, or code, the "Rr" are read-only (const) data, and the
2235       "U" is <undefined>, external symbols referred to.
2236
2237       The test t/porting/libperl.t does this kind of symbol sanity checking
2238       on "libperl.a".
2239
2240       For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2241       doesn't actually hide all symbols inside a big global struct: some
2242       PerlIO_xxx vtables are left visible.  The PERL_GLOBAL_STRUCT_PRIVATE
2243       then hides everything (see how the PERLIO_FUNCS_DECL is used).
2244
2245       All this obviously requires a way for the Perl internal functions to be
2246       either subroutines taking some kind of structure as the first argument,
2247       or subroutines taking nothing as the first argument.  To enable these
2248       two very different ways of building the interpreter, the Perl source
2249       (as it does in so many other situations) makes heavy use of macros and
2250       subroutine naming conventions.
2251
2252       First problem: deciding which functions will be public API functions
2253       and which will be private.  All functions whose names begin "S_" are
2254       private (think "S" for "secret" or "static").  All other functions
2255       begin with "Perl_", but just because a function begins with "Perl_"
2256       does not mean it is part of the API.  (See "Internal Functions".)  The
2257       easiest way to be sure a function is part of the API is to find its
2258       entry in perlapi.  If it exists in perlapi, it's part of the API.  If
2259       it doesn't, and you think it should be (i.e., you need it for your
2260       extension), send mail via perlbug explaining why you think it should
2261       be.
2262
2263       Second problem: there must be a syntax so that the same subroutine
2264       declarations and calls can pass a structure as their first argument, or
2265       pass nothing.  To solve this, the subroutines are named and declared in
2266       a particular way.  Here's a typical start of a static function used
2267       within the Perl guts:
2268
2269         STATIC void
2270         S_incline(pTHX_ char *s)
2271
2272       STATIC becomes "static" in C, and may be #define'd to nothing in some
2273       configurations in the future.
2274
2275       A public function (i.e. part of the internal API, but not necessarily
2276       sanctioned for use in extensions) begins like this:
2277
2278         void
2279         Perl_sv_setiv(pTHX_ SV* dsv, IV num)
2280
2281       "pTHX_" is one of a number of macros (in perl.h) that hide the details
2282       of the interpreter's context.  THX stands for "thread", "this", or
2283       "thingy", as the case may be.  (And no, George Lucas is not involved.
2284       :-) The first character could be 'p' for a prototype, 'a' for argument,
2285       or 'd' for declaration, so we have "pTHX", "aTHX" and "dTHX", and their
2286       variants.
2287
2288       When Perl is built without options that set PERL_IMPLICIT_CONTEXT,
2289       there is no first argument containing the interpreter's context.  The
2290       trailing underscore in the pTHX_ macro indicates that the macro
2291       expansion needs a comma after the context argument because other
2292       arguments follow it.  If PERL_IMPLICIT_CONTEXT is not defined, pTHX_
2293       will be ignored, and the subroutine is not prototyped to take the extra
2294       argument.  The form of the macro without the trailing underscore is
2295       used when there are no additional explicit arguments.
2296
2297       When a core function calls another, it must pass the context.  This is
2298       normally hidden via macros.  Consider "sv_setiv".  It expands into
2299       something like this:
2300
2301           #ifdef PERL_IMPLICIT_CONTEXT
2302             #define sv_setiv(a,b)      Perl_sv_setiv(aTHX_ a, b)
2303             /* can't do this for vararg functions, see below */
2304           #else
2305             #define sv_setiv           Perl_sv_setiv
2306           #endif
2307
2308       This works well, and means that XS authors can gleefully write:
2309
2310           sv_setiv(foo, bar);
2311
2312       and still have it work under all the modes Perl could have been
2313       compiled with.
2314
2315       This doesn't work so cleanly for varargs functions, though, as macros
2316       imply that the number of arguments is known in advance.  Instead we
2317       either need to spell them out fully, passing "aTHX_" as the first
2318       argument (the Perl core tends to do this with functions like
2319       Perl_warner), or use a context-free version.
2320
2321       The context-free version of Perl_warner is called
2322       Perl_warner_nocontext, and does not take the extra argument.  Instead
2323       it does dTHX; to get the context from thread-local storage.  We
2324       "#define warner Perl_warner_nocontext" so that extensions get source
2325       compatibility at the expense of performance.  (Passing an arg is
2326       cheaper than grabbing it from thread-local storage.)
2327
2328       You can ignore [pad]THXx when browsing the Perl headers/sources.  Those
2329       are strictly for use within the core.  Extensions and embedders need
2330       only be aware of [pad]THX.
2331
2332   So what happened to dTHR?
2333       "dTHR" was introduced in perl 5.005 to support the older thread model.
2334       The older thread model now uses the "THX" mechanism to pass context
2335       pointers around, so "dTHR" is not useful any more.  Perl 5.6.0 and
2336       later still have it for backward source compatibility, but it is
2337       defined to be a no-op.
2338
2339   How do I use all this in extensions?
2340       When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call any
2341       functions in the Perl API will need to pass the initial context
2342       argument somehow.  The kicker is that you will need to write it in such
2343       a way that the extension still compiles when Perl hasn't been built
2344       with PERL_IMPLICIT_CONTEXT enabled.
2345
2346       There are three ways to do this.  First, the easy but inefficient way,
2347       which is also the default, in order to maintain source compatibility
2348       with extensions: whenever XSUB.h is #included, it redefines the aTHX
2349       and aTHX_ macros to call a function that will return the context.
2350       Thus, something like:
2351
2352               sv_setiv(sv, num);
2353
2354       in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
2355       in effect:
2356
2357               Perl_sv_setiv(Perl_get_context(), sv, num);
2358
2359       or to this otherwise:
2360
2361               Perl_sv_setiv(sv, num);
2362
2363       You don't have to do anything new in your extension to get this; since
2364       the Perl library provides Perl_get_context(), it will all just work.
2365
2366       The second, more efficient way is to use the following template for
2367       your Foo.xs:
2368
2369               #define PERL_NO_GET_CONTEXT     /* we want efficiency */
2370               #include "EXTERN.h"
2371               #include "perl.h"
2372               #include "XSUB.h"
2373
2374               STATIC void my_private_function(int arg1, int arg2);
2375
2376               STATIC void
2377               my_private_function(int arg1, int arg2)
2378               {
2379                   dTHX;       /* fetch context */
2380                   ... call many Perl API functions ...
