1PERLXS(1) Perl Programmers Reference Guide PERLXS(1)
2
6 perlxs - XS language reference manual
7
9 Introduction
10 XS is an interface description file format used to create an extension
11 interface between Perl and C code (or a C library) which one wishes to
12 use with Perl. The XS interface is combined with the library to create
13 a new library which can then be either dynamically loaded or statically
14 linked into perl. The XS interface description is written in the XS
15 language and is the core component of the Perl extension interface.
16 An XSUB forms the basic unit of the XS interface. After compilation by
18 the xsubpp compiler, each XSUB amounts to a C function definition which
19 will provide the glue between Perl calling conventions and C calling
20 conventions.
21 The glue code pulls the arguments from the Perl stack, converts these
23 Perl values to the formats expected by a C function, call this C
24 function, transfers the return values of the C function back to Perl.
25 Return values here may be a conventional C return value or any C
26 function arguments that may serve as output parameters. These return
27 values may be passed back to Perl either by putting them on the Perl
28 stack, or by modifying the arguments supplied from the Perl side.
29 The above is a somewhat simplified view of what really happens. Since
31 Perl allows more flexible calling conventions than C, XSUBs may do much
32 more in practice, such as checking input parameters for validity,
33 throwing exceptions (or returning undef/empty list) if the return value
34 from the C function indicates failure, calling different C functions
35 based on numbers and types of the arguments, providing an object-
36 oriented interface, etc.
37 Of course, one could write such glue code directly in C. However, this
39 would be a tedious task, especially if one needs to write glue for
40 multiple C functions, and/or one is not familiar enough with the Perl
41 stack discipline and other such arcana. XS comes to the rescue here:
42 instead of writing this glue C code in long-hand, one can write a more
43 concise short-hand description of what should be done by the glue, and
44 let the XS compiler xsubpp handle the rest.
45 The XS language allows one to describe the mapping between how the C
47 routine is used, and how the corresponding Perl routine is used. It
48 also allows creation of Perl routines which are directly translated to
49 C code and which are not related to a pre-existing C function. In
50 cases when the C interface coincides with the Perl interface, the XSUB
51 declaration is almost identical to a declaration of a C function (in
52 K&R style). In such circumstances, there is another tool called "h2xs"
53 that is able to translate an entire C header file into a corresponding
54 XS file that will provide glue to the functions/macros described in the
55 header file.
56 The XS compiler is called xsubpp. This compiler creates the constructs
58 necessary to let an XSUB manipulate Perl values, and creates the glue
59 necessary to let Perl call the XSUB. The compiler uses typemaps to
60 determine how to map C function parameters and output values to Perl
61 values and back. The default typemap (which comes with Perl) handles
62 many common C types. A supplementary typemap may also be needed to
63 handle any special structures and types for the library being linked.
64 A file in XS format starts with a C language section which goes until
66 the first "MODULE =" directive. Other XS directives and XSUB
67 definitions may follow this line. The "language" used in this part of
68 the file is usually referred to as the XS language. xsubpp recognizes
69 and skips POD (see perlpod) in both the C and XS language sections,
70 which allows the XS file to contain embedded documentation.
71 See perlxstut for a tutorial on the whole extension creation process.
73 Note: For some extensions, Dave Beazley's SWIG system may provide a
75 significantly more convenient mechanism for creating the extension glue
76 code. See http://www.swig.org/ for more information.
77 On The Road
79 Many of the examples which follow will concentrate on creating an
80 interface between Perl and the ONC+ RPC bind library functions. The
81 rpcb_gettime() function is used to demonstrate many features of the XS
82 language. This function has two parameters; the first is an input
83 parameter and the second is an output parameter. The function also
84 returns a status value.
85 bool_t rpcb_gettime(const char *host, time_t *timep);
87 From C this function will be called with the following statements.
89 #include <rpc/rpc.h>
91 bool_t status;
92 time_t timep;
93 status = rpcb_gettime( "localhost", &timep );
94 If an XSUB is created to offer a direct translation between this
96 function and Perl, then this XSUB will be used from Perl with the
97 following code. The $status and $timep variables will contain the
98 output of the function.
99 use RPC;
101 $status = rpcb_gettime( "localhost", $timep );
102 The following XS file shows an XS subroutine, or XSUB, which
104 demonstrates one possible interface to the rpcb_gettime() function.
105 This XSUB represents a direct translation between C and Perl and so
106 preserves the interface even from Perl. This XSUB will be invoked from
107 Perl with the usage shown above. Note that the first three #include
108 statements, for "EXTERN.h", "perl.h", and "XSUB.h", will always be
109 present at the beginning of an XS file. This approach and others will
110 be expanded later in this document.
111 #include "EXTERN.h"
113 #include "perl.h"
114 #include "XSUB.h"
115 #include <rpc/rpc.h>
116 MODULE = RPC PACKAGE = RPC
118 bool_t
120 rpcb_gettime(host,timep)
121 char *host
122 time_t &timep
123 OUTPUT:
124 timep
125 Any extension to Perl, including those containing XSUBs, should have a
127 Perl module to serve as the bootstrap which pulls the extension into
128 Perl. This module will export the extension's functions and variables
129 to the Perl program and will cause the extension's XSUBs to be linked
130 into Perl. The following module will be used for most of the examples
131 in this document and should be used from Perl with the "use" command as
132 shown earlier. Perl modules are explained in more detail later in this
133 document.
134 package RPC;
136 require Exporter;
138 require DynaLoader;
139 @ISA = qw(Exporter DynaLoader);
140 @EXPORT = qw( rpcb_gettime );
141 bootstrap RPC;
143 1;
144 Throughout this document a variety of interfaces to the rpcb_gettime()
146 XSUB will be explored. The XSUBs will take their parameters in
147 different orders or will take different numbers of parameters. In each
148 case the XSUB is an abstraction between Perl and the real C
149 rpcb_gettime() function, and the XSUB must always ensure that the real
150 rpcb_gettime() function is called with the correct parameters. This
151 abstraction will allow the programmer to create a more Perl-like
152 interface to the C function.
153 The Anatomy of an XSUB
155 The simplest XSUBs consist of 3 parts: a description of the return
156 value, the name of the XSUB routine and the names of its arguments, and
157 a description of types or formats of the arguments.
158 The following XSUB allows a Perl program to access a C library function
160 called sin(). The XSUB will imitate the C function which takes a
161 single argument and returns a single value.
162 double
164 sin(x)
165 double x
166 Optionally, one can merge the description of types and the list of
168 argument names, rewriting this as
169 double
171 sin(double x)
172 This makes this XSUB look similar to an ANSI C declaration. An
174 optional semicolon is allowed after the argument list, as in
175 double
177 sin(double x);
178 Parameters with C pointer types can have different semantic: C
180 functions with similar declarations
181 bool string_looks_as_a_number(char *s);
183 bool make_char_uppercase(char *c);
184 are used in absolutely incompatible manner. Parameters to these
186 functions could be described xsubpp like this:
187 char * s
189 char &c
190 Both these XS declarations correspond to the "char*" C type, but they
192 have different semantics, see "The & Unary Operator".
193 It is convenient to think that the indirection operator "*" should be
195 considered as a part of the type and the address operator "&" should be
196 considered part of the variable. See "The Typemap" for more info about
197 handling qualifiers and unary operators in C types.
198 The function name and the return type must be placed on separate lines
200 and should be flush left-adjusted.
201 INCORRECT CORRECT
203 double sin(x) double
205 double x sin(x)
206 double x
207 The rest of the function description may be indented or left-adjusted.
209 The following example shows a function with its body left-adjusted.
210 Most examples in this document will indent the body for better
211 readability.
212 CORRECT
214 double
216 sin(x)
217 double x
218 More complicated XSUBs may contain many other sections. Each section
220 of an XSUB starts with the corresponding keyword, such as INIT: or
221 CLEANUP:. However, the first two lines of an XSUB always contain the
222 same data: descriptions of the return type and the names of the
223 function and its parameters. Whatever immediately follows these is
224 considered to be an INPUT: section unless explicitly marked with
225 another keyword. (See "The INPUT: Keyword".)
226 An XSUB section continues until another section-start keyword is found.
228 The Argument Stack
230 The Perl argument stack is used to store the values which are sent as
231 parameters to the XSUB and to store the XSUB's return value(s). In
232 reality all Perl functions (including non-XSUB ones) keep their values
233 on this stack all the same time, each limited to its own range of
234 positions on the stack. In this document the first position on that
235 stack which belongs to the active function will be referred to as
236 position 0 for that function.
237 XSUBs refer to their stack arguments with the macro ST(x), where x
239 refers to a position in this XSUB's part of the stack. Position 0 for
240 that function would be known to the XSUB as ST(0). The XSUB's incoming
241 parameters and outgoing return values always begin at ST(0). For many
242 simple cases the xsubpp compiler will generate the code necessary to
243 handle the argument stack by embedding code fragments found in the
244 typemaps. In more complex cases the programmer must supply the code.
245 The RETVAL Variable
247 The RETVAL variable is a special C variable that is declared
248 automatically for you. The C type of RETVAL matches the return type of
249 the C library function. The xsubpp compiler will declare this variable
250 in each XSUB with non-"void" return type. By default the generated C
251 function will use RETVAL to hold the return value of the C library
252 function being called. In simple cases the value of RETVAL will be
253 placed in ST(0) of the argument stack where it can be received by Perl
254 as the return value of the XSUB.
