This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Describe how Configure patches should be done
[perl5.git] / pod / perlhack.pod
CommitLineData
e8cd7eae
GS
1=head1 NAME
2
3perlhack - How to hack at the Perl internals
4
5=head1 DESCRIPTION
6
7This document attempts to explain how Perl development takes place,
8and ends with some suggestions for people wanting to become bona fide
9porters.
10
11The perl5-porters mailing list is where the Perl standard distribution
12is maintained and developed. The list can get anywhere from 10 to 150
13messages a day, depending on the heatedness of the debate. Most days
14there are two or three patches, extensions, features, or bugs being
15discussed at a time.
16
f8e3975a 17A searchable archive of the list is at either:
e8cd7eae
GS
18
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
20
f8e3975a
IP
21or
22
23 http://archive.develooper.com/perl5-porters@perl.org/
24
e8cd7eae
GS
25List subscribers (the porters themselves) come in several flavours.
26Some are quiet curious lurkers, who rarely pitch in and instead watch
27the ongoing development to ensure they're forewarned of new changes or
28features in Perl. Some are representatives of vendors, who are there
29to make sure that Perl continues to compile and work on their
30platforms. Some patch any reported bug that they know how to fix,
31some are actively patching their pet area (threads, Win32, the regexp
32engine), while others seem to do nothing but complain. In other
33words, it's your usual mix of technical people.
34
35Over this group of porters presides Larry Wall. He has the final word
f6c51b38 36in what does and does not change in the Perl language. Various
b432a672
AL
37releases of Perl are shepherded by a "pumpking", a porter
38responsible for gathering patches, deciding on a patch-by-patch,
f6c51b38 39feature-by-feature basis what will and will not go into the release.
caf100c0 40For instance, Gurusamy Sarathy was the pumpking for the 5.6 release of
961f29c6 41Perl, and Jarkko Hietaniemi was the pumpking for the 5.8 release, and
1a88dbf8 42Rafael Garcia-Suarez holds the pumpking crown for the 5.10 release.
e8cd7eae
GS
43
44In addition, various people are pumpkings for different things. For
961f29c6
MB
45instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the
46I<Configure> pumpkin up till the 5.8 release. For the 5.10 release
47H.Merijn Brand took over.
e8cd7eae
GS
48
49Larry sees Perl development along the lines of the US government:
50there's the Legislature (the porters), the Executive branch (the
51pumpkings), and the Supreme Court (Larry). The legislature can
52discuss and submit patches to the executive branch all they like, but
53the executive branch is free to veto them. Rarely, the Supreme Court
54will side with the executive branch over the legislature, or the
55legislature over the executive branch. Mostly, however, the
56legislature and the executive branch are supposed to get along and
57work out their differences without impeachment or court cases.
58
59You might sometimes see reference to Rule 1 and Rule 2. Larry's power
60as Supreme Court is expressed in The Rules:
61
62=over 4
63
64=item 1
65
66Larry is always by definition right about how Perl should behave.
67This means he has final veto power on the core functionality.
68
69=item 2
70
71Larry is allowed to change his mind about any matter at a later date,
72regardless of whether he previously invoked Rule 1.
73
74=back
75
76Got that? Larry is always right, even when he was wrong. It's rare
77to see either Rule exercised, but they are often alluded to.
78
79New features and extensions to the language are contentious, because
80the criteria used by the pumpkings, Larry, and other porters to decide
81which features should be implemented and incorporated are not codified
82in a few small design goals as with some other languages. Instead,
83the heuristics are flexible and often difficult to fathom. Here is
84one person's list, roughly in decreasing order of importance, of
85heuristics that new features have to be weighed against:
86
87=over 4
88
89=item Does concept match the general goals of Perl?
90
91These haven't been written anywhere in stone, but one approximation
92is:
93
94 1. Keep it fast, simple, and useful.
95 2. Keep features/concepts as orthogonal as possible.
96 3. No arbitrary limits (platforms, data sizes, cultures).
97 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
98 5. Either assimilate new technologies, or build bridges to them.
99
100=item Where is the implementation?
101
102All the talk in the world is useless without an implementation. In
103almost every case, the person or people who argue for a new feature
104will be expected to be the ones who implement it. Porters capable
105of coding new features have their own agendas, and are not available
106to implement your (possibly good) idea.