2381               }
2382
2383               [... etc ...]
2384
2385               MODULE = Foo            PACKAGE = Foo
2386
2387               /* typical XSUB */
2388
2389               void
2390               my_xsub(arg)
2391                       int arg
2392                   CODE:
2393                       my_private_function(arg, 10);
2394
2395       Note that the only two changes from the normal way of writing an
2396       extension is the addition of a "#define PERL_NO_GET_CONTEXT" before
2397       including the Perl headers, followed by a "dTHX;" declaration at the
2398       start of every function that will call the Perl API.  (You'll know
2399       which functions need this, because the C compiler will complain that
2400       there's an undeclared identifier in those functions.)  No changes are
2401       needed for the XSUBs themselves, because the XS() macro is correctly
2402       defined to pass in the implicit context if needed.
2403
2404       The third, even more efficient way is to ape how it is done within the
2405       Perl guts:
2406
2407               #define PERL_NO_GET_CONTEXT     /* we want efficiency */
2408               #include "EXTERN.h"
2409               #include "perl.h"
2410               #include "XSUB.h"
2411
2412               /* pTHX_ only needed for functions that call Perl API */
2413               STATIC void my_private_function(pTHX_ int arg1, int arg2);
2414
2415               STATIC void
2416               my_private_function(pTHX_ int arg1, int arg2)
2417               {
2418                   /* dTHX; not needed here, because THX is an argument */
2419                   ... call Perl API functions ...
2420               }
2421
2422               [... etc ...]
2423
2424               MODULE = Foo            PACKAGE = Foo
2425
2426               /* typical XSUB */
2427
2428               void
2429               my_xsub(arg)
2430                       int arg
2431                   CODE:
2432                       my_private_function(aTHX_ arg, 10);
2433
2434       This implementation never has to fetch the context using a function
2435       call, since it is always passed as an extra argument.  Depending on
2436       your needs for simplicity or efficiency, you may mix the previous two
2437       approaches freely.
2438
2439       Never add a comma after "pTHX" yourself--always use the form of the
2440       macro with the underscore for functions that take explicit arguments,
2441       or the form without the argument for functions with no explicit
2442       arguments.
2443
2444       If one is compiling Perl with the "-DPERL_GLOBAL_STRUCT" the "dVAR"
2445       definition is needed if the Perl global variables (see perlvars.h or
2446       globvar.sym) are accessed in the function and "dTHX" is not used (the
2447       "dTHX" includes the "dVAR" if necessary).  One notices the need for
2448       "dVAR" only with the said compile-time define, because otherwise the
2449       Perl global variables are visible as-is.
2450
2451   Should I do anything special if I call perl from multiple threads?
2452       If you create interpreters in one thread and then proceed to call them
2453       in another, you need to make sure perl's own Thread Local Storage (TLS)
2454       slot is initialized correctly in each of those threads.
2455
2456       The "perl_alloc" and "perl_clone" API functions will automatically set
2457       the TLS slot to the interpreter they created, so that there is no need
2458       to do anything special if the interpreter is always accessed in the
2459       same thread that created it, and that thread did not create or call any
2460       other interpreters afterwards.  If that is not the case, you have to
2461       set the TLS slot of the thread before calling any functions in the Perl
2462       API on that particular interpreter.  This is done by calling the
2463       "PERL_SET_CONTEXT" macro in that thread as the first thing you do:
2464
2465               /* do this before doing anything else with some_perl */
2466               PERL_SET_CONTEXT(some_perl);
2467
2468               ... other Perl API calls on some_perl go here ...
2469
2470   Future Plans and PERL_IMPLICIT_SYS
2471       Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2472       that the interpreter knows about itself and pass it around, so too are
2473       there plans to allow the interpreter to bundle up everything it knows
2474       about the environment it's running on.  This is enabled with the
2475       PERL_IMPLICIT_SYS macro.  Currently it only works with USE_ITHREADS on
2476       Windows.
2477
2478       This allows the ability to provide an extra pointer (called the "host"
2479       environment) for all the system calls.  This makes it possible for all
2480       the system stuff to maintain their own state, broken down into seven C
2481       structures.  These are thin wrappers around the usual system calls (see
2482       win32/perllib.c) for the default perl executable, but for a more
2483       ambitious host (like the one that would do fork() emulation) all the
2484       extra work needed to pretend that different interpreters are actually
2485       different "processes", would be done here.
2486
2487       The Perl engine/interpreter and the host are orthogonal entities.
2488       There could be one or more interpreters in a process, and one or more
2489       "hosts", with free association between them.
2490

Internal Functions

2492       All of Perl's internal functions which will be exposed to the outside
2493       world are prefixed by "Perl_" so that they will not conflict with XS
2494       functions or functions used in a program in which Perl is embedded.
2495       Similarly, all global variables begin with "PL_".  (By convention,
2496       static functions start with "S_".)
2497
2498       Inside the Perl core ("PERL_CORE" defined), you can get at the
2499       functions either with or without the "Perl_" prefix, thanks to a bunch
2500       of defines that live in embed.h.  Note that extension code should not
2501       set "PERL_CORE"; this exposes the full perl internals, and is likely to
2502       cause breakage of the XS in each new perl release.
2503
2504       The file embed.h is generated automatically from embed.pl and
2505       embed.fnc.  embed.pl also creates the prototyping header files for the
2506       internal functions, generates the documentation and a lot of other bits
2507       and pieces.  It's important that when you add a new function to the
2508       core or change an existing one, you change the data in the table in
2509       embed.fnc as well.  Here's a sample entry from that table:
2510
2511           Apd |SV**   |av_fetch   |AV* ar|I32 key|I32 lval
2512
2513       The second column is the return type, the third column the name.
2514       Columns after that are the arguments.  The first column is a set of
2515       flags:
2516
2517       A  This function is a part of the public API.  All such functions
2518          should also have 'd', very few do not.
2519
2520       p  This function has a "Perl_" prefix; i.e. it is defined as
2521          "Perl_av_fetch".
2522
2523       d  This function has documentation using the "apidoc" feature which
2524          we'll look at in a second.  Some functions have 'd' but not 'A';
2525          docs are good.