255 If the XSUB has a return type of "void" then the compiler will not
257 declare a RETVAL variable for that function. When using a PPCODE:
258 section no manipulation of the RETVAL variable is required, the section
259 may use direct stack manipulation to place output values on the stack.
260 If PPCODE: directive is not used, "void" return value should be used
262 only for subroutines which do not return a value, even if CODE:
263 directive is used which sets ST(0) explicitly.
264 Older versions of this document recommended to use "void" return value
266 in such cases. It was discovered that this could lead to segfaults in
267 cases when XSUB was truly "void". This practice is now deprecated, and
268 may be not supported at some future version. Use the return value "SV
269 *" in such cases. (Currently "xsubpp" contains some heuristic code
270 which tries to disambiguate between "truly-void" and "old-practice-
271 declared-as-void" functions. Hence your code is at mercy of this
272 heuristics unless you use "SV *" as return value.)
273 Returning SVs, AVs and HVs through RETVAL
275 When you're using RETVAL to return an "SV *", there's some magic going
276 on behind the scenes that should be mentioned. When you're manipulating
277 the argument stack using the ST(x) macro, for example, you usually have
278 to pay special attention to reference counts. (For more about reference
279 counts, see perlguts.) To make your life easier, the typemap file
280 automatically makes "RETVAL" mortal when you're returning an "SV *".
281 Thus, the following two XSUBs are more or less equivalent:
282 void
284 alpha()
285 PPCODE:
286 ST(0) = newSVpv("Hello World",0);
287 sv_2mortal(ST(0));
288 XSRETURN(1);
289 SV *
291 beta()
292 CODE:
293 RETVAL = newSVpv("Hello World",0);
294 OUTPUT:
295 RETVAL
296 This is quite useful as it usually improves readability. While this
298 works fine for an "SV *", it's unfortunately not as easy to have "AV *"
299 or "HV *" as a return value. You should be able to write:
300 AV *
302 array()
303 CODE:
304 RETVAL = newAV();
305 /* do something with RETVAL */
306 OUTPUT:
307 RETVAL
308 But due to an unfixable bug (fixing it would break lots of existing
310 CPAN modules) in the typemap file, the reference count of the "AV *" is
311 not properly decremented. Thus, the above XSUB would leak memory
312 whenever it is being called. The same problem exists for "HV *".
313 When you're returning an "AV *" or a "HV *", you have to make sure
315 their reference count is decremented by making the AV or HV mortal:
316 AV *
318 array()
319 CODE:
320 RETVAL = newAV();
321 sv_2mortal((SV*)RETVAL);
322 /* do something with RETVAL */
323 OUTPUT:
324 RETVAL
325 And also remember that you don't have to do this for an "SV *".
327 The MODULE Keyword
329 The MODULE keyword is used to start the XS code and to specify the
330 package of the functions which are being defined. All text preceding
331 the first MODULE keyword is considered C code and is passed through to
332 the output with POD stripped, but otherwise untouched. Every XS module
333 will have a bootstrap function which is used to hook the XSUBs into
334 Perl. The package name of this bootstrap function will match the value
335 of the last MODULE statement in the XS source files. The value of
336 MODULE should always remain constant within the same XS file, though
337 this is not required.
338 The following example will start the XS code and will place all
340 functions in a package named RPC.
341 MODULE = RPC
343 The PACKAGE Keyword
345 When functions within an XS source file must be separated into packages
346 the PACKAGE keyword should be used. This keyword is used with the
347 MODULE keyword and must follow immediately after it when used.
348 MODULE = RPC PACKAGE = RPC
350 [ XS code in package RPC ]
352 MODULE = RPC PACKAGE = RPCB
354 [ XS code in package RPCB ]
356 MODULE = RPC PACKAGE = RPC
358 [ XS code in package RPC ]
360 The same package name can be used more than once, allowing for non-
362 contiguous code. This is useful if you have a stronger ordering
363 principle than package names.
364 Although this keyword is optional and in some cases provides redundant
366 information it should always be used. This keyword will ensure that
367 the XSUBs appear in the desired package.
368 The PREFIX Keyword
370 The PREFIX keyword designates prefixes which should be removed from the
371 Perl function names. If the C function is "rpcb_gettime()" and the
372 PREFIX value is "rpcb_" then Perl will see this function as
373 "gettime()".
374 This keyword should follow the PACKAGE keyword when used. If PACKAGE
376 is not used then PREFIX should follow the MODULE keyword.
377 MODULE = RPC PREFIX = rpc_
379 MODULE = RPC PACKAGE = RPCB PREFIX = rpcb_
381 The OUTPUT: Keyword
383 The OUTPUT: keyword indicates that certain function parameters should
384 be updated (new values made visible to Perl) when the XSUB terminates
385 or that certain values should be returned to the calling Perl function.
386 For simple functions which have no CODE: or PPCODE: section, such as
387 the sin() function above, the RETVAL variable is automatically
388 designated as an output value. For more complex functions the xsubpp
389 compiler will need help to determine which variables are output
390 variables.
391 This keyword will normally be used to complement the CODE: keyword.
393 The RETVAL variable is not recognized as an output variable when the
394 CODE: keyword is present. The OUTPUT: keyword is used in this
395 situation to tell the compiler that RETVAL really is an output
396 variable.
397 The OUTPUT: keyword can also be used to indicate that function
399 parameters are output variables. This may be necessary when a
400 parameter has been modified within the function and the programmer
401 would like the update to be seen by Perl.
402 bool_t
404 rpcb_gettime(host,timep)
405 char *host
406 time_t &timep
407 OUTPUT:
408 timep
409 The OUTPUT: keyword will also allow an output parameter to be mapped to
411 a matching piece of code rather than to a typemap.
412 bool_t
414 rpcb_gettime(host,timep)
415 char *host
416 time_t &timep
417 OUTPUT:
418 timep sv_setnv(ST(1), (double)timep);
419 xsubpp emits an automatic "SvSETMAGIC()" for all parameters in the
421 OUTPUT section of the XSUB, except RETVAL. This is the usually desired
422 behavior, as it takes care of properly invoking 'set' magic on output
423 parameters (needed for hash or array element parameters that must be
424 created if they didn't exist). If for some reason, this behavior is
425 not desired, the OUTPUT section may contain a "SETMAGIC: DISABLE" line
426 to disable it for the remainder of the parameters in the OUTPUT
427 section. Likewise, "SETMAGIC: ENABLE" can be used to reenable it for
428 the remainder of the OUTPUT section. See perlguts for more details
429 about 'set' magic.
430 The NO_OUTPUT Keyword
432 The NO_OUTPUT can be placed as the first token of the XSUB. This
433 keyword indicates that while the C subroutine we provide an interface
434 to has a non-"void" return type, the return value of this C subroutine
435 should not be returned from the generated Perl subroutine.
436 With this keyword present "The RETVAL Variable" is created, and in the
438 generated call to the subroutine this variable is assigned to, but the
439 value of this variable is not going to be used in the auto-generated
440 code.
441 This keyword makes sense only if "RETVAL" is going to be accessed by
443 the user-supplied code. It is especially useful to make a function
444 interface more Perl-like, especially when the C return value is just an
445 error condition indicator. For example,
446 NO_OUTPUT int
448 delete_file(char *name)
449 POSTCALL:
450 if (RETVAL != 0)
451 croak("Error %d while deleting file '%s'", RETVAL, name);
452 Here the generated XS function returns nothing on success, and will
454 die() with a meaningful error message on error.
455 The CODE: Keyword
457 This keyword is used in more complicated XSUBs which require special
458 handling for the C function. The RETVAL variable is still declared,
459 but it will not be returned unless it is specified in the OUTPUT:
460 section.
461 The following XSUB is for a C function which requires special handling
463 of its parameters. The Perl usage is given first.
464 $status = rpcb_gettime( "localhost", $timep );
466 The XSUB follows.
468 bool_t
470 rpcb_gettime(host,timep)
471 char *host
472 time_t timep
473 CODE:
474 RETVAL = rpcb_gettime( host, &timep );
475 OUTPUT:
476 timep
477 RETVAL
478 The INIT: Keyword
480 The INIT: keyword allows initialization to be inserted into the XSUB
481 before the compiler generates the call to the C function. Unlike the
482 CODE: keyword above, this keyword does not affect the way the compiler
483 handles RETVAL.
484 bool_t
486 rpcb_gettime(host,timep)
487 char *host
488 time_t &timep
489 INIT:
490 printf("# Host is %s\n", host );
491 OUTPUT:
492 timep
493 Another use for the INIT: section is to check for preconditions before
495 making a call to the C function:
496 long long
498 lldiv(a,b)
499 long long a
500 long long b
501 INIT:
502 if (a == 0 && b == 0)
503 XSRETURN_UNDEF;
504 if (b == 0)
505 croak("lldiv: cannot divide by 0");
506 The NO_INIT Keyword
508 The NO_INIT keyword is used to indicate that a function parameter is
509 being used only as an output value. The xsubpp compiler will normally
510 generate code to read the values of all function parameters from the
511 argument stack and assign them to C variables upon entry to the
512 function. NO_INIT will tell the compiler that some parameters will be
513 used for output rather than for input and that they will be handled
514 before the function terminates.
