3 perlxs - XS language reference manual
9 XS is an interface description file format used to create an extension
10 interface between Perl and C code (or a C library) which one wishes
11 to use with Perl. The XS interface is combined with the library to
12 create a new library which can then be either dynamically loaded
13 or statically linked into perl. The XS interface description is
14 written in the XS language and is the core component of the Perl
17 An B<XSUB> forms the basic unit of the XS interface. After compilation
18 by the B<xsubpp> compiler, each XSUB amounts to a C function definition
19 which will provide the glue between Perl calling conventions and C
22 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 function,
24 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
28 Perl stack, or by modifying the arguments supplied from the Perl side.
30 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-oriented
38 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
43 a more concise short-hand I<description> of what should be done by
44 the glue, and let the XS compiler B<xsubpp> handle the rest.
46 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 cases
50 when the C interface coincides with the Perl interface, the XSUB
51 declaration is almost identical to a declaration of a C function (in K&R
52 style). In such circumstances, there is another tool called C<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
57 The XS compiler is called B<xsubpp>. This compiler creates
58 the constructs necessary to let an XSUB manipulate Perl values, and
59 creates the glue necessary to let Perl call the XSUB. The compiler
60 uses B<typemaps> to determine how to map C function parameters
61 and output values to Perl values and back. The default typemap
62 (which comes with Perl) handles many common C types. A supplementary
63 typemap may also be needed to handle any special structures and types
64 for the library being linked.
66 A file in XS format starts with a C language section which goes until the
67 first C<MODULE =Z<>> directive. Other XS directives and XSUB definitions
68 may follow this line. The "language" used in this part of the file
69 is usually referred to as the XS language. B<xsubpp> recognizes and
70 skips POD (see L<perlpod>) in both the C and XS language sections, which
71 allows the XS file to contain embedded documentation.
73 See L<perlxstut> for a tutorial on the whole extension creation process.
75 Note: For some extensions, Dave Beazley's SWIG system may provide a
76 significantly more convenient mechanism for creating the extension
77 glue code. See http://www.swig.org/ for more information.
81 Many of the examples which follow will concentrate on creating an interface
82 between Perl and the ONC+ RPC bind library functions. The rpcb_gettime()
83 function is used to demonstrate many features of the XS language. This
84 function has two parameters; the first is an input parameter and the second
85 is an output parameter. The function also returns a status value.
87 bool_t rpcb_gettime(const char *host, time_t *timep);
89 From C this function will be called with the following
95 status = rpcb_gettime( "localhost", &timep );
97 If an XSUB is created to offer a direct translation between this function
98 and Perl, then this XSUB will be used from Perl with the following code.
99 The $status and $timep variables will contain the output of the function.
102 $status = rpcb_gettime( "localhost", $timep );
104 The following XS file shows an XS subroutine, or XSUB, which
105 demonstrates one possible interface to the rpcb_gettime()
106 function. This XSUB represents a direct translation between
107 C and Perl and so preserves the interface even from Perl.
108 This XSUB will be invoked from Perl with the usage shown
109 above. Note that the first three #include statements, for
110 C<EXTERN.h>, C<perl.h>, and C<XSUB.h>, will always be present at the
111 beginning of an XS file. This approach and others will be
112 expanded later in this document.
119 MODULE = RPC PACKAGE = RPC
122 rpcb_gettime(host,timep)
128 Any extension to Perl, including those containing XSUBs,
129 should have a Perl module to serve as the bootstrap which
130 pulls the extension into Perl. This module will export the
131 extension's functions and variables to the Perl program and
132 will cause the extension's XSUBs to be linked into Perl.
133 The following module will be used for most of the examples
134 in this document and should be used from Perl with the C<use>
135 command as shown earlier. Perl modules are explained in
136 more detail later in this document.
142 @ISA = qw(Exporter DynaLoader);
143 @EXPORT = qw( rpcb_gettime );
148 Throughout this document a variety of interfaces to the rpcb_gettime()
149 XSUB will be explored. The XSUBs will take their parameters in different
150 orders or will take different numbers of parameters. In each case the
151 XSUB is an abstraction between Perl and the real C rpcb_gettime()
152 function, and the XSUB must always ensure that the real rpcb_gettime()
153 function is called with the correct parameters. This abstraction will
154 allow the programmer to create a more Perl-like interface to the C
157 =head2 The Anatomy of an XSUB
159 The simplest XSUBs consist of 3 parts: a description of the return
160 value, the name of the XSUB routine and the names of its arguments,
161 and a description of types or formats of the arguments.
163 The following XSUB allows a Perl program to access a C library function
164 called sin(). The XSUB will imitate the C function which takes a single
165 argument and returns a single value.
171 Optionally, one can merge the description of types and the list of
172 argument names, rewriting this as
177 This makes this XSUB look similar to an ANSI C declaration. An optional
178 semicolon is allowed after the argument list, as in
183 Parameters with C pointer types can have different semantic: C functions
184 with similar declarations
186 bool string_looks_as_a_number(char *s);
187 bool make_char_uppercase(char *c);
189 are used in absolutely incompatible manner. Parameters to these functions
190 could be described B<xsubpp> like this:
195 Both these XS declarations correspond to the C<char*> C type, but they have
196 different semantics, see L<"The & Unary Operator">.
198 It is convenient to think that the indirection operator
199 C<*> should be considered as a part of the type and the address operator C<&>
200 should be considered part of the variable. See L<"The Typemap">
201 for more info about handling qualifiers and unary operators in C types.
203 The function name and the return type must be placed on
204 separate lines and should be flush left-adjusted.
212 The rest of the function description may be indented or left-adjusted. The
213 following example shows a function with its body left-adjusted. Most
214 examples in this document will indent the body for better readability.
222 More complicated XSUBs may contain many other sections. Each section of
223 an XSUB starts with the corresponding keyword, such as INIT: or CLEANUP:.
224 However, the first two lines of an XSUB always contain the same data:
225 descriptions of the return type and the names of the function and its
226 parameters. Whatever immediately follows these is considered to be
227 an INPUT: section unless explicitly marked with another keyword.
228 (See L<The INPUT: Keyword>.)
230 An XSUB section continues until another section-start keyword is found.
232 =head2 The Argument Stack
234 The Perl argument stack is used to store the values which are
235 sent as parameters to the XSUB and to store the XSUB's
236 return value(s). In reality all Perl functions (including non-XSUB
237 ones) keep their values on this stack all the same time, each limited
238 to its own range of positions on the stack. In this document the
239 first position on that stack which belongs to the active
240 function will be referred to as position 0 for that function.
242 XSUBs refer to their stack arguments with the macro B<ST(x)>, where I<x>
243 refers to a position in this XSUB's part of the stack. Position 0 for that
244 function would be known to the XSUB as ST(0). The XSUB's incoming
245 parameters and outgoing return values always begin at ST(0). For many
246 simple cases the B<xsubpp> compiler will generate the code necessary to
247 handle the argument stack by embedding code fragments found in the
248 typemaps. In more complex cases the programmer must supply the code.
250 =head2 The RETVAL Variable
252 The RETVAL variable is a special C variable that is declared automatically
253 for you. The C type of RETVAL matches the return type of the C library
254 function. The B<xsubpp> compiler will declare this variable in each XSUB
255 with non-C<void> return type. By default the generated C function
256 will use RETVAL to hold the return value of the C library function being
257 called. In simple cases the value of RETVAL will be placed in ST(0) of
258 the argument stack where it can be received by Perl as the return value
261 If the XSUB has a return type of C<void> then the compiler will
262 not declare a RETVAL variable for that function. When using
263 a PPCODE: section no manipulation of the RETVAL variable is required, the
264 section may use direct stack manipulation to place output values on the stack.