2526
2527       Other available flags are:
2528
2529       s  This is a static function and is defined as "STATIC S_whatever", and
2530          usually called within the sources as "whatever(...)".
2531
2532       n  This does not need an interpreter context, so the definition has no
2533          "pTHX", and it follows that callers don't use "aTHX".  (See
2534          "Background and PERL_IMPLICIT_CONTEXT".)
2535
2536       r  This function never returns; "croak", "exit" and friends.
2537
2538       f  This function takes a variable number of arguments, "printf" style.
2539          The argument list should end with "...", like this:
2540
2541              Afprd   |void   |croak          |const char* pat|...
2542
2543       M  This function is part of the experimental development API, and may
2544          change or disappear without notice.
2545
2546       o  This function should not have a compatibility macro to define, say,
2547          "Perl_parse" to "parse".  It must be called as "Perl_parse".
2548
2549       x  This function isn't exported out of the Perl core.
2550
2551       m  This is implemented as a macro.
2552
2553       X  This function is explicitly exported.
2554
2555       E  This function is visible to extensions included in the Perl core.
2556
2557       b  Binary backward compatibility; this function is a macro but also has
2558          a "Perl_" implementation (which is exported).
2559
2560       others
2561          See the comments at the top of "embed.fnc" for others.
2562
2563       If you edit embed.pl or embed.fnc, you will need to run "make
2564       regen_headers" to force a rebuild of embed.h and other auto-generated
2565       files.
2566
2567   Formatted Printing of IVs, UVs, and NVs
2568       If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2569       formatting codes like %d, %ld, %f, you should use the following macros
2570       for portability
2571
2572               IVdf            IV in decimal
2573               UVuf            UV in decimal
2574               UVof            UV in octal
2575               UVxf            UV in hexadecimal
2576               NVef            NV %e-like
2577               NVff            NV %f-like
2578               NVgf            NV %g-like
2579
2580       These will take care of 64-bit integers and long doubles.  For example:
2581
2582               printf("IV is %"IVdf"\n", iv);
2583
2584       The IVdf will expand to whatever is the correct format for the IVs.
2585
2586       Note that there are different "long doubles": Perl will use whatever
2587       the compiler has.
2588
2589       If you are printing addresses of pointers, use UVxf combined with
2590       PTR2UV(), do not use %lx or %p.
2591
2592   Formatted Printing of "Size_t" and "SSize_t"
2593       The most general way to do this is to cast them to a UV or IV, and
2594       print as in the previous section.
2595
2596       But if you're using "PerlIO_printf()", it's less typing and visual
2597       clutter to use the "%z" length modifier (for siZe):
2598
2599               PerlIO_printf("STRLEN is %zu\n", len);
2600
2601       This modifier is not portable, so its use should be restricted to
2602       "PerlIO_printf()".
2603
2604   Pointer-To-Integer and Integer-To-Pointer
2605       Because pointer size does not necessarily equal integer size, use the
2606       follow macros to do it right.
2607
2608               PTR2UV(pointer)
2609               PTR2IV(pointer)
2610               PTR2NV(pointer)
2611               INT2PTR(pointertotype, integer)
2612
2613       For example:
2614
2615               IV  iv = ...;
2616               SV *sv = INT2PTR(SV*, iv);
2617
2618       and
2619
2620               AV *av = ...;
2621               UV  uv = PTR2UV(av);
2622
2623   Exception Handling
2624       There are a couple of macros to do very basic exception handling in XS
2625       modules.  You have to define "NO_XSLOCKS" before including XSUB.h to be
2626       able to use these macros:
2627
2628               #define NO_XSLOCKS
2629               #include "XSUB.h"
2630
2631       You can use these macros if you call code that may croak, but you need
2632       to do some cleanup before giving control back to Perl.  For example:
2633
2634               dXCPT;    /* set up necessary variables */
2635
2636               XCPT_TRY_START {
2637                 code_that_may_croak();
2638               } XCPT_TRY_END
2639
2640               XCPT_CATCH
2641               {
2642                 /* do cleanup here */
2643                 XCPT_RETHROW;
2644               }
2645
2646       Note that you always have to rethrow an exception that has been caught.
2647       Using these macros, it is not possible to just catch the exception and
2648       ignore it.  If you have to ignore the exception, you have to use the
2649       "call_*" function.
2650
2651       The advantage of using the above macros is that you don't have to setup
2652       an extra function for "call_*", and that using these macros is faster
2653       than using "call_*".
2654
2655   Source Documentation
2656       There's an effort going on to document the internal functions and
2657       automatically produce reference manuals from them -- perlapi is one
2658       such manual which details all the functions which are available to XS
2659       writers.  perlintern is the autogenerated manual for the functions
2660       which are not part of the API and are supposedly for internal use only.
2661
2662       Source documentation is created by putting POD comments into the C
2663       source, like this:
2664
2665        /*
2666        =for apidoc sv_setiv
2667
2668        Copies an integer into the given SV.  Does not handle 'set' magic.  See
2669        L<perlapi/sv_setiv_mg>.
2670
2671        =cut
2672        */
2673
2674       Please try and supply some documentation if you add functions to the
2675       Perl core.
2676
2677   Backwards compatibility
2678       The Perl API changes over time.  New functions are added or the
2679       interfaces of existing functions are changed.  The "Devel::PPPort"
2680       module tries to provide compatibility code for some of these changes,
2681       so XS writers don't have to code it themselves when supporting multiple
2682       versions of Perl.
2683
2684       "Devel::PPPort" generates a C header file ppport.h that can also be run
2685       as a Perl script.  To generate ppport.h, run:
2686
2687           perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2688
2689       Besides checking existing XS code, the script can also be used to
2690       retrieve compatibility information for various API calls using the
2691       "--api-info" command line switch.  For example:
2692
2693         % perl ppport.h --api-info=sv_magicext
2694
2695       For details, see "perldoc ppport.h".
2696

Unicode Support

2698       Perl 5.6.0 introduced Unicode support.  It's important for porters and
2699       XS writers to understand this support and make sure that the code they
2700       write does not corrupt Unicode data.
2701
2702   What is Unicode, anyway?
2703       In the olden, less enlightened times, we all used to use ASCII.  Most
2704       of us did, anyway.  The big problem with ASCII is that it's American.