515 The following example shows a variation of the rpcb_gettime() function.
517 This function uses the timep variable only as an output variable and
518 does not care about its initial contents.
519 bool_t
521 rpcb_gettime(host,timep)
522 char *host
523 time_t &timep = NO_INIT
524 OUTPUT:
525 timep
526 Initializing Function Parameters
528 C function parameters are normally initialized with their values from
529 the argument stack (which in turn contains the parameters that were
530 passed to the XSUB from Perl). The typemaps contain the code segments
531 which are used to translate the Perl values to the C parameters. The
532 programmer, however, is allowed to override the typemaps and supply
533 alternate (or additional) initialization code. Initialization code
534 starts with the first "=", ";" or "+" on a line in the INPUT: section.
535 The only exception happens if this ";" terminates the line, then this
536 ";" is quietly ignored.
537 The following code demonstrates how to supply initialization code for
539 function parameters. The initialization code is eval'ed within double
540 quotes by the compiler before it is added to the output so anything
541 which should be interpreted literally [mainly "$", "@", or "\\"] must
542 be protected with backslashes. The variables $var, $arg, and $type can
543 be used as in typemaps.
544 bool_t
546 rpcb_gettime(host,timep)
547 char *host = (char *)SvPV_nolen($arg);
548 time_t &timep = 0;
549 OUTPUT:
550 timep
551 This should not be used to supply default values for parameters. One
553 would normally use this when a function parameter must be processed by
554 another library function before it can be used. Default parameters are
555 covered in the next section.
556 If the initialization begins with "=", then it is output in the
558 declaration for the input variable, replacing the initialization
559 supplied by the typemap. If the initialization begins with ";" or "+",
560 then it is performed after all of the input variables have been
561 declared. In the ";" case the initialization normally supplied by the
562 typemap is not performed. For the "+" case, the declaration for the
563 variable will include the initialization from the typemap. A global
564 variable, %v, is available for the truly rare case where information
565 from one initialization is needed in another initialization.
566 Here's a truly obscure example:
568 bool_t
570 rpcb_gettime(host,timep)
571 time_t &timep; /* \$v{timep}=@{[$v{timep}=$arg]} */
572 char *host + SvOK($v{timep}) ? SvPV_nolen($arg) : NULL;
573 OUTPUT:
574 timep
575 The construct "\$v{timep}=@{[$v{timep}=$arg]}" used in the above
577 example has a two-fold purpose: first, when this line is processed by
578 xsubpp, the Perl snippet "$v{timep}=$arg" is evaluated. Second, the
579 text of the evaluated snippet is output into the generated C file
580 (inside a C comment)! During the processing of "char *host" line, $arg
581 will evaluate to ST(0), and $v{timep} will evaluate to ST(1).
582 Default Parameter Values
584 Default values for XSUB arguments can be specified by placing an
585 assignment statement in the parameter list. The default value may be a
586 number, a string or the special string "NO_INIT". Defaults should
587 always be used on the right-most parameters only.
588 To allow the XSUB for rpcb_gettime() to have a default host value the
590 parameters to the XSUB could be rearranged. The XSUB will then call
591 the real rpcb_gettime() function with the parameters in the correct
592 order. This XSUB can be called from Perl with either of the following
593 statements:
594 $status = rpcb_gettime( $timep, $host );
596 $status = rpcb_gettime( $timep );
598 The XSUB will look like the code which follows. A CODE: block is
600 used to call the real rpcb_gettime() function with the parameters in
601 the correct order for that function.
602 bool_t
604 rpcb_gettime(timep,host="localhost")
605 char *host
606 time_t timep = NO_INIT
607 CODE:
608 RETVAL = rpcb_gettime( host, &timep );
609 OUTPUT:
610 timep
611 RETVAL
612 The PREINIT: Keyword
614 The PREINIT: keyword allows extra variables to be declared immediately
615 before or after the declarations of the parameters from the INPUT:
616 section are emitted.
617 If a variable is declared inside a CODE: section it will follow any
619 typemap code that is emitted for the input parameters. This may result
620 in the declaration ending up after C code, which is C syntax error.
621 Similar errors may happen with an explicit ";"-type or "+"-type
622 initialization of parameters is used (see "Initializing Function
623 Parameters"). Declaring these variables in an INIT: section will not
624 help.
625 In such cases, to force an additional variable to be declared together
627 with declarations of other variables, place the declaration into a
628 PREINIT: section. The PREINIT: keyword may be used one or more times
629 within an XSUB.
630 The following examples are equivalent, but if the code is using complex
632 typemaps then the first example is safer.
633 bool_t
635 rpcb_gettime(timep)
636 time_t timep = NO_INIT
637 PREINIT:
638 char *host = "localhost";
639 CODE:
640 RETVAL = rpcb_gettime( host, &timep );
641 OUTPUT:
642 timep
643 RETVAL
644 For this particular case an INIT: keyword would generate the same C
646 code as the PREINIT: keyword. Another correct, but error-prone
647 example:
648 bool_t
650 rpcb_gettime(timep)
651 time_t timep = NO_INIT
652 CODE:
653 char *host = "localhost";
654 RETVAL = rpcb_gettime( host, &timep );
655 OUTPUT:
656 timep
657 RETVAL
658 Another way to declare "host" is to use a C block in the CODE: section:
660 bool_t
662 rpcb_gettime(timep)
663 time_t timep = NO_INIT
664 CODE:
665 {
666 char *host = "localhost";
667 RETVAL = rpcb_gettime( host, &timep );
668 }
669 OUTPUT:
670 timep
671 RETVAL
672 The ability to put additional declarations before the typemap entries
674 are processed is very handy in the cases when typemap conversions
675 manipulate some global state:
676 MyObject
678 mutate(o)
679 PREINIT:
680 MyState st = global_state;
681 INPUT:
682 MyObject o;
683 CLEANUP:
684 reset_to(global_state, st);
685 Here we suppose that conversion to "MyObject" in the INPUT: section and
687 from MyObject when processing RETVAL will modify a global variable
688 "global_state". After these conversions are performed, we restore the
689 old value of "global_state" (to avoid memory leaks, for example).
690 There is another way to trade clarity for compactness: INPUT sections
692 allow declaration of C variables which do not appear in the parameter
693 list of a subroutine. Thus the above code for mutate() can be
694 rewritten as
695 MyObject
697 mutate(o)
698 MyState st = global_state;
699 MyObject o;
700 CLEANUP:
701 reset_to(global_state, st);
702 and the code for rpcb_gettime() can be rewritten as
704 bool_t
706 rpcb_gettime(timep)
707 time_t timep = NO_INIT
708 char *host = "localhost";
709 C_ARGS:
710 host, &timep
711 OUTPUT:
712 timep
713 RETVAL
714 The SCOPE: Keyword
716 The SCOPE: keyword allows scoping to be enabled for a particular XSUB.
717 If enabled, the XSUB will invoke ENTER and LEAVE automatically.
718 To support potentially complex type mappings, if a typemap entry used
720 by an XSUB contains a comment like "/*scope*/" then scoping will be
721 automatically enabled for that XSUB.
722 To enable scoping:
724 SCOPE: ENABLE
726 To disable scoping:
728 SCOPE: DISABLE
730 The INPUT: Keyword
732 The XSUB's parameters are usually evaluated immediately after entering
733 the XSUB. The INPUT: keyword can be used to force those parameters to
734 be evaluated a little later. The INPUT: keyword can be used multiple
735 times within an XSUB and can be used to list one or more input
736 variables. This keyword is used with the PREINIT: keyword.
737 The following example shows how the input parameter "timep" can be
739 evaluated late, after a PREINIT.
740 bool_t
742 rpcb_gettime(host,timep)
743 char *host
744 PREINIT:
745 time_t tt;
746 INPUT:
747 time_t timep
748 CODE:
749 RETVAL = rpcb_gettime( host, &tt );
750 timep = tt;
751 OUTPUT:
752 timep
753 RETVAL
754 The next example shows each input parameter evaluated late.
756 bool_t
758 rpcb_gettime(host,timep)
759 PREINIT:
760 time_t tt;
761 INPUT:
762 char *host
763 PREINIT:
764 char *h;
765 INPUT:
766 time_t timep
767 CODE:
768 h = host;
769 RETVAL = rpcb_gettime( h, &tt );
770 timep = tt;
771 OUTPUT:
772 timep
773 RETVAL
774 Since INPUT sections allow declaration of C variables which do not
776 appear in the parameter list of a subroutine, this may be shortened to:
777 bool_t
779 rpcb_gettime(host,timep)
780 time_t tt;
781 char *host;
782 char *h = host;
783 time_t timep;
784 CODE:
785 RETVAL = rpcb_gettime( h, &tt );
786 timep = tt;
787 OUTPUT:
788 timep
789 RETVAL
790 (We used our knowledge that input conversion for "char *" is a "simple"
792 one, thus "host" is initialized on the declaration line, and our
793 assignment "h = host" is not performed too early. Otherwise one would
794 need to have the assignment "h = host" in a CODE: or INIT: section.)
795 The IN/OUTLIST/IN_OUTLIST/OUT/IN_OUT Keywords
797 In the list of parameters for an XSUB, one can precede parameter names
798 by the "IN"/"OUTLIST"/"IN_OUTLIST"/"OUT"/"IN_OUT" keywords. "IN"
799 keyword is the default, the other keywords indicate how the Perl
800 interface should differ from the C interface.