266 If PPCODE: directive is not used, C<void> return value should be used
267 only for subroutines which do not return a value, I<even if> CODE:
268 directive is used which sets ST(0) explicitly.
270 Older versions of this document recommended to use C<void> return
271 value in such cases. It was discovered that this could lead to
272 segfaults in cases when XSUB was I<truly> C<void>. This practice is
273 now deprecated, and may be not supported at some future version. Use
274 the return value C<SV *> in such cases. (Currently C<xsubpp> contains
275 some heuristic code which tries to disambiguate between "truly-void"
276 and "old-practice-declared-as-void" functions. Hence your code is at
277 mercy of this heuristics unless you use C<SV *> as return value.)
279 =head2 Returning SVs, AVs and HVs through RETVAL
281 When you're using RETVAL to return an C<SV *>, there's some magic
282 going on behind the scenes that should be mentioned. When you're
283 manipulating the argument stack using the ST(x) macro, for example,
284 you usually have to pay special attention to reference counts. (For
285 more about reference counts, see L<perlguts>.) To make your life
286 easier, the typemap file automatically makes C<RETVAL> mortal when
287 you're returning an C<SV *>. Thus, the following two XSUBs are more
293 ST(0) = newSVpv("Hello World",0);
300 RETVAL = newSVpv("Hello World",0);
304 This is quite useful as it usually improves readability. While
305 this works fine for an C<SV *>, it's unfortunately not as easy
306 to have C<AV *> or C<HV *> as a return value. You I<should> be
313 /* do something with RETVAL */
317 But due to an unfixable bug (fixing it would break lots of existing
318 CPAN modules) in the typemap file, the reference count of the C<AV *>
319 is not properly decremented. Thus, the above XSUB would leak memory
320 whenever it is being called. The same problem exists for C<HV *>.
322 When you're returning an C<AV *> or a C<HV *>, you have make sure
323 their reference count is decremented by making the AV or HV mortal:
329 sv_2mortal((SV*)RETVAL);
330 /* do something with RETVAL */
334 And also remember that you don't have to do this for an C<SV *>.
336 =head2 The MODULE Keyword
338 The MODULE keyword is used to start the XS code and to specify the package
339 of the functions which are being defined. All text preceding the first
340 MODULE keyword is considered C code and is passed through to the output with
341 POD stripped, but otherwise untouched. Every XS module will have a
342 bootstrap function which is used to hook the XSUBs into Perl. The package
343 name of this bootstrap function will match the value of the last MODULE
344 statement in the XS source files. The value of MODULE should always remain
345 constant within the same XS file, though this is not required.
347 The following example will start the XS code and will place
348 all functions in a package named RPC.
352 =head2 The PACKAGE Keyword
354 When functions within an XS source file must be separated into packages
355 the PACKAGE keyword should be used. This keyword is used with the MODULE
356 keyword and must follow immediately after it when used.
358 MODULE = RPC PACKAGE = RPC
360 [ XS code in package RPC ]
362 MODULE = RPC PACKAGE = RPCB
364 [ XS code in package RPCB ]
366 MODULE = RPC PACKAGE = RPC
368 [ XS code in package RPC ]
370 The same package name can be used more than once, allowing for
371 non-contiguous code. This is useful if you have a stronger ordering
372 principle than package names.
374 Although this keyword is optional and in some cases provides redundant
375 information it should always be used. This keyword will ensure that the
376 XSUBs appear in the desired package.
378 =head2 The PREFIX Keyword
380 The PREFIX keyword designates prefixes which should be
381 removed from the Perl function names. If the C function is
382 C<rpcb_gettime()> and the PREFIX value is C<rpcb_> then Perl will
383 see this function as C<gettime()>.
385 This keyword should follow the PACKAGE keyword when used.
386 If PACKAGE is not used then PREFIX should follow the MODULE
389 MODULE = RPC PREFIX = rpc_
391 MODULE = RPC PACKAGE = RPCB PREFIX = rpcb_
393 =head2 The OUTPUT: Keyword
395 The OUTPUT: keyword indicates that certain function parameters should be
396 updated (new values made visible to Perl) when the XSUB terminates or that
397 certain values should be returned to the calling Perl function. For
398 simple functions which have no CODE: or PPCODE: section,
399 such as the sin() function above, the RETVAL variable is
400 automatically designated as an output value. For more complex functions
401 the B<xsubpp> compiler will need help to determine which variables are output
404 This keyword will normally be used to complement the CODE: keyword.
405 The RETVAL variable is not recognized as an output variable when the
406 CODE: keyword is present. The OUTPUT: keyword is used in this
407 situation to tell the compiler that RETVAL really is an output
410 The OUTPUT: keyword can also be used to indicate that function parameters
411 are output variables. This may be necessary when a parameter has been
412 modified within the function and the programmer would like the update to
416 rpcb_gettime(host,timep)
422 The OUTPUT: keyword will also allow an output parameter to
423 be mapped to a matching piece of code rather than to a
427 rpcb_gettime(host,timep)
431 timep sv_setnv(ST(1), (double)timep);
433 B<xsubpp> emits an automatic C<SvSETMAGIC()> for all parameters in the
434 OUTPUT section of the XSUB, except RETVAL. This is the usually desired
435 behavior, as it takes care of properly invoking 'set' magic on output
436 parameters (needed for hash or array element parameters that must be
437 created if they didn't exist). If for some reason, this behavior is
438 not desired, the OUTPUT section may contain a C<SETMAGIC: DISABLE> line
439 to disable it for the remainder of the parameters in the OUTPUT section.
440 Likewise, C<SETMAGIC: ENABLE> can be used to reenable it for the
441 remainder of the OUTPUT section. See L<perlguts> for more details
444 =head2 The NO_OUTPUT Keyword
446 The NO_OUTPUT can be placed as the first token of the XSUB. This keyword
447 indicates that while the C subroutine we provide an interface to has
448 a non-C<void> return type, the return value of this C subroutine should not
449 be returned from the generated Perl subroutine.
451 With this keyword present L<The RETVAL Variable> is created, and in the
452 generated call to the subroutine this variable is assigned to, but the value
453 of this variable is not going to be used in the auto-generated code.
455 This keyword makes sense only if C<RETVAL> is going to be accessed by the
456 user-supplied code. It is especially useful to make a function interface
457 more Perl-like, especially when the C return value is just an error condition
458 indicator. For example,
461 delete_file(char *name)
464 croak("Error %d while deleting file '%s'", RETVAL, name);
466 Here the generated XS function returns nothing on success, and will die()
467 with a meaningful error message on error.
469 =head2 The CODE: Keyword
471 This keyword is used in more complicated XSUBs which require
472 special handling for the C function. The RETVAL variable is
473 still declared, but it will not be returned unless it is specified
474 in the OUTPUT: section.
476 The following XSUB is for a C function which requires special handling of
477 its parameters. The Perl usage is given first.
479 $status = rpcb_gettime( "localhost", $timep );
484 rpcb_gettime(host,timep)
488 RETVAL = rpcb_gettime( host, &timep );
493 =head2 The INIT: Keyword
495 The INIT: keyword allows initialization to be inserted into the XSUB before
496 the compiler generates the call to the C function. Unlike the CODE: keyword
497 above, this keyword does not affect the way the compiler handles RETVAL.