107
108=item Backwards compatibility
109
110It's a cardinal sin to break existing Perl programs. New warnings are
111contentious--some say that a program that emits warnings is not
112broken, while others say it is. Adding keywords has the potential to
113break programs, changing the meaning of existing token sequences or
114functions might break programs.
115
116=item Could it be a module instead?
117
118Perl 5 has extension mechanisms, modules and XS, specifically to avoid
119the need to keep changing the Perl interpreter. You can write modules
120that export functions, you can give those functions prototypes so they
121can be called like built-in functions, you can even write XS code to
122mess with the runtime data structures of the Perl interpreter if you
123want to implement really complicated things. If it can be done in a
124module instead of in the core, it's highly unlikely to be added.
125
126=item Is the feature generic enough?
127
128Is this something that only the submitter wants added to the language,
129or would it be broadly useful? Sometimes, instead of adding a feature
130with a tight focus, the porters might decide to wait until someone
131implements the more generalized feature. For instance, instead of
b432a672 132implementing a "delayed evaluation" feature, the porters are waiting
e8cd7eae
GS
133for a macro system that would permit delayed evaluation and much more.
134
135=item Does it potentially introduce new bugs?
136
137Radical rewrites of large chunks of the Perl interpreter have the
138potential to introduce new bugs. The smaller and more localized the
139change, the better.
140
141=item Does it preclude other desirable features?
142
143A patch is likely to be rejected if it closes off future avenues of
144development. For instance, a patch that placed a true and final
145interpretation on prototypes is likely to be rejected because there
146are still options for the future of prototypes that haven't been
147addressed.
148
149=item Is the implementation robust?
150
151Good patches (tight code, complete, correct) stand more chance of
152going in. Sloppy or incorrect patches might be placed on the back
153burner until the pumpking has time to fix, or might be discarded
154altogether without further notice.
155
156=item Is the implementation generic enough to be portable?
157
158The worst patches make use of a system-specific features. It's highly
353c6505 159unlikely that non-portable additions to the Perl language will be
e8cd7eae
GS
160accepted.
161
a936dd3c
NC
162=item Is the implementation tested?
163
164Patches which change behaviour (fixing bugs or introducing new features)
165must include regression tests to verify that everything works as expected.
166Without tests provided by the original author, how can anyone else changing
167perl in the future be sure that they haven't unwittingly broken the behaviour
168the patch implements? And without tests, how can the patch's author be
9d077eaa 169confident that his/her hard work put into the patch won't be accidentally
a936dd3c
NC
170thrown away by someone in the future?
171
e8cd7eae
GS
172=item Is there enough documentation?
173
174Patches without documentation are probably ill-thought out or
175incomplete. Nothing can be added without documentation, so submitting
176a patch for the appropriate manpages as well as the source code is
a936dd3c 177always a good idea.
e8cd7eae
GS
178
179=item Is there another way to do it?
180
b432a672
AL
181Larry said "Although the Perl Slogan is I<There's More Than One Way
182to Do It>, I hesitate to make 10 ways to do something". This is a
e8cd7eae
GS
183tricky heuristic to navigate, though--one man's essential addition is
184another man's pointless cruft.
185
186=item Does it create too much work?
187
188Work for the pumpking, work for Perl programmers, work for module
189authors, ... Perl is supposed to be easy.
190
f6c51b38
GS
191=item Patches speak louder than words
192
193Working code is always preferred to pie-in-the-sky ideas. A patch to
194add a feature stands a much higher chance of making it to the language
195than does a random feature request, no matter how fervently argued the
b432a672 196request might be. This ties into "Will it be useful?", as the fact
f6c51b38
GS
197that someone took the time to make the patch demonstrates a strong
198desire for the feature.
199
e8cd7eae
GS
200=back
201
b432a672
AL
202If you're on the list, you might hear the word "core" bandied
203around. It refers to the standard distribution. "Hacking on the
204core" means you're changing the C source code to the Perl
205interpreter. "A core module" is one that ships with Perl.
e8cd7eae 206
a1f349fd
MB
207=head2 Keeping in sync
208
e8cd7eae 209The source code to the Perl interpreter, in its different versions, is
b16c2e4a
RGS
210kept in a repository managed by the git revision control system. The
211pumpkings and a few others have write access to the repository to check in
212changes.
2be4c08b 213
b16c2e4a 214How to clone and use the git perl repository is described in L<perlrepository>.