801 Parameters preceded by "OUTLIST"/"IN_OUTLIST"/"OUT"/"IN_OUT" keywords
803 are considered to be used by the C subroutine via pointers.
804 "OUTLIST"/"OUT" keywords indicate that the C subroutine does not
805 inspect the memory pointed by this parameter, but will write through
806 this pointer to provide additional return values.
807 Parameters preceded by "OUTLIST" keyword do not appear in the usage
809 signature of the generated Perl function.
810 Parameters preceded by "IN_OUTLIST"/"IN_OUT"/"OUT" do appear as
812 parameters to the Perl function. With the exception of
813 "OUT"-parameters, these parameters are converted to the corresponding C
814 type, then pointers to these data are given as arguments to the C
815 function. It is expected that the C function will write through these
816 pointers.
817 The return list of the generated Perl function consists of the C return
819 value from the function (unless the XSUB is of "void" return type or
820 "The NO_OUTPUT Keyword" was used) followed by all the "OUTLIST" and
821 "IN_OUTLIST" parameters (in the order of appearance). On the return
822 from the XSUB the "IN_OUT"/"OUT" Perl parameter will be modified to
823 have the values written by the C function.
824 For example, an XSUB
826 void
828 day_month(OUTLIST day, IN unix_time, OUTLIST month)
829 int day
830 int unix_time
831 int month
832 should be used from Perl as
834 my ($day, $month) = day_month(time);
836 The C signature of the corresponding function should be
838 void day_month(int *day, int unix_time, int *month);
840 The "IN"/"OUTLIST"/"IN_OUTLIST"/"IN_OUT"/"OUT" keywords can be mixed
842 with ANSI-style declarations, as in
843 void
845 day_month(OUTLIST int day, int unix_time, OUTLIST int month)
846 (here the optional "IN" keyword is omitted).
848 The "IN_OUT" parameters are identical with parameters introduced with
850 "The & Unary Operator" and put into the "OUTPUT:" section (see "The
851 OUTPUT: Keyword"). The "IN_OUTLIST" parameters are very similar, the
852 only difference being that the value C function writes through the
853 pointer would not modify the Perl parameter, but is put in the output
854 list.
855 The "OUTLIST"/"OUT" parameter differ from "IN_OUTLIST"/"IN_OUT"
857 parameters only by the initial value of the Perl parameter not being
858 read (and not being given to the C function - which gets some garbage
859 instead). For example, the same C function as above can be interfaced
860 with as
861 void day_month(OUT int day, int unix_time, OUT int month);
863 or
865 void
867 day_month(day, unix_time, month)
868 int &day = NO_INIT
869 int unix_time
870 int &month = NO_INIT
871 OUTPUT:
872 day
873 month
874 However, the generated Perl function is called in very C-ish style:
876 my ($day, $month);
878 day_month($day, time, $month);
879 The "length(NAME)" Keyword
881 If one of the input arguments to the C function is the length of a
882 string argument "NAME", one can substitute the name of the length-
883 argument by "length(NAME)" in the XSUB declaration. This argument must
884 be omitted when the generated Perl function is called. E.g.,
885 void
887 dump_chars(char *s, short l)
888 {
889 short n = 0;
890 while (n < l) {
891 printf("s[%d] = \"\\%#03o\"\n", n, (int)s[n]);
892 n++;
893 }
894 }
895 MODULE = x PACKAGE = x
897 void dump_chars(char *s, short length(s))
899 should be called as "dump_chars($string)".
901 This directive is supported with ANSI-type function declarations only.
903 Variable-length Parameter Lists
905 XSUBs can have variable-length parameter lists by specifying an
906 ellipsis "(...)" in the parameter list. This use of the ellipsis is
907 similar to that found in ANSI C. The programmer is able to determine
908 the number of arguments passed to the XSUB by examining the "items"
909 variable which the xsubpp compiler supplies for all XSUBs. By using
910 this mechanism one can create an XSUB which accepts a list of
911 parameters of unknown length.
912 The host parameter for the rpcb_gettime() XSUB can be optional so the
914 ellipsis can be used to indicate that the XSUB will take a variable
915 number of parameters. Perl should be able to call this XSUB with
916 either of the following statements.
917 $status = rpcb_gettime( $timep, $host );
919 $status = rpcb_gettime( $timep );
921 The XS code, with ellipsis, follows.
923 bool_t
925 rpcb_gettime(timep, ...)
926 time_t timep = NO_INIT
927 PREINIT:
928 char *host = "localhost";
929 CODE:
930 if( items > 1 )
931 host = (char *)SvPV_nolen(ST(1));
932 RETVAL = rpcb_gettime( host, &timep );
933 OUTPUT:
934 timep
935 RETVAL
936 The C_ARGS: Keyword
938 The C_ARGS: keyword allows creating of XSUBS which have different
939 calling sequence from Perl than from C, without a need to write CODE:
940 or PPCODE: section. The contents of the C_ARGS: paragraph is put as
941 the argument to the called C function without any change.
942 For example, suppose that a C function is declared as
944 symbolic nth_derivative(int n, symbolic function, int flags);
946 and that the default flags are kept in a global C variable
948 "default_flags". Suppose that you want to create an interface which is
949 called as
950 $second_deriv = $function->nth_derivative(2);
952 To do this, declare the XSUB as
954 symbolic
956 nth_derivative(function, n)
957 symbolic function
958 int n
959 C_ARGS:
960 n, function, default_flags
961 The PPCODE: Keyword
963 The PPCODE: keyword is an alternate form of the CODE: keyword and is
964 used to tell the xsubpp compiler that the programmer is supplying the
965 code to control the argument stack for the XSUBs return values.
966 Occasionally one will want an XSUB to return a list of values rather
967 than a single value. In these cases one must use PPCODE: and then
968 explicitly push the list of values on the stack. The PPCODE: and CODE:
969 keywords should not be used together within the same XSUB.
970 The actual difference between PPCODE: and CODE: sections is in the
972 initialization of "SP" macro (which stands for the current Perl stack
973 pointer), and in the handling of data on the stack when returning from
974 an XSUB. In CODE: sections SP preserves the value which was on entry
975 to the XSUB: SP is on the function pointer (which follows the last
976 parameter). In PPCODE: sections SP is moved backward to the beginning
977 of the parameter list, which allows "PUSH*()" macros to place output
978 values in the place Perl expects them to be when the XSUB returns back
979 to Perl.
980 The generated trailer for a CODE: section ensures that the number of
982 return values Perl will see is either 0 or 1 (depending on the
983 "void"ness of the return value of the C function, and heuristics
984 mentioned in "The RETVAL Variable"). The trailer generated for a
985 PPCODE: section is based on the number of return values and on the
986 number of times "SP" was updated by "[X]PUSH*()" macros.
987 Note that macros ST(i), "XST_m*()" and "XSRETURN*()" work equally well
989 in CODE: sections and PPCODE: sections.
990 The following XSUB will call the C rpcb_gettime() function and will
992 return its two output values, timep and status, to Perl as a single
993 list.
994 void
996 rpcb_gettime(host)
997 char *host
998 PREINIT:
999 time_t timep;
1000 bool_t status;
1001 PPCODE:
1002 status = rpcb_gettime( host, &timep );
1003 EXTEND(SP, 2);
1004 PUSHs(sv_2mortal(newSViv(status)));
1005 PUSHs(sv_2mortal(newSViv(timep)));
1006 Notice that the programmer must supply the C code necessary to have the
1008 real rpcb_gettime() function called and to have the return values
1009 properly placed on the argument stack.
1010 The "void" return type for this function tells the xsubpp compiler that
1012 the RETVAL variable is not needed or used and that it should not be
1013 created. In most scenarios the void return type should be used with
1014 the PPCODE: directive.
1015 The EXTEND() macro is used to make room on the argument stack for 2
1017 return values. The PPCODE: directive causes the xsubpp compiler to
1018 create a stack pointer available as "SP", and it is this pointer which
1019 is being used in the EXTEND() macro. The values are then pushed onto
1020 the stack with the PUSHs() macro.
1021 Now the rpcb_gettime() function can be used from Perl with the
1023 following statement.
1024 ($status, $timep) = rpcb_gettime("localhost");
1026 When handling output parameters with a PPCODE section, be sure to
1028 handle 'set' magic properly. See perlguts for details about 'set'
1029 magic.
1030 Returning Undef And Empty Lists
1032 Occasionally the programmer will want to return simply "undef" or an
1033 empty list if a function fails rather than a separate status value.
1034 The rpcb_gettime() function offers just this situation. If the
1035 function succeeds we would like to have it return the time and if it
1036 fails we would like to have undef returned. In the following Perl code
1037 the value of $timep will either be undef or it will be a valid time.
1038 $timep = rpcb_gettime( "localhost" );
1040 The following XSUB uses the "SV *" return type as a mnemonic only, and
1042 uses a CODE: block to indicate to the compiler that the programmer has
1043 supplied all the necessary code. The sv_newmortal() call will
1044 initialize the return value to undef, making that the default return
1045 value.
1046 SV *
1048 rpcb_gettime(host)
1049 char * host
1050 PREINIT:
1051 time_t timep;
1052 bool_t x;
1053 CODE:
1054 ST(0) = sv_newmortal();
1055 if( rpcb_gettime( host, &timep ) )
1056 sv_setnv( ST(0), (double)timep);
1057 The next example demonstrates how one would place an explicit undef in
1059 the return value, should the need arise.