500 rpcb_gettime(host,timep)
504 printf("# Host is %s\n", host );
508 Another use for the INIT: section is to check for preconditions before
509 making a call to the C function:
516 if (a == 0 && b == 0)
519 croak("lldiv: cannot divide by 0");
521 =head2 The NO_INIT Keyword
523 The NO_INIT keyword is used to indicate that a function
524 parameter is being used only as an output value. The B<xsubpp>
525 compiler will normally generate code to read the values of
526 all function parameters from the argument stack and assign
527 them to C variables upon entry to the function. NO_INIT
528 will tell the compiler that some parameters will be used for
529 output rather than for input and that they will be handled
530 before the function terminates.
532 The following example shows a variation of the rpcb_gettime() function.
533 This function uses the timep variable only as an output variable and does
534 not care about its initial contents.
537 rpcb_gettime(host,timep)
539 time_t &timep = NO_INIT
543 =head2 Initializing Function Parameters
545 C function parameters are normally initialized with their values from
546 the argument stack (which in turn contains the parameters that were
547 passed to the XSUB from Perl). The typemaps contain the
548 code segments which are used to translate the Perl values to
549 the C parameters. The programmer, however, is allowed to
550 override the typemaps and supply alternate (or additional)
551 initialization code. Initialization code starts with the first
552 C<=>, C<;> or C<+> on a line in the INPUT: section. The only
553 exception happens if this C<;> terminates the line, then this C<;>
556 The following code demonstrates how to supply initialization code for
557 function parameters. The initialization code is eval'd within double
558 quotes by the compiler before it is added to the output so anything
559 which should be interpreted literally [mainly C<$>, C<@>, or C<\\>]
560 must be protected with backslashes. The variables $var, $arg,
561 and $type can be used as in typemaps.
564 rpcb_gettime(host,timep)
565 char *host = (char *)SvPV($arg,PL_na);
570 This should not be used to supply default values for parameters. One
571 would normally use this when a function parameter must be processed by
572 another library function before it can be used. Default parameters are
573 covered in the next section.
575 If the initialization begins with C<=>, then it is output in
576 the declaration for the input variable, replacing the initialization
577 supplied by the typemap. If the initialization
578 begins with C<;> or C<+>, then it is performed after
579 all of the input variables have been declared. In the C<;>
580 case the initialization normally supplied by the typemap is not performed.
581 For the C<+> case, the declaration for the variable will include the
582 initialization from the typemap. A global
583 variable, C<%v>, is available for the truly rare case where
584 information from one initialization is needed in another
587 Here's a truly obscure example:
590 rpcb_gettime(host,timep)
591 time_t &timep ; /* \$v{timep}=@{[$v{timep}=$arg]} */
592 char *host + SvOK($v{timep}) ? SvPV($arg,PL_na) : NULL;
596 The construct C<\$v{timep}=@{[$v{timep}=$arg]}> used in the above
597 example has a two-fold purpose: first, when this line is processed by
598 B<xsubpp>, the Perl snippet C<$v{timep}=$arg> is evaluated. Second,
599 the text of the evaluated snippet is output into the generated C file
600 (inside a C comment)! During the processing of C<char *host> line,
601 $arg will evaluate to C<ST(0)>, and C<$v{timep}> will evaluate to
604 =head2 Default Parameter Values
606 Default values for XSUB arguments can be specified by placing an
607 assignment statement in the parameter list. The default value may
608 be a number, a string or the special string C<NO_INIT>. Defaults should
609 always be used on the right-most parameters only.
611 To allow the XSUB for rpcb_gettime() to have a default host
612 value the parameters to the XSUB could be rearranged. The
613 XSUB will then call the real rpcb_gettime() function with
614 the parameters in the correct order. This XSUB can be called
615 from Perl with either of the following statements:
617 $status = rpcb_gettime( $timep, $host );
619 $status = rpcb_gettime( $timep );
621 The XSUB will look like the code which follows. A CODE:
622 block is used to call the real rpcb_gettime() function with
623 the parameters in the correct order for that function.
626 rpcb_gettime(timep,host="localhost")
628 time_t timep = NO_INIT
630 RETVAL = rpcb_gettime( host, &timep );
635 =head2 The PREINIT: Keyword
637 The PREINIT: keyword allows extra variables to be declared immediately
638 before or after the declarations of the parameters from the INPUT: section
641 If a variable is declared inside a CODE: section it will follow any typemap
642 code that is emitted for the input parameters. This may result in the
643 declaration ending up after C code, which is C syntax error. Similar
644 errors may happen with an explicit C<;>-type or C<+>-type initialization of
645 parameters is used (see L<"Initializing Function Parameters">). Declaring
646 these variables in an INIT: section will not help.
648 In such cases, to force an additional variable to be declared together
649 with declarations of other variables, place the declaration into a
650 PREINIT: section. The PREINIT: keyword may be used one or more times
653 The following examples are equivalent, but if the code is using complex
654 typemaps then the first example is safer.
658 time_t timep = NO_INIT
660 char *host = "localhost";
662 RETVAL = rpcb_gettime( host, &timep );
667 For this particular case an INIT: keyword would generate the
668 same C code as the PREINIT: keyword. Another correct, but error-prone example:
672 time_t timep = NO_INIT
674 char *host = "localhost";
675 RETVAL = rpcb_gettime( host, &timep );
680 Another way to declare C<host> is to use a C block in the CODE: section:
684 time_t timep = NO_INIT
687 char *host = "localhost";
688 RETVAL = rpcb_gettime( host, &timep );
694 The ability to put additional declarations before the typemap entries are
695 processed is very handy in the cases when typemap conversions manipulate
701 MyState st = global_state;
705 reset_to(global_state, st);
707 Here we suppose that conversion to C<MyObject> in the INPUT: section and from
708 MyObject when processing RETVAL will modify a global variable C<global_state>.
709 After these conversions are performed, we restore the old value of
710 C<global_state> (to avoid memory leaks, for example).
712 There is another way to trade clarity for compactness: INPUT sections allow
713 declaration of C variables which do not appear in the parameter list of
714 a subroutine. Thus the above code for mutate() can be rewritten as
718 MyState st = global_state;
721 reset_to(global_state, st);
723 and the code for rpcb_gettime() can be rewritten as
727 time_t timep = NO_INIT
728 char *host = "localhost";
735 =head2 The SCOPE: Keyword
737 The SCOPE: keyword allows scoping to be enabled for a particular XSUB. If
738 enabled, the XSUB will invoke ENTER and LEAVE automatically.
740 To support potentially complex type mappings, if a typemap entry used
741 by an XSUB contains a comment like C</*scope*/> then scoping will
742 be automatically enabled for that XSUB.
752 =head2 The INPUT: Keyword
754 The XSUB's parameters are usually evaluated immediately after entering the
755 XSUB. The INPUT: keyword can be used to force those parameters to be
756 evaluated a little later. The INPUT: keyword can be used multiple times
757 within an XSUB and can be used to list one or more input variables. This
758 keyword is used with the PREINIT: keyword.
760 The following example shows how the input parameter C<timep> can be
761 evaluated late, after a PREINIT.
764 rpcb_gettime(host,timep)
771 RETVAL = rpcb_gettime( host, &tt );
777 The next example shows each input parameter evaluated late.
780 rpcb_gettime(host,timep)
791 RETVAL = rpcb_gettime( h, &tt );
797 Since INPUT sections allow declaration of C variables which do not appear
798 in the parameter list of a subroutine, this may be shortened to:
801 rpcb_gettime(host,timep)
807 RETVAL = rpcb_gettime( h, &tt );
813 (We used our knowledge that input conversion for C<char *> is a "simple" one,
814 thus C<host> is initialized on the declaration line, and our assignment
815 C<h = host> is not performed too early. Otherwise one would need to have the
816 assignment C<h = host> in a CODE: or INIT: section.)