2be4c08b 215
b16c2e4a 216You can also choose to use rsync to get a copy of the current source tree
fe749c9a 217for the bleadperl branch and all maintenance branches :
0cfb3454 218
b16c2e4a
RGS
219 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-current .
220 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.10.x .
221 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.8.x .
222 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.6.x .
223 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.005xx .
224
225(Add the C<--delete> option to remove leftover files)
0cfb3454
GS
226
227You may also want to subscribe to the perl5-changes mailing list to
228receive a copy of each patch that gets submitted to the maintenance
229and development "branches" of the perl repository. See
230http://lists.perl.org/ for subscription information.
231
a1f349fd
MB
232If you are a member of the perl5-porters mailing list, it is a good
233thing to keep in touch with the most recent changes. If not only to
234verify if what you would have posted as a bug report isn't already
235solved in the most recent available perl development branch, also
236known as perl-current, bleading edge perl, bleedperl or bleadperl.
2be4c08b
GS
237
238Needless to say, the source code in perl-current is usually in a perpetual
239state of evolution. You should expect it to be very buggy. Do B<not> use
240it for any purpose other than testing and development.
e8cd7eae 241
3fd28c4e 242=head2 Perlbug administration
52315700 243
902821cc
RGS
244There is a single remote administrative interface for modifying bug status,
245category, open issues etc. using the B<RT> bugtracker system, maintained
246by Robert Spier. Become an administrator, and close any bugs you can get
3fd28c4e 247your sticky mitts on:
52315700 248
39417508 249 http://bugs.perl.org/
52315700 250
3fd28c4e 251To email the bug system administrators:
52315700 252
3fd28c4e 253 "perlbug-admin" <perlbug-admin@perl.org>
52315700 254
a1f349fd
MB
255=head2 Submitting patches
256
f7e1e956
MS
257Always submit patches to I<perl5-porters@perl.org>. If you're
258patching a core module and there's an author listed, send the author a
259copy (see L<Patching a core module>). This lets other porters review
260your patch, which catches a surprising number of errors in patches.
b16c2e4a
RGS
261Please patch against the latest B<development> version. (e.g., even if
262you're fixing a bug in the 5.8 track, patch against the C<blead> branch in
263the git repository.)
824d470b
RGS
264
265If changes are accepted, they are applied to the development branch. Then
fe749c9a 266the maintenance pumpking decides which of those patches is to be
b16c2e4a
RGS
267backported to the maint branch. Only patches that survive the heat of the
268development branch get applied to maintenance versions.
f7e1e956 269
2c8694a7
JH
270Your patch should update the documentation and test suite. See
271L<Writing a test>. If you have added or removed files in the distribution,
272edit the MANIFEST file accordingly, sort the MANIFEST file using
273C<make manisort>, and include those changes as part of your patch.
e8cd7eae 274
824d470b
RGS
275Patching documentation also follows the same order: if accepted, a patch
276is first applied to B<development>, and if relevant then it's backported
277to B<maintenance>. (With an exception for some patches that document
278behaviour that only appears in the maintenance branch, but which has
279changed in the development version.)
280
e8cd7eae
GS
281To report a bug in Perl, use the program I<perlbug> which comes with
282Perl (if you can't get Perl to work, send mail to the address
f18956b7 283I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
e8cd7eae 284I<perlbug> feeds into the automated bug-tracking system, access to
902821cc 285which is provided through the web at http://rt.perl.org/rt3/ . It
e8cd7eae
GS
286often pays to check the archives of the perl5-porters mailing list to
287see whether the bug you're reporting has been reported before, and if
288so whether it was considered a bug. See above for the location of
289the searchable archives.
290
f224927c 291The CPAN testers ( http://testers.cpan.org/ ) are a group of
ba139f7d 292volunteers who test CPAN modules on a variety of platforms. Perl
d3e8af89
GS
293Smokers ( http://www.nntp.perl.org/group/perl.daily-build and
294http://www.nntp.perl.org/group/perl.daily-build.reports/ )
902821cc 295automatically test Perl source releases on platforms with various
d3e8af89
GS
296configurations. Both efforts welcome volunteers. In order to get
297involved in smoke testing of the perl itself visit
298L<http://search.cpan.org/dist/Test-Smoke>. In order to start smoke
68cbce50
CBW
299testing CPAN modules visit L<http://search.cpan.org/dist/CPANPLUS-YACSmoke/>
300or L<http://search.cpan.org/dist/minismokebox/> or
d3e8af89 301L<http://search.cpan.org/dist/CPAN-Reporter/>.
e8cd7eae 302
e8cd7eae
GS
303It's a good idea to read and lurk for a while before chipping in.