1060 SV *
1062 rpcb_gettime(host)
1063 char * host
1064 PREINIT:
1065 time_t timep;
1066 bool_t x;
1067 CODE:
1068 if( rpcb_gettime( host, &timep ) ){
1069 ST(0) = sv_newmortal();
1070 sv_setnv( ST(0), (double)timep);
1071 }
1072 else{
1073 ST(0) = &PL_sv_undef;
1074 }
1075 To return an empty list one must use a PPCODE: block and then not push
1077 return values on the stack.
1078 void
1080 rpcb_gettime(host)
1081 char *host
1082 PREINIT:
1083 time_t timep;
1084 PPCODE:
1085 if( rpcb_gettime( host, &timep ) )
1086 PUSHs(sv_2mortal(newSViv(timep)));
1087 else{
1088 /* Nothing pushed on stack, so an empty
1089 * list is implicitly returned. */
1090 }
1091 Some people may be inclined to include an explicit "return" in the
1093 above XSUB, rather than letting control fall through to the end. In
1094 those situations "XSRETURN_EMPTY" should be used, instead. This will
1095 ensure that the XSUB stack is properly adjusted. Consult perlapi for
1096 other "XSRETURN" macros.
1097 Since "XSRETURN_*" macros can be used with CODE blocks as well, one can
1099 rewrite this example as:
1100 int
1102 rpcb_gettime(host)
1103 char *host
1104 PREINIT:
1105 time_t timep;
1106 CODE:
1107 RETVAL = rpcb_gettime( host, &timep );
1108 if (RETVAL == 0)
1109 XSRETURN_UNDEF;
1110 OUTPUT:
1111 RETVAL
1112 In fact, one can put this check into a POSTCALL: section as well.
1114 Together with PREINIT: simplifications, this leads to:
1115 int
1117 rpcb_gettime(host)
1118 char *host
1119 time_t timep;
1120 POSTCALL:
1121 if (RETVAL == 0)
1122 XSRETURN_UNDEF;
1123 The REQUIRE: Keyword
1125 The REQUIRE: keyword is used to indicate the minimum version of the
1126 xsubpp compiler needed to compile the XS module. An XS module which
1127 contains the following statement will compile with only xsubpp version
1128 1.922 or greater:
1129 REQUIRE: 1.922
1131 The CLEANUP: Keyword
1133 This keyword can be used when an XSUB requires special cleanup
1134 procedures before it terminates. When the CLEANUP: keyword is used it
1135 must follow any CODE:, PPCODE:, or OUTPUT: blocks which are present in
1136 the XSUB. The code specified for the cleanup block will be added as
1137 the last statements in the XSUB.
1138 The POSTCALL: Keyword
1140 This keyword can be used when an XSUB requires special procedures
1141 executed after the C subroutine call is performed. When the POSTCALL:
1142 keyword is used it must precede OUTPUT: and CLEANUP: blocks which are
1143 present in the XSUB.
1144 See examples in "The NO_OUTPUT Keyword" and "Returning Undef And Empty
1146 Lists".
1147 The POSTCALL: block does not make a lot of sense when the C subroutine
1149 call is supplied by user by providing either CODE: or PPCODE: section.
1150 The BOOT: Keyword
1152 The BOOT: keyword is used to add code to the extension's bootstrap
1153 function. The bootstrap function is generated by the xsubpp compiler
1154 and normally holds the statements necessary to register any XSUBs with
1155 Perl. With the BOOT: keyword the programmer can tell the compiler to
1156 add extra statements to the bootstrap function.
1157 This keyword may be used any time after the first MODULE keyword and
1159 should appear on a line by itself. The first blank line after the
1160 keyword will terminate the code block.
1161 BOOT:
1163 # The following message will be printed when the
1164 # bootstrap function executes.
1165 printf("Hello from the bootstrap!\n");
1166 The VERSIONCHECK: Keyword
1168 The VERSIONCHECK: keyword corresponds to xsubpp's "-versioncheck" and
1169 "-noversioncheck" options. This keyword overrides the command line
1170 options. Version checking is enabled by default. When version
1171 checking is enabled the XS module will attempt to verify that its
1172 version matches the version of the PM module.
1173 To enable version checking:
1175 VERSIONCHECK: ENABLE
1177 To disable version checking:
1179 VERSIONCHECK: DISABLE
1181 Note that if the version of the PM module is an NV (a floating point
1183 number), it will be stringified with a possible loss of precision
1184 (currently chopping to nine decimal places) so that it may not match
1185 the version of the XS module anymore. Quoting the $VERSION declaration
1186 to make it a string is recommended if long version numbers are used.
1187 The PROTOTYPES: Keyword
1189 The PROTOTYPES: keyword corresponds to xsubpp's "-prototypes" and
1190 "-noprototypes" options. This keyword overrides the command line
1191 options. Prototypes are enabled by default. When prototypes are
1192 enabled XSUBs will be given Perl prototypes. This keyword may be used
1193 multiple times in an XS module to enable and disable prototypes for
1194 different parts of the module.
1195 To enable prototypes:
1197 PROTOTYPES: ENABLE
1199 To disable prototypes:
1201 PROTOTYPES: DISABLE
1203 The PROTOTYPE: Keyword
1205 This keyword is similar to the PROTOTYPES: keyword above but can be
1206 used to force xsubpp to use a specific prototype for the XSUB. This
1207 keyword overrides all other prototype options and keywords but affects
1208 only the current XSUB. Consult "Prototypes" in perlsub for information
1209 about Perl prototypes.
1210 bool_t
1212 rpcb_gettime(timep, ...)
1213 time_t timep = NO_INIT
1214 PROTOTYPE: $;$
1215 PREINIT:
1216 char *host = "localhost";
1217 CODE:
1218 if( items > 1 )
1219 host = (char *)SvPV_nolen(ST(1));
1220 RETVAL = rpcb_gettime( host, &timep );
1221 OUTPUT:
1222 timep
1223 RETVAL
1224 If the prototypes are enabled, you can disable it locally for a given
1226 XSUB as in the following example:
1227 void
1229 rpcb_gettime_noproto()
1230 PROTOTYPE: DISABLE
1231 ...
1232 The ALIAS: Keyword
1234 The ALIAS: keyword allows an XSUB to have two or more unique Perl names
1235 and to know which of those names was used when it was invoked. The
1236 Perl names may be fully-qualified with package names. Each alias is
1237 given an index. The compiler will setup a variable called "ix" which
1238 contain the index of the alias which was used. When the XSUB is called
1239 with its declared name "ix" will be 0.
1240 The following example will create aliases "FOO::gettime()" and
1242 "BAR::getit()" for this function.
1243 bool_t
1245 rpcb_gettime(host,timep)
1246 char *host
1247 time_t &timep
1248 ALIAS:
1249 FOO::gettime = 1
1250 BAR::getit = 2
1251 INIT:
1252 printf("# ix = %d\n", ix );
1253 OUTPUT:
1254 timep
1255 The OVERLOAD: Keyword
1257 Instead of writing an overloaded interface using pure Perl, you can
1258 also use the OVERLOAD keyword to define additional Perl names for your
1259 functions (like the ALIAS: keyword above). However, the overloaded
1260 functions must be defined with three parameters (except for the
1261 nomethod() function which needs four parameters). If any function has
1262 the OVERLOAD: keyword, several additional lines will be defined in the
1263 c file generated by xsubpp in order to register with the overload
1264 magic.
1265 Since blessed objects are actually stored as RV's, it is useful to use
1267 the typemap features to preprocess parameters and extract the actual SV
1268 stored within the blessed RV. See the sample for T_PTROBJ_SPECIAL
1269 below.
1270 To use the OVERLOAD: keyword, create an XS function which takes three
1272 input parameters ( or use the c style '...' definition) like this:
1273 SV *
1275 cmp (lobj, robj, swap)
1276 My_Module_obj lobj
1277 My_Module_obj robj
1278 IV swap
1279 OVERLOAD: cmp <=>
1280 { /* function defined here */}
1281 In this case, the function will overload both of the three way
1283 comparison operators. For all overload operations using non-alpha
1284 characters, you must type the parameter without quoting, separating
1285 multiple overloads with whitespace. Note that "" (the stringify
1286 overload) should be entered as \"\" (i.e. escaped).
1287 The FALLBACK: Keyword
1289 In addition to the OVERLOAD keyword, if you need to control how Perl
1290 autogenerates missing overloaded operators, you can set the FALLBACK
1291 keyword in the module header section, like this:
1292 MODULE = RPC PACKAGE = RPC
1294 FALLBACK: TRUE
1296 ...
1297 where FALLBACK can take any of the three values TRUE, FALSE, or UNDEF.
1299 If you do not set any FALLBACK value when using OVERLOAD, it defaults
1300 to UNDEF. FALLBACK is not used except when one or more functions using
1301 OVERLOAD have been defined. Please see "Fallback" in overload for more
1302 details.
1303 The INTERFACE: Keyword
1305 This keyword declares the current XSUB as a keeper of the given calling
1306 signature. If some text follows this keyword, it is considered as a
1307 list of functions which have this signature, and should be attached to
1308 the current XSUB.