818 =head2 The IN/OUTLIST/IN_OUTLIST/OUT/IN_OUT Keywords
820 In the list of parameters for an XSUB, one can precede parameter names
821 by the C<IN>/C<OUTLIST>/C<IN_OUTLIST>/C<OUT>/C<IN_OUT> keywords.
822 C<IN> keyword is the default, the other keywords indicate how the Perl
823 interface should differ from the C interface.
825 Parameters preceded by C<OUTLIST>/C<IN_OUTLIST>/C<OUT>/C<IN_OUT>
826 keywords are considered to be used by the C subroutine I<via
827 pointers>. C<OUTLIST>/C<OUT> keywords indicate that the C subroutine
828 does not inspect the memory pointed by this parameter, but will write
829 through this pointer to provide additional return values.
831 Parameters preceded by C<OUTLIST> keyword do not appear in the usage
832 signature of the generated Perl function.
834 Parameters preceded by C<IN_OUTLIST>/C<IN_OUT>/C<OUT> I<do> appear as
835 parameters to the Perl function. With the exception of
836 C<OUT>-parameters, these parameters are converted to the corresponding
837 C type, then pointers to these data are given as arguments to the C
838 function. It is expected that the C function will write through these
841 The return list of the generated Perl function consists of the C return value
842 from the function (unless the XSUB is of C<void> return type or
843 C<The NO_OUTPUT Keyword> was used) followed by all the C<OUTLIST>
844 and C<IN_OUTLIST> parameters (in the order of appearance). On the
845 return from the XSUB the C<IN_OUT>/C<OUT> Perl parameter will be
846 modified to have the values written by the C function.
851 day_month(OUTLIST day, IN unix_time, OUTLIST month)
856 should be used from Perl as
858 my ($day, $month) = day_month(time);
860 The C signature of the corresponding function should be
862 void day_month(int *day, int unix_time, int *month);
864 The C<IN>/C<OUTLIST>/C<IN_OUTLIST>/C<IN_OUT>/C<OUT> keywords can be
865 mixed with ANSI-style declarations, as in
868 day_month(OUTLIST int day, int unix_time, OUTLIST int month)
870 (here the optional C<IN> keyword is omitted).
872 The C<IN_OUT> parameters are identical with parameters introduced with
873 L<The & Unary Operator> and put into the C<OUTPUT:> section (see
874 L<The OUTPUT: Keyword>). The C<IN_OUTLIST> parameters are very similar,
875 the only difference being that the value C function writes through the
876 pointer would not modify the Perl parameter, but is put in the output
879 The C<OUTLIST>/C<OUT> parameter differ from C<IN_OUTLIST>/C<IN_OUT>
880 parameters only by the initial value of the Perl parameter not
881 being read (and not being given to the C function - which gets some
882 garbage instead). For example, the same C function as above can be
885 void day_month(OUT int day, int unix_time, OUT int month);
890 day_month(day, unix_time, month)
898 However, the generated Perl function is called in very C-ish style:
901 day_month($day, time, $month);
903 =head2 The C<length(NAME)> Keyword
905 If one of the input arguments to the C function is the length of a string
906 argument C<NAME>, one can substitute the name of the length-argument by
907 C<length(NAME)> in the XSUB declaration. This argument must be omited when
908 the generated Perl function is called. E.g.,
911 dump_chars(char *s, short l)
915 printf("s[%d] = \"\\%#03o\"\n", n, (int)s[n]);
920 MODULE = x PACKAGE = x
922 void dump_chars(char *s, short length(s))
924 should be called as C<dump_chars($string)>.
926 This directive is supported with ANSI-type function declarations only.
928 =head2 Variable-length Parameter Lists
930 XSUBs can have variable-length parameter lists by specifying an ellipsis
931 C<(...)> in the parameter list. This use of the ellipsis is similar to that
932 found in ANSI C. The programmer is able to determine the number of
933 arguments passed to the XSUB by examining the C<items> variable which the
934 B<xsubpp> compiler supplies for all XSUBs. By using this mechanism one can
935 create an XSUB which accepts a list of parameters of unknown length.
937 The I<host> parameter for the rpcb_gettime() XSUB can be
938 optional so the ellipsis can be used to indicate that the
939 XSUB will take a variable number of parameters. Perl should
940 be able to call this XSUB with either of the following statements.
942 $status = rpcb_gettime( $timep, $host );
944 $status = rpcb_gettime( $timep );
946 The XS code, with ellipsis, follows.
949 rpcb_gettime(timep, ...)
950 time_t timep = NO_INIT
952 char *host = "localhost";
956 host = (char *)SvPV(ST(1), n_a);
957 RETVAL = rpcb_gettime( host, &timep );
962 =head2 The C_ARGS: Keyword
964 The C_ARGS: keyword allows creating of XSUBS which have different
965 calling sequence from Perl than from C, without a need to write
966 CODE: or PPCODE: section. The contents of the C_ARGS: paragraph is
967 put as the argument to the called C function without any change.
969 For example, suppose that a C function is declared as
971 symbolic nth_derivative(int n, symbolic function, int flags);
973 and that the default flags are kept in a global C variable
974 C<default_flags>. Suppose that you want to create an interface which
977 $second_deriv = $function->nth_derivative(2);
979 To do this, declare the XSUB as
982 nth_derivative(function, n)
986 n, function, default_flags
988 =head2 The PPCODE: Keyword
990 The PPCODE: keyword is an alternate form of the CODE: keyword and is used
991 to tell the B<xsubpp> compiler that the programmer is supplying the code to
992 control the argument stack for the XSUBs return values. Occasionally one
993 will want an XSUB to return a list of values rather than a single value.
994 In these cases one must use PPCODE: and then explicitly push the list of
995 values on the stack. The PPCODE: and CODE: keywords should not be used
996 together within the same XSUB.
998 The actual difference between PPCODE: and CODE: sections is in the
999 initialization of C<SP> macro (which stands for the I<current> Perl
1000 stack pointer), and in the handling of data on the stack when returning
1001 from an XSUB. In CODE: sections SP preserves the value which was on
1002 entry to the XSUB: SP is on the function pointer (which follows the
1003 last parameter). In PPCODE: sections SP is moved backward to the
1004 beginning of the parameter list, which allows C<PUSH*()> macros
1005 to place output values in the place Perl expects them to be when
1006 the XSUB returns back to Perl.
1008 The generated trailer for a CODE: section ensures that the number of return
1009 values Perl will see is either 0 or 1 (depending on the C<void>ness of the
1010 return value of the C function, and heuristics mentioned in
1011 L<"The RETVAL Variable">). The trailer generated for a PPCODE: section
1012 is based on the number of return values and on the number of times
1013 C<SP> was updated by C<[X]PUSH*()> macros.
1015 Note that macros C<ST(i)>, C<XST_m*()> and C<XSRETURN*()> work equally
1016 well in CODE: sections and PPCODE: sections.
1018 The following XSUB will call the C rpcb_gettime() function
1019 and will return its two output values, timep and status, to
1020 Perl as a single list.
1029 status = rpcb_gettime( host, &timep );
1031 PUSHs(sv_2mortal(newSViv(status)));
1032 PUSHs(sv_2mortal(newSViv(timep)));
1034 Notice that the programmer must supply the C code necessary
1035 to have the real rpcb_gettime() function called and to have
1036 the return values properly placed on the argument stack.
1038 The C<void> return type for this function tells the B<xsubpp> compiler that
1039 the RETVAL variable is not needed or used and that it should not be created.
1040 In most scenarios the void return type should be used with the PPCODE:
1043 The EXTEND() macro is used to make room on the argument
1044 stack for 2 return values. The PPCODE: directive causes the
1045 B<xsubpp> compiler to create a stack pointer available as C<SP>, and it
1046 is this pointer which is being used in the EXTEND() macro.