2705       Well, no, that's not actually the problem; the problem is that it's not
2706       particularly useful for people who don't use the Roman alphabet.  What
2707       used to happen was that particular languages would stick their own
2708       alphabet in the upper range of the sequence, between 128 and 255.  Of
2709       course, we then ended up with plenty of variants that weren't quite
2710       ASCII, and the whole point of it being a standard was lost.
2711
2712       Worse still, if you've got a language like Chinese or Japanese that has
2713       hundreds or thousands of characters, then you really can't fit them
2714       into a mere 256, so they had to forget about ASCII altogether, and
2715       build their own systems using pairs of numbers to refer to one
2716       character.
2717
2718       To fix this, some people formed Unicode, Inc. and produced a new
2719       character set containing all the characters you can possibly think of
2720       and more.  There are several ways of representing these characters, and
2721       the one Perl uses is called UTF-8.  UTF-8 uses a variable number of
2722       bytes to represent a character.  You can learn more about Unicode and
2723       Perl's Unicode model in perlunicode.
2724
2725       (On EBCDIC platforms, Perl uses instead UTF-EBCDIC, which is a form of
2726       UTF-8 adapted for EBCDIC platforms.  Below, we just talk about UTF-8.
2727       UTF-EBCDIC is like UTF-8, but the details are different.  The macros
2728       hide the differences from you, just remember that the particular
2729       numbers and bit patterns presented below will differ in UTF-EBCDIC.)
2730
2731   How can I recognise a UTF-8 string?
2732       You can't.  This is because UTF-8 data is stored in bytes just like
2733       non-UTF-8 data.  The Unicode character 200, (0xC8 for you hex types)
2734       capital E with a grave accent, is represented by the two bytes
2735       "v196.172".  Unfortunately, the non-Unicode string "chr(196).chr(172)"
2736       has that byte sequence as well.  So you can't tell just by looking --
2737       this is what makes Unicode input an interesting problem.
2738
2739       In general, you either have to know what you're dealing with, or you
2740       have to guess.  The API function "is_utf8_string" can help; it'll tell
2741       you if a string contains only valid UTF-8 characters, and the chances
2742       of a non-UTF-8 string looking like valid UTF-8 become very small very
2743       quickly with increasing string length.  On a character-by-character
2744       basis, "isUTF8_CHAR" will tell you whether the current character in a
2745       string is valid UTF-8.
2746
2747   How does UTF-8 represent Unicode characters?
2748       As mentioned above, UTF-8 uses a variable number of bytes to store a
2749       character.  Characters with values 0...127 are stored in one byte, just
2750       like good ol' ASCII.  Character 128 is stored as "v194.128"; this
2751       continues up to character 191, which is "v194.191".  Now we've run out
2752       of bits (191 is binary 10111111) so we move on; character 192 is
2753       "v195.128".  And so it goes on, moving to three bytes at character
2754       2048.  "Unicode Encodings" in perlunicode has pictures of how this
2755       works.
2756
2757       Assuming you know you're dealing with a UTF-8 string, you can find out
2758       how long the first character in it is with the "UTF8SKIP" macro:
2759
2760           char *utf = "\305\233\340\240\201";
2761           I32 len;
2762
2763           len = UTF8SKIP(utf); /* len is 2 here */
2764           utf += len;
2765           len = UTF8SKIP(utf); /* len is 3 here */
2766
2767       Another way to skip over characters in a UTF-8 string is to use
2768       "utf8_hop", which takes a string and a number of characters to skip
2769       over.  You're on your own about bounds checking, though, so don't use
2770       it lightly.
2771
2772       All bytes in a multi-byte UTF-8 character will have the high bit set,
2773       so you can test if you need to do something special with this character
2774       like this (the "UTF8_IS_INVARIANT()" is a macro that tests whether the
2775       byte is encoded as a single byte even in UTF-8):
2776
2777           U8 *utf;     /* Initialize this to point to the beginning of the
2778                           sequence to convert */
2779           U8 *utf_end; /* Initialize this to 1 beyond the end of the sequence
2780                           pointed to by 'utf' */
2781           UV uv;       /* Returned code point; note: a UV, not a U8, not a
2782                           char */
2783           STRLEN len; /* Returned length of character in bytes */
2784
2785           if (!UTF8_IS_INVARIANT(*utf))
2786               /* Must treat this as UTF-8 */
2787               uv = utf8_to_uvchr_buf(utf, utf_end, &len);
2788           else
2789               /* OK to treat this character as a byte */
2790               uv = *utf;
2791
2792       You can also see in that example that we use "utf8_to_uvchr_buf" to get
2793       the value of the character; the inverse function "uvchr_to_utf8" is
2794       available for putting a UV into UTF-8:
2795
2796           if (!UVCHR_IS_INVARIANT(uv))
2797               /* Must treat this as UTF8 */
2798               utf8 = uvchr_to_utf8(utf8, uv);
2799           else
2800               /* OK to treat this character as a byte */
2801               *utf8++ = uv;
2802
2803       You must convert characters to UVs using the above functions if you're
2804       ever in a situation where you have to match UTF-8 and non-UTF-8
2805       characters.  You may not skip over UTF-8 characters in this case.  If
2806       you do this, you'll lose the ability to match hi-bit non-UTF-8
2807       characters; for instance, if your UTF-8 string contains "v196.172", and
2808       you skip that character, you can never match a "chr(200)" in a
2809       non-UTF-8 string.  So don't do that!
2810
2811       (Note that we don't have to test for invariant characters in the
2812       examples above.  The functions work on any well-formed UTF-8 input.
2813       It's just that its faster to avoid the function overhead when it's not
2814       needed.)
2815
2816   How does Perl store UTF-8 strings?
2817       Currently, Perl deals with UTF-8 strings and non-UTF-8 strings slightly
2818       differently.  A flag in the SV, "SVf_UTF8", indicates that the string
2819       is internally encoded as UTF-8.  Without it, the byte value is the
2820       codepoint number and vice versa.  This flag is only meaningful if the
2821       SV is "SvPOK" or immediately after stringification via "SvPV" or a
2822       similar macro.  You can check and manipulate this flag with the
2823       following macros:
2824
2825           SvUTF8(sv)
2826           SvUTF8_on(sv)
2827           SvUTF8_off(sv)
2828
2829       This flag has an important effect on Perl's treatment of the string: if
2830       UTF-8 data is not properly distinguished, regular expressions,
2831       "length", "substr" and other string handling operations will have
2832       undesirable (wrong) results.