304That way you'll get to see the dynamic of the conversations, learn the
305personalities of the players, and hopefully be better prepared to make
306a useful contribution when do you speak up.
307
308If after all this you still think you want to join the perl5-porters
f6c51b38
GS
309mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
310unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
e8cd7eae 311
a422fd2d
SC
312To hack on the Perl guts, you'll need to read the following things:
313
314=over 3
315
316=item L<perlguts>
317
318This is of paramount importance, since it's the documentation of what
319goes where in the Perl source. Read it over a couple of times and it
320might start to make sense - don't worry if it doesn't yet, because the
321best way to study it is to read it in conjunction with poking at Perl
322source, and we'll do that later on.
323
2e5b5004
RGS
324Gisle Aas's illustrated perlguts (also known as I<illguts>) is wonderful,
325although a little out of date with regard to some size details; the
326various SV structures have since been reworked for smaller memory footprint.
327The fundamentals are right however, and the pictures are very helpful.
de10be12 328
2e5b5004 329L<http://www.perl.org/tpc/1998/Perl_Language_and_Modules/Perl%20Illustrated/>
a422fd2d
SC
330
331=item L<perlxstut> and L<perlxs>
332
333A working knowledge of XSUB programming is incredibly useful for core
334hacking; XSUBs use techniques drawn from the PP code, the portion of the
335guts that actually executes a Perl program. It's a lot gentler to learn
336those techniques from simple examples and explanation than from the core
337itself.
338
339=item L<perlapi>
340
341The documentation for the Perl API explains what some of the internal
342functions do, as well as the many macros used in the source.
343
344=item F<Porting/pumpkin.pod>
345
346This is a collection of words of wisdom for a Perl porter; some of it is
347only useful to the pumpkin holder, but most of it applies to anyone
348wanting to go about Perl development.
349
350=item The perl5-porters FAQ
351
902821cc
RGS
352This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html .
353It contains hints on reading perl5-porters, information on how
354perl5-porters works and how Perl development in general works.
a422fd2d
SC
355
356=back
357
358=head2 Finding Your Way Around
359
360Perl maintenance can be split into a number of areas, and certain people
361(pumpkins) will have responsibility for each area. These areas sometimes
362correspond to files or directories in the source kit. Among the areas are:
363
364=over 3
365
366=item Core modules
367
c53fdc5e
RF
368Modules shipped as part of the Perl core live in various subdirectories, where
369two are dedicated to core-only modules, and two are for the dual-life modules
370which live on CPAN and may be maintained separately with respect to the Perl
371core:
372
373 lib/ is for pure-Perl modules, which exist in the core only.
374
375 ext/ is for XS extensions, and modules with special Makefile.PL requirements, which exist in the core only.
376
377 cpan/ is for dual-life modules, where the CPAN module is canonical (should be patched first).
378
379 dist/ is for dual-life modules, where the blead source is canonical.
a422fd2d 380
f7e1e956
MS
381=item Tests
382
383There are tests for nearly all the modules, built-ins and major bits
384of functionality. Test files all have a .t suffix. Module tests live
385in the F<lib/> and F<ext/> directories next to the module being
386tested. Others live in F<t/>. See L<Writing a test>
387
a422fd2d
SC
388=item Documentation
389
390Documentation maintenance includes looking after everything in the
391F<pod/> directory, (as well as contributing new documentation) and
392the documentation to the modules in core.
393
394=item Configure
395
99c47ece 396The Configure process is the way we make Perl portable across the
a422fd2d 397myriad of operating systems it supports. Responsibility for the
99c47ece
MB
398Configure, build and installation process, as well as the overall
399portability of the core code rests with the Configure pumpkin -
400others help out with individual operating systems.
401
402The three files that fall under his/her resposibility are Configure,
403config_h.SH, and Porting/Glossary (and a whole bunch of small related
404files that are less important here). The Configure pumpkin decides how
405patches to these are dealt with. Currently, the Configure pumpkin will
406accept patches in most common formats, even directly to these files.