1047 The values are then pushed onto the stack with the PUSHs()
1050 Now the rpcb_gettime() function can be used from Perl with
1051 the following statement.
1053 ($status, $timep) = rpcb_gettime("localhost");
1055 When handling output parameters with a PPCODE section, be sure to handle
1056 'set' magic properly. See L<perlguts> for details about 'set' magic.
1058 =head2 Returning Undef And Empty Lists
1060 Occasionally the programmer will want to return simply
1061 C<undef> or an empty list if a function fails rather than a
1062 separate status value. The rpcb_gettime() function offers
1063 just this situation. If the function succeeds we would like
1064 to have it return the time and if it fails we would like to
1065 have undef returned. In the following Perl code the value
1066 of $timep will either be undef or it will be a valid time.
1068 $timep = rpcb_gettime( "localhost" );
1070 The following XSUB uses the C<SV *> return type as a mnemonic only,
1071 and uses a CODE: block to indicate to the compiler
1072 that the programmer has supplied all the necessary code. The
1073 sv_newmortal() call will initialize the return value to undef, making that
1074 the default return value.
1083 ST(0) = sv_newmortal();
1084 if( rpcb_gettime( host, &timep ) )
1085 sv_setnv( ST(0), (double)timep);
1087 The next example demonstrates how one would place an explicit undef in the
1088 return value, should the need arise.
1097 ST(0) = sv_newmortal();
1098 if( rpcb_gettime( host, &timep ) ){
1099 sv_setnv( ST(0), (double)timep);
1102 ST(0) = &PL_sv_undef;
1105 To return an empty list one must use a PPCODE: block and
1106 then not push return values on the stack.
1114 if( rpcb_gettime( host, &timep ) )
1115 PUSHs(sv_2mortal(newSViv(timep)));
1117 /* Nothing pushed on stack, so an empty
1118 * list is implicitly returned. */
1121 Some people may be inclined to include an explicit C<return> in the above
1122 XSUB, rather than letting control fall through to the end. In those
1123 situations C<XSRETURN_EMPTY> should be used, instead. This will ensure that
1124 the XSUB stack is properly adjusted. Consult L<perlapi> for other
1127 Since C<XSRETURN_*> macros can be used with CODE blocks as well, one can
1128 rewrite this example as:
1136 RETVAL = rpcb_gettime( host, &timep );
1142 In fact, one can put this check into a POSTCALL: section as well. Together
1143 with PREINIT: simplifications, this leads to:
1153 =head2 The REQUIRE: Keyword
1155 The REQUIRE: keyword is used to indicate the minimum version of the
1156 B<xsubpp> compiler needed to compile the XS module. An XS module which
1157 contains the following statement will compile with only B<xsubpp> version
1162 =head2 The CLEANUP: Keyword
1164 This keyword can be used when an XSUB requires special cleanup procedures
1165 before it terminates. When the CLEANUP: keyword is used it must follow
1166 any CODE:, PPCODE:, or OUTPUT: blocks which are present in the XSUB. The
1167 code specified for the cleanup block will be added as the last statements
1170 =head2 The POSTCALL: Keyword
1172 This keyword can be used when an XSUB requires special procedures
1173 executed after the C subroutine call is performed. When the POSTCALL:
1174 keyword is used it must precede OUTPUT: and CLEANUP: blocks which are
1175 present in the XSUB.
1177 See examples in L<"The NO_OUTPUT Keyword"> and L<"Returning Undef And Empty Lists">.
1179 The POSTCALL: block does not make a lot of sense when the C subroutine
1180 call is supplied by user by providing either CODE: or PPCODE: section.
1182 =head2 The BOOT: Keyword
1184 The BOOT: keyword is used to add code to the extension's bootstrap
1185 function. The bootstrap function is generated by the B<xsubpp> compiler and
1186 normally holds the statements necessary to register any XSUBs with Perl.
1187 With the BOOT: keyword the programmer can tell the compiler to add extra
1188 statements to the bootstrap function.
1190 This keyword may be used any time after the first MODULE keyword and should
1191 appear on a line by itself. The first blank line after the keyword will
1192 terminate the code block.
1195 # The following message will be printed when the
1196 # bootstrap function executes.
1197 printf("Hello from the bootstrap!\n");
1199 =head2 The VERSIONCHECK: Keyword
1201 The VERSIONCHECK: keyword corresponds to B<xsubpp>'s C<-versioncheck> and
1202 C<-noversioncheck> options. This keyword overrides the command line
1203 options. Version checking is enabled by default. When version checking is
1204 enabled the XS module will attempt to verify that its version matches the
1205 version of the PM module.
1207 To enable version checking:
1209 VERSIONCHECK: ENABLE
1211 To disable version checking:
1213 VERSIONCHECK: DISABLE
1215 =head2 The PROTOTYPES: Keyword
1217 The PROTOTYPES: keyword corresponds to B<xsubpp>'s C<-prototypes> and
1218 C<-noprototypes> options. This keyword overrides the command line options.
1219 Prototypes are enabled by default. When prototypes are enabled XSUBs will
1220 be given Perl prototypes. This keyword may be used multiple times in an XS
1221 module to enable and disable prototypes for different parts of the module.
1223 To enable prototypes:
1227 To disable prototypes:
1231 =head2 The PROTOTYPE: Keyword
1233 This keyword is similar to the PROTOTYPES: keyword above but can be used to
1234 force B<xsubpp> to use a specific prototype for the XSUB. This keyword
1235 overrides all other prototype options and keywords but affects only the
1236 current XSUB. Consult L<perlsub/Prototypes> for information about Perl
1240 rpcb_gettime(timep, ...)
1241 time_t timep = NO_INIT
1244 char *host = "localhost";
1248 host = (char *)SvPV(ST(1), n_a);
1249 RETVAL = rpcb_gettime( host, &timep );
1254 If the prototypes are enabled, you can disable it locally for a given
1255 XSUB as in the following example:
1258 rpcb_gettime_noproto()
1262 =head2 The ALIAS: Keyword
1264 The ALIAS: keyword allows an XSUB to have two or more unique Perl names
1265 and to know which of those names was used when it was invoked. The Perl
1266 names may be fully-qualified with package names. Each alias is given an
1267 index. The compiler will setup a variable called C<ix> which contain the
1268 index of the alias which was used. When the XSUB is called with its
1269 declared name C<ix> will be 0.
1271 The following example will create aliases C<FOO::gettime()> and
1272 C<BAR::getit()> for this function.
1275 rpcb_gettime(host,timep)
1282 printf("# ix = %d\n", ix );
1286 =head2 The OVERLOAD: Keyword
1288 Instead of writing an overloaded interface using pure Perl, you
1289 can also use the OVERLOAD keyword to define additional Perl names
1290 for your functions (like the ALIAS: keyword above). However, the
1291 overloaded functions must be defined with three parameters (except
1292 for the nomethod() function which needs four parameters). If any
1293 function has the OVERLOAD: keyword, several additional lines
1294 will be defined in the c file generated by xsubpp in order to
1295 register with the overload magic.
1297 Since blessed objects are actually stored as RV's, it is useful
1298 to use the typemap features to preprocess parameters and extract
1299 the actual SV stored within the blessed RV. See the sample for
1300 T_PTROBJ_SPECIAL below.
1302 To use the OVERLOAD: keyword, create an XS function which takes
1303 three input parameters ( or use the c style '...' definition) like
1307 cmp (lobj, robj, swap)
1312 { /* function defined here */}
1314 In this case, the function will overload both of the three way
1315 comparison operators. For all overload operations using non-alpha
1316 characters, you must type the parameter without quoting, seperating
1317 multiple overloads with whitespace. Note that "" (the stringify
1318 overload) should be entered as \"\" (i.e. escaped).