2833
2834       The problem comes when you have, for instance, a string that isn't
2835       flagged as UTF-8, and contains a byte sequence that could be UTF-8 --
2836       especially when combining non-UTF-8 and UTF-8 strings.
2837
2838       Never forget that the "SVf_UTF8" flag is separate from the PV value;
2839       you need to be sure you don't accidentally knock it off while you're
2840       manipulating SVs.  More specifically, you cannot expect to do this:
2841
2842           SV *sv;
2843           SV *nsv;
2844           STRLEN len;
2845           char *p;
2846
2847           p = SvPV(sv, len);
2848           frobnicate(p);
2849           nsv = newSVpvn(p, len);
2850
2851       The "char*" string does not tell you the whole story, and you can't
2852       copy or reconstruct an SV just by copying the string value.  Check if
2853       the old SV has the UTF8 flag set (after the "SvPV" call), and act
2854       accordingly:
2855
2856           p = SvPV(sv, len);
2857           is_utf8 = SvUTF8(sv);
2858           frobnicate(p, is_utf8);
2859           nsv = newSVpvn(p, len);
2860           if (is_utf8)
2861               SvUTF8_on(nsv);
2862
2863       In the above, your "frobnicate" function has been changed to be made
2864       aware of whether or not it's dealing with UTF-8 data, so that it can
2865       handle the string appropriately.
2866
2867       Since just passing an SV to an XS function and copying the data of the
2868       SV is not enough to copy the UTF8 flags, even less right is just
2869       passing a "char *" to an XS function.
2870
2871       For full generality, use the "DO_UTF8" macro to see if the string in an
2872       SV is to be treated as UTF-8.  This takes into account if the call to
2873       the XS function is being made from within the scope of "use bytes".  If
2874       so, the underlying bytes that comprise the UTF-8 string are to be
2875       exposed, rather than the character they represent.  But this pragma
2876       should only really be used for debugging and perhaps low-level testing
2877       at the byte level.  Hence most XS code need not concern itself with
2878       this, but various areas of the perl core do need to support it.
2879
2880       And this isn't the whole story.  Starting in Perl v5.12, strings that
2881       aren't encoded in UTF-8 may also be treated as Unicode under various
2882       conditions (see "ASCII Rules versus Unicode Rules" in perlunicode).
2883       This is only really a problem for characters whose ordinals are between
2884       128 and 255, and their behavior varies under ASCII versus Unicode rules
2885       in ways that your code cares about (see "The "Unicode Bug"" in
2886       perlunicode).  There is no published API for dealing with this, as it
2887       is subject to change, but you can look at the code for "pp_lc" in pp.c
2888       for an example as to how it's currently done.
2889
2890   How do I convert a string to UTF-8?
2891       If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to
2892       upgrade the non-UTF-8 strings to UTF-8.  If you've got an SV, the
2893       easiest way to do this is:
2894
2895           sv_utf8_upgrade(sv);
2896
2897       However, you must not do this, for example:
2898
2899           if (!SvUTF8(left))
2900               sv_utf8_upgrade(left);
2901
2902       If you do this in a binary operator, you will actually change one of
2903       the strings that came into the operator, and, while it shouldn't be
2904       noticeable by the end user, it can cause problems in deficient code.
2905
2906       Instead, "bytes_to_utf8" will give you a UTF-8-encoded copy of its
2907       string argument.  This is useful for having the data available for
2908       comparisons and so on, without harming the original SV.  There's also
2909       "utf8_to_bytes" to go the other way, but naturally, this will fail if
2910       the string contains any characters above 255 that can't be represented
2911       in a single byte.
2912
2913   How do I compare strings?
2914       "sv_cmp" in perlapi and "sv_cmp_flags" in perlapi do a lexigraphic
2915       comparison of two SV's, and handle UTF-8ness properly.  Note, however,
2916       that Unicode specifies a much fancier mechanism for collation,
2917       available via the Unicode::Collate module.
2918
2919       To just compare two strings for equality/non-equality, you can just use
2920       "memEQ()" and "memNE()" as usual, except the strings must be both UTF-8
2921       or not UTF-8 encoded.
2922
2923       To compare two strings case-insensitively, use "foldEQ_utf8()" (the
2924       strings don't have to have the same UTF-8ness).
2925
2926   Is there anything else I need to know?
2927       Not really.  Just remember these things:
2928
2929       ·  There's no way to tell if a "char *" or "U8 *" string is UTF-8 or
2930          not.  But you can tell if an SV is to be treated as UTF-8 by calling
2931          "DO_UTF8" on it, after stringifying it with "SvPV" or a similar
2932          macro.  And, you can tell if SV is actually UTF-8 (even if it is not
2933          to be treated as such) by looking at its "SvUTF8" flag (again after
2934          stringifying it).  Don't forget to set the flag if something should
2935          be UTF-8.  Treat the flag as part of the PV, even though it's not --
2936          if you pass on the PV to somewhere, pass on the flag too.
2937
2938       ·  If a string is UTF-8, always use "utf8_to_uvchr_buf" to get at the
2939          value, unless "UTF8_IS_INVARIANT(*s)" in which case you can use *s.
2940
2941       ·  When writing a character UV to a UTF-8 string, always use
2942          "uvchr_to_utf8", unless "UVCHR_IS_INVARIANT(uv))" in which case you
2943          can use "*s = uv".
2944
2945       ·  Mixing UTF-8 and non-UTF-8 strings is tricky.  Use "bytes_to_utf8"
2946          to get a new string which is UTF-8 encoded, and then combine them.
2947

Custom Operators

2949       Custom operator support is an experimental feature that allows you to
2950       define your own ops.  This is primarily to allow the building of
2951       interpreters for other languages in the Perl core, but it also allows
2952       optimizations through the creation of "macro-ops" (ops which perform
2953       the functions of multiple ops which are usually executed together, such
2954       as "gvsv, gvsv, add".)