1309 For example, if you have 4 C functions multiply(), divide(), add(),
1311 subtract() all having the signature:
1312 symbolic f(symbolic, symbolic);
1314 you can make them all to use the same XSUB using this:
1316 symbolic
1318 interface_s_ss(arg1, arg2)
1319 symbolic arg1
1320 symbolic arg2
1321 INTERFACE:
1322 multiply divide
1323 add subtract
1324 (This is the complete XSUB code for 4 Perl functions!) Four generated
1326 Perl function share names with corresponding C functions.
1327 The advantage of this approach comparing to ALIAS: keyword is that
1329 there is no need to code a switch statement, each Perl function (which
1330 shares the same XSUB) knows which C function it should call.
1331 Additionally, one can attach an extra function remainder() at runtime
1332 by using
1333 CV *mycv = newXSproto("Symbolic::remainder",
1335 XS_Symbolic_interface_s_ss, __FILE__, "$$");
1336 XSINTERFACE_FUNC_SET(mycv, remainder);
1337 say, from another XSUB. (This example supposes that there was no
1339 INTERFACE_MACRO: section, otherwise one needs to use something else
1340 instead of "XSINTERFACE_FUNC_SET", see the next section.)
1341 The INTERFACE_MACRO: Keyword
1343 This keyword allows one to define an INTERFACE using a different way to
1344 extract a function pointer from an XSUB. The text which follows this
1345 keyword should give the name of macros which would extract/set a
1346 function pointer. The extractor macro is given return type, "CV*", and
1347 "XSANY.any_dptr" for this "CV*". The setter macro is given cv, and the
1348 function pointer.
1349 The default value is "XSINTERFACE_FUNC" and "XSINTERFACE_FUNC_SET". An
1351 INTERFACE keyword with an empty list of functions can be omitted if
1352 INTERFACE_MACRO keyword is used.
1353 Suppose that in the previous example functions pointers for multiply(),
1355 divide(), add(), subtract() are kept in a global C array "fp[]" with
1356 offsets being "multiply_off", "divide_off", "add_off", "subtract_off".
1357 Then one can use
1358 #define XSINTERFACE_FUNC_BYOFFSET(ret,cv,f) \
1360 ((XSINTERFACE_CVT_ANON(ret))fp[CvXSUBANY(cv).any_i32])
1361 #define XSINTERFACE_FUNC_BYOFFSET_set(cv,f) \
1362 CvXSUBANY(cv).any_i32 = CAT2( f, _off )
1363 in C section,
1365 symbolic
1367 interface_s_ss(arg1, arg2)
1368 symbolic arg1
1369 symbolic arg2
1370 INTERFACE_MACRO:
1371 XSINTERFACE_FUNC_BYOFFSET
1372 XSINTERFACE_FUNC_BYOFFSET_set
1373 INTERFACE:
1374 multiply divide
1375 add subtract
1376 in XSUB section.
1378 The INCLUDE: Keyword
1380 This keyword can be used to pull other files into the XS module. The
1381 other files may have XS code. INCLUDE: can also be used to run a
1382 command to generate the XS code to be pulled into the module.
1383 The file Rpcb1.xsh contains our "rpcb_gettime()" function:
1385 bool_t
1387 rpcb_gettime(host,timep)
1388 char *host
1389 time_t &timep
1390 OUTPUT:
1391 timep
1392 The XS module can use INCLUDE: to pull that file into it.
1394 INCLUDE: Rpcb1.xsh
1396 If the parameters to the INCLUDE: keyword are followed by a pipe ("|")
1398 then the compiler will interpret the parameters as a command.
1399 INCLUDE: cat Rpcb1.xsh |
1401 The CASE: Keyword
1403 The CASE: keyword allows an XSUB to have multiple distinct parts with
1404 each part acting as a virtual XSUB. CASE: is greedy and if it is used
1405 then all other XS keywords must be contained within a CASE:. This
1406 means nothing may precede the first CASE: in the XSUB and anything
1407 following the last CASE: is included in that case.
1408 A CASE: might switch via a parameter of the XSUB, via the "ix" ALIAS:
1410 variable (see "The ALIAS: Keyword"), or maybe via the "items" variable
1411 (see "Variable-length Parameter Lists"). The last CASE: becomes the
1412 default case if it is not associated with a conditional. The following
1413 example shows CASE switched via "ix" with a function "rpcb_gettime()"
1414 having an alias "x_gettime()". When the function is called as
1415 "rpcb_gettime()" its parameters are the usual "(char *host, time_t
1416 *timep)", but when the function is called as "x_gettime()" its
1417 parameters are reversed, "(time_t *timep, char *host)".
1418 long
1420 rpcb_gettime(a,b)
1421 CASE: ix == 1
1422 ALIAS:
1423 x_gettime = 1
1424 INPUT:
1425 # 'a' is timep, 'b' is host
1426 char *b
1427 time_t a = NO_INIT
1428 CODE:
1429 RETVAL = rpcb_gettime( b, &a );
1430 OUTPUT:
1431 a
1432 RETVAL
1433 CASE:
1434 # 'a' is host, 'b' is timep
1435 char *a
1436 time_t &b = NO_INIT
1437 OUTPUT:
1438 b
1439 RETVAL
1440 That function can be called with either of the following statements.
1442 Note the different argument lists.
1443 $status = rpcb_gettime( $host, $timep );
1445 $status = x_gettime( $timep, $host );
1447 The & Unary Operator
1449 The "&" unary operator in the INPUT: section is used to tell xsubpp
1450 that it should convert a Perl value to/from C using the C type to the
1451 left of "&", but provide a pointer to this value when the C function is
1452 called.
1453 This is useful to avoid a CODE: block for a C function which takes a
1455 parameter by reference. Typically, the parameter should be not a
1456 pointer type (an "int" or "long" but not an "int*" or "long*").
1457 The following XSUB will generate incorrect C code. The xsubpp compiler
1459 will turn this into code which calls "rpcb_gettime()" with parameters
1460 "(char *host, time_t timep)", but the real "rpcb_gettime()" wants the
1461 "timep" parameter to be of type "time_t*" rather than "time_t".
1462 bool_t
1464 rpcb_gettime(host,timep)
1465 char *host
1466 time_t timep
1467 OUTPUT:
1468 timep
1469 That problem is corrected by using the "&" operator. The xsubpp
1471 compiler will now turn this into code which calls "rpcb_gettime()"
1472 correctly with parameters "(char *host, time_t *timep)". It does this
1473 by carrying the "&" through, so the function call looks like
1474 "rpcb_gettime(host, &timep)".
1475 bool_t
1477 rpcb_gettime(host,timep)
1478 char *host
1479 time_t &timep
1480 OUTPUT:
1481 timep
1482 Inserting POD, Comments and C Preprocessor Directives
1484 C preprocessor directives are allowed within BOOT:, PREINIT: INIT:,
1485 CODE:, PPCODE:, POSTCALL:, and CLEANUP: blocks, as well as outside the
1486 functions. Comments are allowed anywhere after the MODULE keyword.
1487 The compiler will pass the preprocessor directives through untouched
1488 and will remove the commented lines. POD documentation is allowed at
1489 any point, both in the C and XS language sections. POD must be
1490 terminated with a "=cut" command; "xsubpp" will exit with an error if
1491 it does not. It is very unlikely that human generated C code will be
1492 mistaken for POD, as most indenting styles result in whitespace in
1493 front of any line starting with "=". Machine generated XS files may
1494 fall into this trap unless care is taken to ensure that a space breaks
1495 the sequence "\n=".
1496 Comments can be added to XSUBs by placing a "#" as the first non-
1498 whitespace of a line. Care should be taken to avoid making the comment
1499 look like a C preprocessor directive, lest it be interpreted as such.
1500 The simplest way to prevent this is to put whitespace in front of the
1501 "#".
1502 If you use preprocessor directives to choose one of two versions of a
1504 function, use
1505 #if ... version1
1507 #else /* ... version2 */
1508 #endif
1509 and not
1511 #if ... version1
1513 #endif
1514 #if ... version2
1515 #endif
1516 because otherwise xsubpp will believe that you made a duplicate
1518 definition of the function. Also, put a blank line before the
1519 #else/#endif so it will not be seen as part of the function body.
1520 Using XS With C++
1522 If an XSUB name contains "::", it is considered to be a C++ method.
1523 The generated Perl function will assume that its first argument is an
1524 object pointer. The object pointer will be stored in a variable called
1525 THIS. The object should have been created by C++ with the new()
1526 function and should be blessed by Perl with the sv_setref_pv() macro.
1527 The blessing of the object by Perl can be handled by a typemap. An
1528 example typemap is shown at the end of this section.
1529 If the return type of the XSUB includes "static", the method is
1531 considered to be a static method. It will call the C++ function using
1532 the class::method() syntax. If the method is not static the function
1533 will be called using the THIS->method() syntax.
1534 The next examples will use the following C++ class.
1536 class color {
1538 public:
1539 color();
1540 ~color();
1541 int blue();
1542 void set_blue( int );
1543 private:
1545 int c_blue;
1546 };
1547 The XSUBs for the blue() and set_blue() methods are defined with the
1549 class name but the parameter for the object (THIS, or "self") is
1550 implicit and is not listed.