1320 =head2 The FALLBACK: Keyword
1322 In addition to the OVERLOAD keyword, if you need to control how
1323 Perl autogenerates missing overloaded operators, you can set the
1324 FALLBACK keyword in the module header section, like this:
1326 MODULE = RPC PACKAGE = RPC
1331 where FALLBACK can take any of the three values TRUE, FALSE, or
1332 UNDEF. If you do not set any FALLBACK value when using OVERLOAD,
1333 it defaults to UNDEF. FALLBACK is not used except when one or
1334 more functions using OVERLOAD have been defined. Please see
1335 L<overload/Fallback> for more details.
1337 =head2 The INTERFACE: Keyword
1339 This keyword declares the current XSUB as a keeper of the given
1340 calling signature. If some text follows this keyword, it is
1341 considered as a list of functions which have this signature, and
1342 should be attached to the current XSUB.
1344 For example, if you have 4 C functions multiply(), divide(), add(),
1345 subtract() all having the signature:
1347 symbolic f(symbolic, symbolic);
1349 you can make them all to use the same XSUB using this:
1352 interface_s_ss(arg1, arg2)
1359 (This is the complete XSUB code for 4 Perl functions!) Four generated
1360 Perl function share names with corresponding C functions.
1362 The advantage of this approach comparing to ALIAS: keyword is that there
1363 is no need to code a switch statement, each Perl function (which shares
1364 the same XSUB) knows which C function it should call. Additionally, one
1365 can attach an extra function remainder() at runtime by using
1367 CV *mycv = newXSproto("Symbolic::remainder",
1368 XS_Symbolic_interface_s_ss, __FILE__, "$$");
1369 XSINTERFACE_FUNC_SET(mycv, remainder);
1371 say, from another XSUB. (This example supposes that there was no
1372 INTERFACE_MACRO: section, otherwise one needs to use something else instead of
1373 C<XSINTERFACE_FUNC_SET>, see the next section.)
1375 =head2 The INTERFACE_MACRO: Keyword
1377 This keyword allows one to define an INTERFACE using a different way
1378 to extract a function pointer from an XSUB. The text which follows
1379 this keyword should give the name of macros which would extract/set a
1380 function pointer. The extractor macro is given return type, C<CV*>,
1381 and C<XSANY.any_dptr> for this C<CV*>. The setter macro is given cv,
1382 and the function pointer.
1384 The default value is C<XSINTERFACE_FUNC> and C<XSINTERFACE_FUNC_SET>.
1385 An INTERFACE keyword with an empty list of functions can be omitted if
1386 INTERFACE_MACRO keyword is used.
1388 Suppose that in the previous example functions pointers for
1389 multiply(), divide(), add(), subtract() are kept in a global C array
1390 C<fp[]> with offsets being C<multiply_off>, C<divide_off>, C<add_off>,
1391 C<subtract_off>. Then one can use
1393 #define XSINTERFACE_FUNC_BYOFFSET(ret,cv,f) \
1394 ((XSINTERFACE_CVT(ret,))fp[CvXSUBANY(cv).any_i32])
1395 #define XSINTERFACE_FUNC_BYOFFSET_set(cv,f) \
1396 CvXSUBANY(cv).any_i32 = CAT2( f, _off )
1401 interface_s_ss(arg1, arg2)
1405 XSINTERFACE_FUNC_BYOFFSET
1406 XSINTERFACE_FUNC_BYOFFSET_set
1413 =head2 The INCLUDE: Keyword
1415 This keyword can be used to pull other files into the XS module. The other
1416 files may have XS code. INCLUDE: can also be used to run a command to
1417 generate the XS code to be pulled into the module.
1419 The file F<Rpcb1.xsh> contains our C<rpcb_gettime()> function:
1422 rpcb_gettime(host,timep)
1428 The XS module can use INCLUDE: to pull that file into it.
1432 If the parameters to the INCLUDE: keyword are followed by a pipe (C<|>) then
1433 the compiler will interpret the parameters as a command.
1435 INCLUDE: cat Rpcb1.xsh |
1437 =head2 The CASE: Keyword
1439 The CASE: keyword allows an XSUB to have multiple distinct parts with each
1440 part acting as a virtual XSUB. CASE: is greedy and if it is used then all
1441 other XS keywords must be contained within a CASE:. This means nothing may
1442 precede the first CASE: in the XSUB and anything following the last CASE: is
1443 included in that case.
1445 A CASE: might switch via a parameter of the XSUB, via the C<ix> ALIAS:
1446 variable (see L<"The ALIAS: Keyword">), or maybe via the C<items> variable
1447 (see L<"Variable-length Parameter Lists">). The last CASE: becomes the
1448 B<default> case if it is not associated with a conditional. The following
1449 example shows CASE switched via C<ix> with a function C<rpcb_gettime()>
1450 having an alias C<x_gettime()>. When the function is called as
1451 C<rpcb_gettime()> its parameters are the usual C<(char *host, time_t *timep)>,
1452 but when the function is called as C<x_gettime()> its parameters are
1453 reversed, C<(time_t *timep, char *host)>.
1461 # 'a' is timep, 'b' is host
1465 RETVAL = rpcb_gettime( b, &a );
1470 # 'a' is host, 'b' is timep
1477 That function can be called with either of the following statements. Note
1478 the different argument lists.
1480 $status = rpcb_gettime( $host, $timep );
1482 $status = x_gettime( $timep, $host );
1484 =head2 The & Unary Operator
1486 The C<&> unary operator in the INPUT: section is used to tell B<xsubpp>
1487 that it should convert a Perl value to/from C using the C type to the left
1488 of C<&>, but provide a pointer to this value when the C function is called.
1490 This is useful to avoid a CODE: block for a C function which takes a parameter
1491 by reference. Typically, the parameter should be not a pointer type (an
1492 C<int> or C<long> but not an C<int*> or C<long*>).
1494 The following XSUB will generate incorrect C code. The B<xsubpp> compiler will
1495 turn this into code which calls C<rpcb_gettime()> with parameters C<(char
1496 *host, time_t timep)>, but the real C<rpcb_gettime()> wants the C<timep>
1497 parameter to be of type C<time_t*> rather than C<time_t>.
1500 rpcb_gettime(host,timep)
1506 That problem is corrected by using the C<&> operator. The B<xsubpp> compiler
1507 will now turn this into code which calls C<rpcb_gettime()> correctly with
1508 parameters C<(char *host, time_t *timep)>. It does this by carrying the
1509 C<&> through, so the function call looks like C<rpcb_gettime(host, &timep)>.
1512 rpcb_gettime(host,timep)
1518 =head2 Inserting POD, Comments and C Preprocessor Directives
1520 C preprocessor directives are allowed within BOOT:, PREINIT: INIT:, CODE:,
1521 PPCODE:, POSTCALL:, and CLEANUP: blocks, as well as outside the functions.
1522 Comments are allowed anywhere after the MODULE keyword. The compiler will
1523 pass the preprocessor directives through untouched and will remove the
1524 commented lines. POD documentation is allowed at any point, both in the
1525 C and XS language sections. POD must be terminated with a C<=cut> command;
1526 C<xsubpp> will exit with an error if it does not. It is very unlikely that
1527 human generated C code will be mistaken for POD, as most indenting styles
1528 result in whitespace in front of any line starting with C<=>. Machine
1529 generated XS files may fall into this trap unless care is taken to
1530 ensure that a space breaks the sequence "\n=".