2955
2956       This feature is implemented as a new op type, "OP_CUSTOM".  The Perl
2957       core does not "know" anything special about this op type, and so it
2958       will not be involved in any optimizations.  This also means that you
2959       can define your custom ops to be any op structure -- unary, binary,
2960       list and so on -- you like.
2961
2962       It's important to know what custom operators won't do for you.  They
2963       won't let you add new syntax to Perl, directly.  They won't even let
2964       you add new keywords, directly.  In fact, they won't change the way
2965       Perl compiles a program at all.  You have to do those changes yourself,
2966       after Perl has compiled the program.  You do this either by
2967       manipulating the op tree using a "CHECK" block and the "B::Generate"
2968       module, or by adding a custom peephole optimizer with the "optimize"
2969       module.
2970
2971       When you do this, you replace ordinary Perl ops with custom ops by
2972       creating ops with the type "OP_CUSTOM" and the "op_ppaddr" of your own
2973       PP function.  This should be defined in XS code, and should look like
2974       the PP ops in "pp_*.c".  You are responsible for ensuring that your op
2975       takes the appropriate number of values from the stack, and you are
2976       responsible for adding stack marks if necessary.
2977
2978       You should also "register" your op with the Perl interpreter so that it
2979       can produce sensible error and warning messages.  Since it is possible
2980       to have multiple custom ops within the one "logical" op type
2981       "OP_CUSTOM", Perl uses the value of "o->op_ppaddr" to determine which
2982       custom op it is dealing with.  You should create an "XOP" structure for
2983       each ppaddr you use, set the properties of the custom op with
2984       "XopENTRY_set", and register the structure against the ppaddr using
2985       "Perl_custom_op_register".  A trivial example might look like:
2986
2987           static XOP my_xop;
2988           static OP *my_pp(pTHX);
2989
2990           BOOT:
2991               XopENTRY_set(&my_xop, xop_name, "myxop");
2992               XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2993               Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2994
2995       The available fields in the structure are:
2996
2997       xop_name
2998           A short name for your op.  This will be included in some error
2999           messages, and will also be returned as "$op->name" by the B module,
3000           so it will appear in the output of module like B::Concise.
3001
3002       xop_desc
3003           A short description of the function of the op.
3004
3005       xop_class
3006           Which of the various *OP structures this op uses.  This should be
3007           one of the "OA_*" constants from op.h, namely
3008
3009           OA_BASEOP
3010           OA_UNOP
3011           OA_BINOP
3012           OA_LOGOP
3013           OA_LISTOP
3014           OA_PMOP
3015           OA_SVOP
3016           OA_PADOP
3017           OA_PVOP_OR_SVOP
3018               This should be interpreted as '"PVOP"' only.  The "_OR_SVOP" is
3019               because the only core "PVOP", "OP_TRANS", can sometimes be a
3020               "SVOP" instead.
3021
3022           OA_LOOP
3023           OA_COP
3024
3025           The other "OA_*" constants should not be used.
3026
3027       xop_peep
3028           This member is of type "Perl_cpeep_t", which expands to "void
3029           (*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)".  If it is set, this
3030           function will be called from "Perl_rpeep" when ops of this type are
3031           encountered by the peephole optimizer.  o is the OP that needs
3032           optimizing; oldop is the previous OP optimized, whose "op_next"
3033           points to o.
3034
3035       "B::Generate" directly supports the creation of custom ops by name.
3036

Dynamic Scope and the Context Stack

3038       Note: this section describes a non-public internal API that is subject
3039       to change without notice.
3040
3041   Introduction to the context stack
3042       In Perl, dynamic scoping refers to the runtime nesting of things like
3043       subroutine calls, evals etc, as well as the entering and exiting of
3044       block scopes. For example, the restoring of a "local"ised variable is
3045       determined by the dynamic scope.
3046
3047       Perl tracks the dynamic scope by a data structure called the context
3048       stack, which is an array of "PERL_CONTEXT" structures, and which is
3049       itself a big union for all the types of context. Whenever a new scope
3050       is entered (such as a block, a "for" loop, or a subroutine call), a new
3051       context entry is pushed onto the stack. Similarly when leaving a block
3052       or returning from a subroutine call etc. a context is popped. Since the
3053       context stack represents the current dynamic scope, it can be searched.
3054       For example, "next LABEL" searches back through the stack looking for a
3055       loop context that matches the label; "return" pops contexts until it
3056       finds a sub or eval context or similar; "caller" examines sub contexts
3057       on the stack.
3058
3059       Each context entry is labelled with a context type, "cx_type". Typical
3060       context types are "CXt_SUB", "CXt_EVAL" etc., as well as "CXt_BLOCK"
3061       and "CXt_NULL" which represent a basic scope (as pushed by "pp_enter")
3062       and a sort block. The type determines which part of the context union
3063       are valid.
3064
3065       The main division in the context struct is between a substitution scope
3066       ("CXt_SUBST") and block scopes, which are everything else. The former
3067       is just used while executing "s///e", and won't be discussed further
3068       here.
3069
3070       All the block scope types share a common base, which corresponds to
3071       "CXt_BLOCK". This stores the old values of various scope-related
3072       variables like "PL_curpm", as well as information about the current
3073       scope, such as "gimme". On scope exit, the old variables are restored.
3074
3075       Particular block scope types store extra per-type information. For
3076       example, "CXt_SUB" stores the currently executing CV, while the various
3077       for loop types might hold the original loop variable SV. On scope exit,
3078       the per-type data is processed; for example the CV has its reference
3079       count decremented, and the original loop variable is restored.
3080
3081       The macro "cxstack" returns the base of the current context stack,
3082       while "cxstack_ix" is the index of the current frame within that stack.
3083
3084       In fact, the context stack is actually part of a stack-of-stacks
3085       system; whenever something unusual is done such as calling a "DESTROY"
3086       or tie handler, a new stack is pushed, then popped at the end.
3087
3088       Note that the API described here changed considerably in perl 5.24;
3089       prior to that, big macros like "PUSHBLOCK" and "POPSUB" were used; in
3090       5.24 they were replaced by the inline static functions described below.
3091       In addition, the ordering and detail of how these macros/function work
3092       changed in many ways, often subtly. In particular they didn't handle
3093       saving the savestack and temps stack positions, and required additional
3094       "ENTER", "SAVETMPS" and "LEAVE" compared to the new functions. The old-
3095       style macros will not be described further.