1551 int
1553 color::blue()
1554 void
1556 color::set_blue( val )
1557 int val
1558 Both Perl functions will expect an object as the first parameter. In
1560 the generated C++ code the object is called "THIS", and the method call
1561 will be performed on this object. So in the C++ code the blue() and
1562 set_blue() methods will be called as this:
1563 RETVAL = THIS->blue();
1565 THIS->set_blue( val );
1567 You could also write a single get/set method using an optional
1569 argument:
1570 int
1572 color::blue( val = NO_INIT )
1573 int val
1574 PROTOTYPE $;$
1575 CODE:
1576 if (items > 1)
1577 THIS->set_blue( val );
1578 RETVAL = THIS->blue();
1579 OUTPUT:
1580 RETVAL
1581 If the function's name is DESTROY then the C++ "delete" function will
1583 be called and "THIS" will be given as its parameter. The generated C++
1584 code for
1585 void
1587 color::DESTROY()
1588 will look like this:
1590 color *THIS = ...; // Initialized as in typemap
1592 delete THIS;
1594 If the function's name is new then the C++ "new" function will be
1596 called to create a dynamic C++ object. The XSUB will expect the class
1597 name, which will be kept in a variable called "CLASS", to be given as
1598 the first argument.
1599 color *
1601 color::new()
1602 The generated C++ code will call "new".
1604 RETVAL = new color();
1606 The following is an example of a typemap that could be used for this
1608 C++ example.
1609 TYPEMAP
1611 color * O_OBJECT
1612 OUTPUT
1614 # The Perl object is blessed into 'CLASS', which should be a
1615 # char* having the name of the package for the blessing.
1616 O_OBJECT
1617 sv_setref_pv( $arg, CLASS, (void*)$var );
1618 INPUT
1620 O_OBJECT
1621 if( sv_isobject($arg) && (SvTYPE(SvRV($arg)) == SVt_PVMG) )
1622 $var = ($type)SvIV((SV*)SvRV( $arg ));
1623 else{
1624 warn( \"${Package}::$func_name() -- $var is not a blessed SV reference\" );
1625 XSRETURN_UNDEF;
1626 }
1627 Interface Strategy
1629 When designing an interface between Perl and a C library a straight
1630 translation from C to XS (such as created by "h2xs -x") is often
1631 sufficient. However, sometimes the interface will look very C-like and
1632 occasionally nonintuitive, especially when the C function modifies one
1633 of its parameters, or returns failure inband (as in "negative return
1634 values mean failure"). In cases where the programmer wishes to create
1635 a more Perl-like interface the following strategy may help to identify
1636 the more critical parts of the interface.
1637 Identify the C functions with input/output or output parameters. The
1639 XSUBs for these functions may be able to return lists to Perl.
1640 Identify the C functions which use some inband info as an indication of
1642 failure. They may be candidates to return undef or an empty list in
1643 case of failure. If the failure may be detected without a call to the
1644 C function, you may want to use an INIT: section to report the failure.
1645 For failures detectable after the C function returns one may want to
1646 use a POSTCALL: section to process the failure. In more complicated
1647 cases use CODE: or PPCODE: sections.
1648 If many functions use the same failure indication based on the return
1650 value, you may want to create a special typedef to handle this
1651 situation. Put
1652 typedef int negative_is_failure;
1654 near the beginning of XS file, and create an OUTPUT typemap entry for
1656 "negative_is_failure" which converts negative values to "undef", or
1657 maybe croak()s. After this the return value of type
1658 "negative_is_failure" will create more Perl-like interface.
1659 Identify which values are used by only the C and XSUB functions
1661 themselves, say, when a parameter to a function should be a contents of
1662 a global variable. If Perl does not need to access the contents of the
1663 value then it may not be necessary to provide a translation for that
1664 value from C to Perl.
1665 Identify the pointers in the C function parameter lists and return
1667 values. Some pointers may be used to implement input/output or output
1668 parameters, they can be handled in XS with the "&" unary operator, and,
1669 possibly, using the NO_INIT keyword. Some others will require handling
1670 of types like "int *", and one needs to decide what a useful Perl
1671 translation will do in such a case. When the semantic is clear, it is
1672 advisable to put the translation into a typemap file.
1673 Identify the structures used by the C functions. In many cases it may
1675 be helpful to use the T_PTROBJ typemap for these structures so they can
1676 be manipulated by Perl as blessed objects. (This is handled
1677 automatically by "h2xs -x".)
1678 If the same C type is used in several different contexts which require
1680 different translations, "typedef" several new types mapped to this C
1681 type, and create separate typemap entries for these new types. Use
1682 these types in declarations of return type and parameters to XSUBs.
1683 Perl Objects And C Structures
1685 When dealing with C structures one should select either T_PTROBJ or
1686 T_PTRREF for the XS type. Both types are designed to handle pointers
1687 to complex objects. The T_PTRREF type will allow the Perl object to be
1688 unblessed while the T_PTROBJ type requires that the object be blessed.
1689 By using T_PTROBJ one can achieve a form of type-checking because the
1690 XSUB will attempt to verify that the Perl object is of the expected
1691 type.
1692 The following XS code shows the getnetconfigent() function which is
1694 used with ONC+ TIRPC. The getnetconfigent() function will return a
1695 pointer to a C structure and has the C prototype shown below. The
1696 example will demonstrate how the C pointer will become a Perl
1697 reference. Perl will consider this reference to be a pointer to a
1698 blessed object and will attempt to call a destructor for the object. A
1699 destructor will be provided in the XS source to free the memory used by
1700 getnetconfigent(). Destructors in XS can be created by specifying an
1701 XSUB function whose name ends with the word DESTROY. XS destructors
1702 can be used to free memory which may have been malloc'd by another
1703 XSUB.
1704 struct netconfig *getnetconfigent(const char *netid);
1706 A "typedef" will be created for "struct netconfig". The Perl object
1708 will be blessed in a class matching the name of the C type, with the
1709 tag "Ptr" appended, and the name should not have embedded spaces if it
1710 will be a Perl package name. The destructor will be placed in a class
1711 corresponding to the class of the object and the PREFIX keyword will be
1712 used to trim the name to the word DESTROY as Perl will expect.
1713 typedef struct netconfig Netconfig;
1715 MODULE = RPC PACKAGE = RPC
1717 Netconfig *
1719 getnetconfigent(netid)
1720 char *netid
1721 MODULE = RPC PACKAGE = NetconfigPtr PREFIX = rpcb_
1723 void
1725 rpcb_DESTROY(netconf)
1726 Netconfig *netconf
1727 CODE:
1728 printf("Now in NetconfigPtr::DESTROY\n");
1729 free( netconf );
1730 This example requires the following typemap entry. Consult the typemap
1732 section for more information about adding new typemaps for an
1733 extension.
1734 TYPEMAP
1736 Netconfig * T_PTROBJ
1737 This example will be used with the following Perl statements.
1739 use RPC;
1741 $netconf = getnetconfigent("udp");
1742 When Perl destroys the object referenced by $netconf it will send the
1744 object to the supplied XSUB DESTROY function. Perl cannot determine,
1745 and does not care, that this object is a C struct and not a Perl
1746 object. In this sense, there is no difference between the object
1747 created by the getnetconfigent() XSUB and an object created by a normal
1748 Perl subroutine.
1749 The Typemap
1751 The typemap is a collection of code fragments which are used by the
1752 xsubpp compiler to map C function parameters and values to Perl values.
1753 The typemap file may consist of three sections labelled "TYPEMAP",
1754 "INPUT", and "OUTPUT". An unlabelled initial section is assumed to be
1755 a "TYPEMAP" section. The INPUT section tells the compiler how to
1756 translate Perl values into variables of certain C types. The OUTPUT
1757 section tells the compiler how to translate the values from certain C
1758 types into values Perl can understand. The TYPEMAP section tells the
1759 compiler which of the INPUT and OUTPUT code fragments should be used to
1760 map a given C type to a Perl value. The section labels "TYPEMAP",
1761 "INPUT", or "OUTPUT" must begin in the first column on a line by
1762 themselves, and must be in uppercase.
1763 The default typemap in the "lib/ExtUtils" directory of the Perl source
1765 contains many useful types which can be used by Perl extensions. Some
1766 extensions define additional typemaps which they keep in their own
1767 directory. These additional typemaps may reference INPUT and OUTPUT
1768 maps in the main typemap. The xsubpp compiler will allow the
1769 extension's own typemap to override any mappings which are in the
1770 default typemap.
1771 Most extensions which require a custom typemap will need only the
1773 TYPEMAP section of the typemap file. The custom typemap used in the
1774 getnetconfigent() example shown earlier demonstrates what may be the
1775 typical use of extension typemaps. That typemap is used to equate a C
1776 structure with the T_PTROBJ typemap. The typemap used by
1777 getnetconfigent() is shown here. Note that the C type is separated
1778 from the XS type with a tab and that the C unary operator "*" is
1779 considered to be a part of the C type name.
1780 TYPEMAP
1782 Netconfig *<tab>T_PTROBJ
1783 Here's a more complicated example: suppose that you wanted "struct
1785 netconfig" to be blessed into the class "Net::Config". One way to do
1786 this is to use underscores (_) to separate package names, as follows:
1787 typedef struct netconfig * Net_Config;
1789 And then provide a typemap entry "T_PTROBJ_SPECIAL" that maps
1791 underscores to double-colons (::), and declare "Net_Config" to be of
1792 that type:
1793 TYPEMAP
1795 Net_Config T_PTROBJ_SPECIAL
1796 INPUT
1798 T_PTROBJ_SPECIAL
1799 if (sv_derived_from($arg, \"${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\")) {
1800 IV tmp = SvIV((SV*)SvRV($arg));
1801 $var = INT2PTR($type, tmp);
1802 }
1803 else
1804 croak(\"$var is not of type ${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\")
1805 OUTPUT
1807 T_PTROBJ_SPECIAL
1808 sv_setref_pv($arg, \"${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\",
1809 (void*)$var);
1810 The INPUT and OUTPUT sections substitute underscores for double-colons
1812 on the fly, giving the desired effect. This example demonstrates some
1813 of the power and versatility of the typemap facility.