1532 Comments can be added to XSUBs by placing a C<#> as the first
1533 non-whitespace of a line. Care should be taken to avoid making the
1534 comment look like a C preprocessor directive, lest it be interpreted as
1535 such. The simplest way to prevent this is to put whitespace in front of
1538 If you use preprocessor directives to choose one of two
1539 versions of a function, use
1542 #else /* ... version2 */
1552 because otherwise B<xsubpp> will believe that you made a duplicate
1553 definition of the function. Also, put a blank line before the
1554 #else/#endif so it will not be seen as part of the function body.
1556 =head2 Using XS With C++
1558 If an XSUB name contains C<::>, it is considered to be a C++ method.
1559 The generated Perl function will assume that
1560 its first argument is an object pointer. The object pointer
1561 will be stored in a variable called THIS. The object should
1562 have been created by C++ with the new() function and should
1563 be blessed by Perl with the sv_setref_pv() macro. The
1564 blessing of the object by Perl can be handled by a typemap. An example
1565 typemap is shown at the end of this section.
1567 If the return type of the XSUB includes C<static>, the method is considered
1568 to be a static method. It will call the C++
1569 function using the class::method() syntax. If the method is not static
1570 the function will be called using the THIS-E<gt>method() syntax.
1572 The next examples will use the following C++ class.
1579 void set_blue( int );
1585 The XSUBs for the blue() and set_blue() methods are defined with the class
1586 name but the parameter for the object (THIS, or "self") is implicit and is
1593 color::set_blue( val )
1596 Both Perl functions will expect an object as the first parameter. In the
1597 generated C++ code the object is called C<THIS>, and the method call will
1598 be performed on this object. So in the C++ code the blue() and set_blue()
1599 methods will be called as this:
1601 RETVAL = THIS->blue();
1603 THIS->set_blue( val );
1605 You could also write a single get/set method using an optional argument:
1608 color::blue( val = NO_INIT )
1613 THIS->set_blue( val );
1614 RETVAL = THIS->blue();
1618 If the function's name is B<DESTROY> then the C++ C<delete> function will be
1619 called and C<THIS> will be given as its parameter. The generated C++ code for
1624 will look like this:
1626 color *THIS = ...; // Initialized as in typemap
1630 If the function's name is B<new> then the C++ C<new> function will be called
1631 to create a dynamic C++ object. The XSUB will expect the class name, which
1632 will be kept in a variable called C<CLASS>, to be given as the first
1638 The generated C++ code will call C<new>.
1640 RETVAL = new color();
1642 The following is an example of a typemap that could be used for this C++
1649 # The Perl object is blessed into 'CLASS', which should be a
1650 # char* having the name of the package for the blessing.
1652 sv_setref_pv( $arg, CLASS, (void*)$var );
1656 if( sv_isobject($arg) && (SvTYPE(SvRV($arg)) == SVt_PVMG) )
1657 $var = ($type)SvIV((SV*)SvRV( $arg ));
1659 warn( \"${Package}::$func_name() -- $var is not a blessed SV reference\" );
1663 =head2 Interface Strategy
1665 When designing an interface between Perl and a C library a straight
1666 translation from C to XS (such as created by C<h2xs -x>) is often sufficient.
1667 However, sometimes the interface will look
1668 very C-like and occasionally nonintuitive, especially when the C function
1669 modifies one of its parameters, or returns failure inband (as in "negative
1670 return values mean failure"). In cases where the programmer wishes to
1671 create a more Perl-like interface the following strategy may help to
1672 identify the more critical parts of the interface.
1674 Identify the C functions with input/output or output parameters. The XSUBs for
1675 these functions may be able to return lists to Perl.
1677 Identify the C functions which use some inband info as an indication
1678 of failure. They may be
1679 candidates to return undef or an empty list in case of failure. If the
1680 failure may be detected without a call to the C function, you may want to use
1681 an INIT: section to report the failure. For failures detectable after the C
1682 function returns one may want to use a POSTCALL: section to process the
1683 failure. In more complicated cases use CODE: or PPCODE: sections.
1685 If many functions use the same failure indication based on the return value,
1686 you may want to create a special typedef to handle this situation. Put
1688 typedef int negative_is_failure;
1690 near the beginning of XS file, and create an OUTPUT typemap entry
1691 for C<negative_is_failure> which converts negative values to C<undef>, or
1692 maybe croak()s. After this the return value of type C<negative_is_failure>
1693 will create more Perl-like interface.
1695 Identify which values are used by only the C and XSUB functions
1696 themselves, say, when a parameter to a function should be a contents of a
1697 global variable. If Perl does not need to access the contents of the value
1698 then it may not be necessary to provide a translation for that value
1701 Identify the pointers in the C function parameter lists and return
1702 values. Some pointers may be used to implement input/output or
1703 output parameters, they can be handled in XS with the C<&> unary operator,
1704 and, possibly, using the NO_INIT keyword.
1705 Some others will require handling of types like C<int *>, and one needs
1706 to decide what a useful Perl translation will do in such a case. When
1707 the semantic is clear, it is advisable to put the translation into a typemap
1710 Identify the structures used by the C functions. In many
1711 cases it may be helpful to use the T_PTROBJ typemap for
1712 these structures so they can be manipulated by Perl as
1713 blessed objects. (This is handled automatically by C<h2xs -x>.)
1715 If the same C type is used in several different contexts which require
1716 different translations, C<typedef> several new types mapped to this C type,
1717 and create separate F<typemap> entries for these new types. Use these
1718 types in declarations of return type and parameters to XSUBs.
1720 =head2 Perl Objects And C Structures
1722 When dealing with C structures one should select either
1723 B<T_PTROBJ> or B<T_PTRREF> for the XS type. Both types are
1724 designed to handle pointers to complex objects. The
1725 T_PTRREF type will allow the Perl object to be unblessed
1726 while the T_PTROBJ type requires that the object be blessed.
1727 By using T_PTROBJ one can achieve a form of type-checking
1728 because the XSUB will attempt to verify that the Perl object
1729 is of the expected type.
1731 The following XS code shows the getnetconfigent() function which is used
1732 with ONC+ TIRPC. The getnetconfigent() function will return a pointer to a
1733 C structure and has the C prototype shown below. The example will
1734 demonstrate how the C pointer will become a Perl reference. Perl will
1735 consider this reference to be a pointer to a blessed object and will
1736 attempt to call a destructor for the object. A destructor will be
1737 provided in the XS source to free the memory used by getnetconfigent().
1738 Destructors in XS can be created by specifying an XSUB function whose name
1739 ends with the word B<DESTROY>. XS destructors can be used to free memory
1740 which may have been malloc'd by another XSUB.
1742 struct netconfig *getnetconfigent(const char *netid);
1744 A C<typedef> will be created for C<struct netconfig>. The Perl
1745 object will be blessed in a class matching the name of the C
1746 type, with the tag C<Ptr> appended, and the name should not
1747 have embedded spaces if it will be a Perl package name. The
1748 destructor will be placed in a class corresponding to the
1749 class of the object and the PREFIX keyword will be used to
1750 trim the name to the word DESTROY as Perl will expect.
1752 typedef struct netconfig Netconfig;
1754 MODULE = RPC PACKAGE = RPC
1757 getnetconfigent(netid)
1760 MODULE = RPC PACKAGE = NetconfigPtr PREFIX = rpcb_
1763 rpcb_DESTROY(netconf)
1766 printf("Now in NetconfigPtr::DESTROY\n");
1769 This example requires the following typemap entry. Consult the typemap
1770 section for more information about adding new typemaps for an extension.