3096
3097   Pushing contexts
3098       For pushing a new context, the two basic functions are "cx =
3099       cx_pushblock()", which pushes a new basic context block and returns its
3100       address, and a family of similar functions with names like
3101       "cx_pushsub(cx)" which populate the additional type-dependent fields in
3102       the "cx" struct. Note that "CXt_NULL" and "CXt_BLOCK" don't have their
3103       own push functions, as they don't store any data beyond that pushed by
3104       "cx_pushblock".
3105
3106       The fields of the context struct and the arguments to the "cx_*"
3107       functions are subject to change between perl releases, representing
3108       whatever is convenient or efficient for that release.
3109
3110       A typical context stack pushing can be found in "pp_entersub"; the
3111       following shows a simplified and stripped-down example of a non-XS
3112       call, along with comments showing roughly what each function does.
3113
3114        dMARK;
3115        U8 gimme      = GIMME_V;
3116        bool hasargs  = cBOOL(PL_op->op_flags & OPf_STACKED);
3117        OP *retop     = PL_op->op_next;
3118        I32 old_ss_ix = PL_savestack_ix;
3119        CV *cv        = ....;
3120
3121        /* ... make mortal copies of stack args which are PADTMPs here ... */
3122
3123        /* ... do any additional savestack pushes here ... */
3124
3125        /* Now push a new context entry of type 'CXt_SUB'; initially just
3126         * doing the actions common to all block types: */
3127
3128        cx = cx_pushblock(CXt_SUB, gimme, MARK, old_ss_ix);
3129
3130            /* this does (approximately):
3131                CXINC;              /* cxstack_ix++ (grow if necessary) */
3132                cx = CX_CUR();      /* and get the address of new frame */
3133                cx->cx_type        = CXt_SUB;
3134                cx->blk_gimme      = gimme;
3135                cx->blk_oldsp      = MARK - PL_stack_base;
3136                cx->blk_oldsaveix  = old_ss_ix;
3137                cx->blk_oldcop     = PL_curcop;
3138                cx->blk_oldmarksp  = PL_markstack_ptr - PL_markstack;
3139                cx->blk_oldscopesp = PL_scopestack_ix;
3140                cx->blk_oldpm      = PL_curpm;
3141                cx->blk_old_tmpsfloor = PL_tmps_floor;
3142
3143                PL_tmps_floor        = PL_tmps_ix;
3144            */
3145
3146
3147        /* then update the new context frame with subroutine-specific info,
3148         * such as the CV about to be executed: */
3149
3150        cx_pushsub(cx, cv, retop, hasargs);
3151
3152            /* this does (approximately):
3153                cx->blk_sub.cv          = cv;
3154                cx->blk_sub.olddepth    = CvDEPTH(cv);
3155                cx->blk_sub.prevcomppad = PL_comppad;
3156                cx->cx_type            |= (hasargs) ? CXp_HASARGS : 0;
3157                cx->blk_sub.retop       = retop;
3158                SvREFCNT_inc_simple_void_NN(cv);
3159            */
3160
3161       Note that "cx_pushblock()" sets two new floors: for the args stack (to
3162       "MARK") and the temps stack (to "PL_tmps_ix"). While executing at this
3163       scope level, every "nextstate" (amongst others) will reset the args and
3164       tmps stack levels to these floors. Note that since "cx_pushblock" uses
3165       the current value of "PL_tmps_ix" rather than it being passed as an
3166       arg, this dictates at what point "cx_pushblock" should be called. In
3167       particular, any new mortals which should be freed only on scope exit
3168       (rather than at the next "nextstate") should be created first.
3169
3170       Most callers of "cx_pushblock" simply set the new args stack floor to
3171       the top of the previous stack frame, but for "CXt_LOOP_LIST" it stores
3172       the items being iterated over on the stack, and so sets "blk_oldsp" to
3173       the top of these items instead. Note that, contrary to its name,
3174       "blk_oldsp" doesn't always represent the value to restore "PL_stack_sp"
3175       to on scope exit.
3176
3177       Note the early capture of "PL_savestack_ix" to "old_ss_ix", which is
3178       later passed as an arg to "cx_pushblock". In the case of "pp_entersub",
3179       this is because, although most values needing saving are stored in
3180       fields of the context struct, an extra value needs saving only when the
3181       debugger is running, and it doesn't make sense to bloat the struct for
3182       this rare case. So instead it is saved on the savestack. Since this
3183       value gets calculated and saved before the context is pushed, it is
3184       necessary to pass the old value of "PL_savestack_ix" to "cx_pushblock",
3185       to ensure that the saved value gets freed during scope exit.  For most
3186       users of "cx_pushblock", where nothing needs pushing on the save stack,
3187       "PL_savestack_ix" is just passed directly as an arg to "cx_pushblock".
3188
3189       Note that where possible, values should be saved in the context struct
3190       rather than on the save stack; it's much faster that way.
3191
3192       Normally "cx_pushblock" should be immediately followed by the
3193       appropriate "cx_pushfoo", with nothing between them; this is because if
3194       code in-between could die (e.g. a warning upgraded to fatal), then the
3195       context stack unwinding code in "dounwind" would see (in the example
3196       above) a "CXt_SUB" context frame, but without all the subroutine-
3197       specific fields set, and crashes would soon ensue.
3198
3199       Where the two must be separate, initially set the type to "CXt_NULL" or
3200       "CXt_BLOCK", and later change it to "CXt_foo" when doing the
3201       "cx_pushfoo". This is exactly what "pp_enteriter" does, once it's
3202       determined which type of loop it's pushing.
3203
3204   Popping contexts
3205       Contexts are popped using "cx_popsub()" etc. and "cx_popblock()". Note
3206       however, that unlike "cx_pushblock", neither of these functions
3207       actually decrement the current context stack index; this is done
3208       separately using "CX_POP()".
3209
3210       There are two main ways that contexts are popped. During normal
3211       execution as scopes are exited, functions like "pp_leave",
3212       "pp_leaveloop" and "pp_leavesub" process and pop just one context using
3213       "cx_popfoo" and "cx_popblock". On the other hand, things like
3214       "pp_return" and "next" may have to pop back several scopes until a sub
3215       or loop context is found, and exceptions (such as "die") need to pop
3216       back contexts until an eval context is found. Both of these are
3217       accomplished by "dounwind()", which is capable of processing and
3218       popping all contexts above the target one.