1814 The INT2PTR macro (defined in perl.h) casts an integer to a pointer, of
1816 a given type, taking care of the possible different size of integers
1817 and pointers. There are also PTR2IV, PTR2UV, PTR2NV macros, to map the
1818 other way, which may be useful in OUTPUT sections.
1819 Safely Storing Static Data in XS
1821 Starting with Perl 5.8, a macro framework has been defined to allow
1822 static data to be safely stored in XS modules that will be accessed
1823 from a multi-threaded Perl.
1824 Although primarily designed for use with multi-threaded Perl, the
1826 macros have been designed so that they will work with non-threaded Perl
1827 as well.
1828 It is therefore strongly recommended that these macros be used by all
1830 XS modules that make use of static data.
1831 The easiest way to get a template set of macros to use is by specifying
1833 the "-g" ("--global") option with h2xs (see h2xs).
1834 Below is an example module that makes use of the macros.
1836 #include "EXTERN.h"
1838 #include "perl.h"
1839 #include "XSUB.h"
1840 /* Global Data */
1842 #define MY_CXT_KEY "BlindMice::_guts" XS_VERSION
1844 typedef struct {
1846 int count;
1847 char name[3][100];
1848 } my_cxt_t;
1849 START_MY_CXT
1851 MODULE = BlindMice PACKAGE = BlindMice
1853 BOOT:
1855 {
1856 MY_CXT_INIT;
1857 MY_CXT.count = 0;
1858 strcpy(MY_CXT.name[0], "None");
1859 strcpy(MY_CXT.name[1], "None");
1860 strcpy(MY_CXT.name[2], "None");
1861 }
1862 int
1864 newMouse(char * name)
1865 char * name;
1866 PREINIT:
1867 dMY_CXT;
1868 CODE:
1869 if (MY_CXT.count >= 3) {
1870 warn("Already have 3 blind mice");
1871 RETVAL = 0;
1872 }
1873 else {
1874 RETVAL = ++ MY_CXT.count;
1875 strcpy(MY_CXT.name[MY_CXT.count - 1], name);
1876 }
1877 char *
1879 get_mouse_name(index)
1880 int index
1881 CODE:
1882 dMY_CXT;
1883 RETVAL = MY_CXT.lives ++;
1884 if (index > MY_CXT.count)
1885 croak("There are only 3 blind mice.");
1886 else
1887 RETVAL = newSVpv(MY_CXT.name[index - 1]);
1888 void
1890 CLONE(...)
1891 CODE:
1892 MY_CXT_CLONE;
1893 REFERENCE
1895 MY_CXT_KEY
1897 This macro is used to define a unique key to refer to the static
1898 data for an XS module. The suggested naming scheme, as used by
1899 h2xs, is to use a string that consists of the module name, the
1900 string "::_guts" and the module version number.
1901 #define MY_CXT_KEY "MyModule::_guts" XS_VERSION
1903 typedef my_cxt_t
1905 This struct typedef must always be called "my_cxt_t" -- the other
1906 "CXT*" macros assume the existence of the "my_cxt_t" typedef name.
1907 Declare a typedef named "my_cxt_t" that is a structure that
1909 contains all the data that needs to be interpreter-local.
1910 typedef struct {
1912 int some_value;
1913 } my_cxt_t;
1914 START_MY_CXT
1916 Always place the START_MY_CXT macro directly after the declaration
1917 of "my_cxt_t".
1918 MY_CXT_INIT
1920 The MY_CXT_INIT macro initialises storage for the "my_cxt_t"
1921 struct.
1922 It must be called exactly once -- typically in a BOOT: section. If
1924 you are maintaining multiple interpreters, it should be called
1925 once in each interpreter instance, except for interpreters cloned
1926 from existing ones. (But see "MY_CXT_CLONE" below.)
1927 dMY_CXT
1929 Use the dMY_CXT macro (a declaration) in all the functions that
1930 access MY_CXT.
1931 MY_CXT
1933 Use the MY_CXT macro to access members of the "my_cxt_t" struct.
1934 For example, if "my_cxt_t" is
1935 typedef struct {
1937 int index;
1938 } my_cxt_t;
1939 then use this to access the "index" member
1941 dMY_CXT;
1943 MY_CXT.index = 2;
1944 aMY_CXT/pMY_CXT
1946 "dMY_CXT" may be quite expensive to calculate, and to avoid the
1947 overhead of invoking it in each function it is possible to pass
1948 the declaration onto other functions using the "aMY_CXT"/"pMY_CXT"
1949 macros, eg
1950 void sub1() {
1952 dMY_CXT;
1953 MY_CXT.index = 1;
1954 sub2(aMY_CXT);
1955 }
1956 void sub2(pMY_CXT) {
1958 MY_CXT.index = 2;
1959 }
1960 Analogously to "pTHX", there are equivalent forms for when the
1962 macro is the first or last in multiple arguments, where an
1963 underscore represents a comma, i.e. "_aMY_CXT", "aMY_CXT_",
1964 "_pMY_CXT" and "pMY_CXT_".
1965 MY_CXT_CLONE
1967 By default, when a new interpreter is created as a copy of an
1968 existing one (eg via "threads->create()"), both interpreters share
1969 the same physical my_cxt_t structure. Calling "MY_CXT_CLONE"
1970 (typically via the package's "CLONE()" function), causes a byte-
1971 for-byte copy of the structure to be taken, and any future dMY_CXT
1972 will cause the copy to be accessed instead.
1973 MY_CXT_INIT_INTERP(my_perl)
1975 dMY_CXT_INTERP(my_perl)
1976 These are versions of the macros which take an explicit
1977 interpreter as an argument.
1978 Note that these macros will only work together within the same source
1980 file; that is, a dMY_CTX in one source file will access a different
1981 structure than a dMY_CTX in another source file.
1982 Thread-aware system interfaces
1984 Starting from Perl 5.8, in C/C++ level Perl knows how to wrap
1985 system/library interfaces that have thread-aware versions (e.g.
1986 getpwent_r()) into frontend macros (e.g. getpwent()) that correctly
1987 handle the multithreaded interaction with the Perl interpreter. This
1988 will happen transparently, the only thing you need to do is to
1989 instantiate a Perl interpreter.
1990 This wrapping happens always when compiling Perl core source (PERL_CORE
1992 is defined) or the Perl core extensions (PERL_EXT is defined). When
1993 compiling XS code outside of Perl core the wrapping does not take
1994 place. Note, however, that intermixing the _r-forms (as Perl compiled
1995 for multithreaded operation will do) and the _r-less forms is neither
1996 well-defined (inconsistent results, data corruption, or even crashes
1997 become more likely), nor is it very portable.
1998
2000 File "RPC.xs": Interface to some ONC+ RPC bind library functions.
2001 #include "EXTERN.h"
2003 #include "perl.h"
2004 #include "XSUB.h"
2005 #include <rpc/rpc.h>
2007 typedef struct netconfig Netconfig;
2009 MODULE = RPC PACKAGE = RPC
2011 SV *
2013 rpcb_gettime(host="localhost")
2014 char *host
2015 PREINIT:
2016 time_t timep;
2017 CODE:
2018 ST(0) = sv_newmortal();
2019 if( rpcb_gettime( host, &timep ) )
2020 sv_setnv( ST(0), (double)timep );
2021 Netconfig *
2023 getnetconfigent(netid="udp")
2024 char *netid
2025 MODULE = RPC PACKAGE = NetconfigPtr PREFIX = rpcb_
2027 void
2029 rpcb_DESTROY(netconf)
2030 Netconfig *netconf
2031 CODE:
2032 printf("NetconfigPtr::DESTROY\n");
2033 free( netconf );
2034 File "typemap": Custom typemap for RPC.xs.
2036 TYPEMAP
2038 Netconfig * T_PTROBJ
2039 File "RPC.pm": Perl module for the RPC extension.
2041 package RPC;
2043 require Exporter;
2045 require DynaLoader;
2046 @ISA = qw(Exporter DynaLoader);
2047 @EXPORT = qw(rpcb_gettime getnetconfigent);
2048 bootstrap RPC;
2050 1;
2051 File "rpctest.pl": Perl test program for the RPC extension.
2053 use RPC;
2055 $netconf = getnetconfigent();
2057 $a = rpcb_gettime();
2058 print "time = $a\n";
2059 print "netconf = $netconf\n";
2060 $netconf = getnetconfigent("tcp");
2062 $a = rpcb_gettime("poplar");
2063 print "time = $a\n";
2064 print "netconf = $netconf\n";
2065
2067 This document covers features supported by "xsubpp" 1.935.
2068
2070 Originally written by Dean Roehrich <roehrich@cray.com>.
2071 Maintained since 1996 by The Perl Porters <perlbug@perl.org>.
2073perl v5.10.1 2009-07-27 PERLXS(1)