1773 Netconfig * T_PTROBJ
1775 This example will be used with the following Perl statements.
1778 $netconf = getnetconfigent("udp");
1780 When Perl destroys the object referenced by $netconf it will send the
1781 object to the supplied XSUB DESTROY function. Perl cannot determine, and
1782 does not care, that this object is a C struct and not a Perl object. In
1783 this sense, there is no difference between the object created by the
1784 getnetconfigent() XSUB and an object created by a normal Perl subroutine.
1788 The typemap is a collection of code fragments which are used by the B<xsubpp>
1789 compiler to map C function parameters and values to Perl values. The
1790 typemap file may consist of three sections labelled C<TYPEMAP>, C<INPUT>, and
1791 C<OUTPUT>. An unlabelled initial section is assumed to be a C<TYPEMAP>
1792 section. The INPUT section tells
1793 the compiler how to translate Perl values
1794 into variables of certain C types. The OUTPUT section tells the compiler
1795 how to translate the values from certain C types into values Perl can
1796 understand. The TYPEMAP section tells the compiler which of the INPUT and
1797 OUTPUT code fragments should be used to map a given C type to a Perl value.
1798 The section labels C<TYPEMAP>, C<INPUT>, or C<OUTPUT> must begin
1799 in the first column on a line by themselves, and must be in uppercase.
1801 The default typemap in the C<lib/ExtUtils> directory of the Perl source
1802 contains many useful types which can be used by Perl extensions. Some
1803 extensions define additional typemaps which they keep in their own directory.
1804 These additional typemaps may reference INPUT and OUTPUT maps in the main
1805 typemap. The B<xsubpp> compiler will allow the extension's own typemap to
1806 override any mappings which are in the default typemap.
1808 Most extensions which require a custom typemap will need only the TYPEMAP
1809 section of the typemap file. The custom typemap used in the
1810 getnetconfigent() example shown earlier demonstrates what may be the typical
1811 use of extension typemaps. That typemap is used to equate a C structure
1812 with the T_PTROBJ typemap. The typemap used by getnetconfigent() is shown
1813 here. Note that the C type is separated from the XS type with a tab and
1814 that the C unary operator C<*> is considered to be a part of the C type name.
1817 Netconfig *<tab>T_PTROBJ
1819 Here's a more complicated example: suppose that you wanted C<struct
1820 netconfig> to be blessed into the class C<Net::Config>. One way to do
1821 this is to use underscores (_) to separate package names, as follows:
1823 typedef struct netconfig * Net_Config;
1825 And then provide a typemap entry C<T_PTROBJ_SPECIAL> that maps underscores to
1826 double-colons (::), and declare C<Net_Config> to be of that type:
1830 Net_Config T_PTROBJ_SPECIAL
1834 if (sv_derived_from($arg, \"${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\")) {
1835 IV tmp = SvIV((SV*)SvRV($arg));
1836 $var = INT2PTR($type, tmp);
1839 croak(\"$var is not of type ${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\")
1843 sv_setref_pv($arg, \"${(my $ntt=$ntype)=~s/_/::/g;\$ntt}\",
1846 The INPUT and OUTPUT sections substitute underscores for double-colons
1847 on the fly, giving the desired effect. This example demonstrates some
1848 of the power and versatility of the typemap facility.
1850 The INT2PTR macro (defined in perl.h) casts an integer to a pointer,
1851 of a given type, taking care of the possible different size of integers
1852 and pointers. There are also PTR2IV, PTR2UV, PTR2NV macros,
1853 to map the other way, which may be useful in OUTPUT sections.
1855 =head2 Safely Storing Static Data in XS
1857 Starting with Perl 5.8, a macro framework has been defined to allow
1858 static data to be safely stored in XS modules that will be accessed from
1859 a multi-threaded Perl.
1861 Although primarily designed for use with multi-threaded Perl, the macros
1862 have been designed so that they will work with non-threaded Perl as well.
1864 It is therefore strongly recommended that these macros be used by all
1865 XS modules that make use of static data.
1867 The easiest way to get a template set of macros to use is by specifying
1868 the C<-g> (C<--global>) option with h2xs (see L<h2xs>).
1870 Below is an example module that makes use of the macros.
1878 #define MY_CXT_KEY "BlindMice::_guts" XS_VERSION
1887 MODULE = BlindMice PACKAGE = BlindMice
1893 strcpy(MY_CXT.name[0], "None");
1894 strcpy(MY_CXT.name[1], "None");
1895 strcpy(MY_CXT.name[2], "None");
1899 newMouse(char * name)
1904 if (MY_CXT.count >= 3) {
1905 warn("Already have 3 blind mice") ;
1909 RETVAL = ++ MY_CXT.count;
1910 strcpy(MY_CXT.name[MY_CXT.count - 1], name);
1914 get_mouse_name(index)
1918 RETVAL = MY_CXT.lives ++;
1919 if (index > MY_CXT.count)
1920 croak("There are only 3 blind mice.");
1922 RETVAL = newSVpv(MY_CXT.name[index - 1]);
1931 This macro is used to define a unique key to refer to the static data
1932 for an XS module. The suggested naming scheme, as used by h2xs, is to
1933 use a string that consists of the module name, the string "::_guts"
1934 and the module version number.
1936 #define MY_CXT_KEY "MyModule::_guts" XS_VERSION
1938 =item typedef my_cxt_t
1940 This struct typedef I<must> always be called C<my_cxt_t> -- the other
1941 C<CXT*> macros assume the existence of the C<my_cxt_t> typedef name.
1943 Declare a typedef named C<my_cxt_t> that is a structure that contains
1944 all the data that needs to be interpreter-local.
1952 Always place the START_MY_CXT macro directly after the declaration
1957 The MY_CXT_INIT macro initialises storage for the C<my_cxt_t> struct.
1959 It I<must> be called exactly once -- typically in a BOOT: section.
1963 Use the dMY_CXT macro (a declaration) in all the functions that access
1968 Use the MY_CXT macro to access members of the C<my_cxt_t> struct. For
1969 example, if C<my_cxt_t> is
1975 then use this to access the C<index> member
1984 File C<RPC.xs>: Interface to some ONC+ RPC bind library functions.
1990 #include <rpc/rpc.h>
1992 typedef struct netconfig Netconfig;
1994 MODULE = RPC PACKAGE = RPC
1997 rpcb_gettime(host="localhost")
2002 ST(0) = sv_newmortal();
2003 if( rpcb_gettime( host, &timep ) )
2004 sv_setnv( ST(0), (double)timep );
2007 getnetconfigent(netid="udp")
2010 MODULE = RPC PACKAGE = NetconfigPtr PREFIX = rpcb_
2013 rpcb_DESTROY(netconf)
2016 printf("NetconfigPtr::DESTROY\n");
2019 File C<typemap>: Custom typemap for RPC.xs.
2022 Netconfig * T_PTROBJ
2024 File C<RPC.pm>: Perl module for the RPC extension.
2030 @ISA = qw(Exporter DynaLoader);
2031 @EXPORT = qw(rpcb_gettime getnetconfigent);
2036 File C<rpctest.pl>: Perl test program for the RPC extension.
2040 $netconf = getnetconfigent();
2041 $a = rpcb_gettime();
2042 print "time = $a\n";
2043 print "netconf = $netconf\n";
2045 $netconf = getnetconfigent("tcp");
2046 $a = rpcb_gettime("poplar");
2047 print "time = $a\n";
2048 print "netconf = $netconf\n";
2053 This document covers features supported by C<xsubpp> 1.935.
2057 Originally written by Dean Roehrich <F<roehrich@cray.com>>.
2059 Maintained since 1996 by The Perl Porters <F<perlbug@perl.org>>.