3219
3220       Here is a typical example of context popping, as found in "pp_leavesub"
3221       (simplified slightly):
3222
3223        U8 gimme;
3224        PERL_CONTEXT *cx;
3225        SV **oldsp;
3226        OP *retop;
3227
3228        cx = CX_CUR();
3229
3230        gimme = cx->blk_gimme;
3231        oldsp = PL_stack_base + cx->blk_oldsp; /* last arg of previous frame */
3232
3233        if (gimme == G_VOID)
3234            PL_stack_sp = oldsp;
3235        else
3236            leave_adjust_stacks(oldsp, oldsp, gimme, 0);
3237
3238        CX_LEAVE_SCOPE(cx);
3239        cx_popsub(cx);
3240        cx_popblock(cx);
3241        retop = cx->blk_sub.retop;
3242        CX_POP(cx);
3243
3244        return retop;
3245
3246       The steps above are in a very specific order, designed to be the
3247       reverse order of when the context was pushed. The first thing to do is
3248       to copy and/or protect any any return arguments and free any temps in
3249       the current scope. Scope exits like an rvalue sub normally return a
3250       mortal copy of their return args (as opposed to lvalue subs). It is
3251       important to make this copy before the save stack is popped or
3252       variables are restored, or bad things like the following can happen:
3253
3254           sub f { my $x =...; $x }  # $x freed before we get to copy it
3255           sub f { /(...)/;    $1 }  # PL_curpm restored before $1 copied
3256
3257       Although we wish to free any temps at the same time, we have to be
3258       careful not to free any temps which are keeping return args alive; nor
3259       to free the temps we have just created while mortal copying return
3260       args. Fortunately, "leave_adjust_stacks()" is capable of making mortal
3261       copies of return args, shifting args down the stack, and only
3262       processing those entries on the temps stack that are safe to do so.
3263
3264       In void context no args are returned, so it's more efficient to skip
3265       calling "leave_adjust_stacks()". Also in void context, a "nextstate" op
3266       is likely to be imminently called which will do a "FREETMPS", so
3267       there's no need to do that either.
3268
3269       The next step is to pop savestack entries: "CX_LEAVE_SCOPE(cx)" is just
3270       defined as "<LEAVE_SCOPE(cx-"blk_oldsaveix)>>. Note that during the
3271       popping, it's possible for perl to call destructors, call "STORE" to
3272       undo localisations of tied vars, and so on. Any of these can die or
3273       call "exit()". In this case, "dounwind()" will be called, and the
3274       current context stack frame will be re-processed. Thus it is vital that
3275       all steps in popping a context are done in such a way to support
3276       reentrancy.  The other alternative, of decrementing "cxstack_ix" before
3277       processing the frame, would lead to leaks and the like if something
3278       died halfway through, or overwriting of the current frame.
3279
3280       "CX_LEAVE_SCOPE" itself is safely re-entrant: if only half the
3281       savestack items have been popped before dying and getting trapped by
3282       eval, then the "CX_LEAVE_SCOPE"s in "dounwind" or "pp_leaveeval" will
3283       continue where the first one left off.
3284
3285       The next step is the type-specific context processing; in this case
3286       "cx_popsub". In part, this looks like:
3287
3288           cv = cx->blk_sub.cv;
3289           CvDEPTH(cv) = cx->blk_sub.olddepth;
3290           cx->blk_sub.cv = NULL;
3291           SvREFCNT_dec(cv);
3292
3293       where its processing the just-executed CV. Note that before it
3294       decrements the CV's reference count, it nulls the "blk_sub.cv". This
3295       means that if it re-enters, the CV won't be freed twice. It also means
3296       that you can't rely on such type-specific fields having useful values
3297       after the return from "cx_popfoo".
3298
3299       Next, "cx_popblock" restores all the various interpreter vars to their
3300       previous values or previous high water marks; it expands to:
3301
3302           PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
3303           PL_scopestack_ix = cx->blk_oldscopesp;
3304           PL_curpm         = cx->blk_oldpm;
3305           PL_curcop        = cx->blk_oldcop;
3306           PL_tmps_floor    = cx->blk_old_tmpsfloor;
3307
3308       Note that it doesn't restore "PL_stack_sp"; as mentioned earlier, which
3309       value to restore it to depends on the context type (specifically "for
3310       (list) {}"), and what args (if any) it returns; and that will already
3311       have been sorted out earlier by "leave_adjust_stacks()".
3312
3313       Finally, the context stack pointer is actually decremented by
3314       "CX_POP(cx)".  After this point, it's possible that that the current
3315       context frame could be overwritten by other contexts being pushed.
3316       Although things like ties and "DESTROY" are supposed to work within a
3317       new context stack, it's best not to assume this. Indeed on debugging
3318       builds, "CX_POP(cx)" deliberately sets "cx" to null to detect code that
3319       is still relying on the field values in that context frame. Note in the
3320       "pp_leavesub()" example above, we grab "blk_sub.retop" before calling
3321       "CX_POP".
3322
3323   Redoing contexts
3324       Finally, there is "cx_topblock(cx)", which acts like a
3325       super-"nextstate" as regards to resetting various vars to their base
3326       values. It is used in places like "pp_next", "pp_redo" and "pp_goto"
3327       where rather than exiting a scope, we want to re-initialise the scope.
3328       As well as resetting "PL_stack_sp" like "nextstate", it also resets
3329       "PL_markstack_ptr", "PL_scopestack_ix" and "PL_curpm". Note that it
3330       doesn't do a "FREETMPS".
3331

AUTHORS

3333       Until May 1997, this document was maintained by Jeff Okamoto
3334       <okamoto@corp.hp.com>.  It is now maintained as part of Perl itself by
3335       the Perl 5 Porters <perl5-porters@perl.org>.
3336
3337       With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
3338       Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
3339       Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
3340       Stephen McCamant, and Gurusamy Sarathy.
3341

SEE ALSO

3343       perlapi, perlintern, perlxs, perlembed
3344
3345
3346
3347perl v5.28.2                      2018-11-01                       PERLGUTS(1)
Impressum