407Other committers are allowed to commit to these files under the strict
408condition that they will inform the Configure pumpkin, either on IRC
409(if he/she happens to be around) or through (personal) e-mail.
a422fd2d
SC
410
411The files involved are the operating system directories, (F<win32/>,
412F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
413and F<Makefile>, as well as the metaconfig files which generate
414F<Configure>. (metaconfig isn't included in the core distribution.)
415
99c47ece
MB
416See http://perl5.git.perl.org/metaconfig.git/blob/HEAD:/README for a
417description of the full process involved.
418
a422fd2d
SC
419=item Interpreter
420
421And of course, there's the core of the Perl interpreter itself. Let's
422have a look at that in a little more detail.
423
424=back
425
426Before we leave looking at the layout, though, don't forget that
427F<MANIFEST> contains not only the file names in the Perl distribution,
428but short descriptions of what's in them, too. For an overview of the
429important files, try this:
430
431 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
432
433=head2 Elements of the interpreter
434
435The work of the interpreter has two main stages: compiling the code
436into the internal representation, or bytecode, and then executing it.
437L<perlguts/Compiled code> explains exactly how the compilation stage
438happens.
439
440Here is a short breakdown of perl's operation:
441
442=over 3
443
444=item Startup
445
446The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
447This is very high-level code, enough to fit on a single screen, and it
448resembles the code found in L<perlembed>; most of the real action takes
449place in F<perl.c>
450
9df8f87f
LB
451F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at
452make time, so you should make perl to follow this along.
453
a422fd2d 454First, F<perlmain.c> allocates some memory and constructs a Perl
9df8f87f 455interpreter, along these lines:
a422fd2d
SC
456
457 1 PERL_SYS_INIT3(&argc,&argv,&env);
458 2
459 3 if (!PL_do_undump) {
460 4 my_perl = perl_alloc();
461 5 if (!my_perl)
462 6 exit(1);
463 7 perl_construct(my_perl);
464 8 PL_perl_destruct_level = 0;
465 9 }
466
467Line 1 is a macro, and its definition is dependent on your operating
468system. Line 3 references C<PL_do_undump>, a global variable - all
469global variables in Perl start with C<PL_>. This tells you whether the
470current running program was created with the C<-u> flag to perl and then
471F<undump>, which means it's going to be false in any sane context.
472
473Line 4 calls a function in F<perl.c> to allocate memory for a Perl
474interpreter. It's quite a simple function, and the guts of it looks like
475this:
476
477 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
478
479Here you see an example of Perl's system abstraction, which we'll see
480later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
481own C<malloc> as defined in F<malloc.c> if you selected that option at
482configure time.
483
9df8f87f
LB
484Next, in line 7, we construct the interpreter using perl_construct,
485also in F<perl.c>; this sets up all the special variables that Perl
486needs, the stacks, and so on.
a422fd2d
SC
487
488Now we pass Perl the command line options, and tell it to go:
489
490 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
9df8f87f
LB
491 if (!exitstatus)
492 perl_run(my_perl);
493
494 exitstatus = perl_destruct(my_perl);
a422fd2d 495
9df8f87f 496 perl_free(my_perl);
a422fd2d
SC
497
498C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
499in F<perl.c>, which processes the command line options, sets up any
500statically linked XS modules, opens the program and calls C<yyparse> to
501parse it.
502
503=item Parsing
504
505The aim of this stage is to take the Perl source, and turn it into an op
506tree. We'll see what one of those looks like later. Strictly speaking,
507there's three things going on here.
508
509C<yyparse>, the parser, lives in F<perly.c>, although you're better off
510reading the original YACC input in F<perly.y>. (Yes, Virginia, there
511B<is> a YACC grammar for Perl!) The job of the parser is to take your
b432a672 512code and "understand" it, splitting it into sentences, deciding which
a422fd2d
SC
513operands go with which operators and so on.
514
515The parser is nobly assisted by the lexer, which chunks up your input
516into tokens, and decides what type of thing each token is: a variable
517name, an operator, a bareword, a subroutine, a core function, and so on.
518The main point of entry to the lexer is C<yylex>, and that and its
519associated routines can be found in F<toke.c>. Perl isn't much like
520other computer languages; it's highly context sensitive at times, it can
521be tricky to work out what sort of token something is, or where a token
522ends. As such, there's a lot of interplay between the tokeniser and the
523parser, which can get pretty frightening if you're not used to it.
524
525As the parser understands a Perl program, it builds up a tree of
526operations for the interpreter to perform during execution. The routines
527which construct and link together the various operations are to be found
528in F<op.c>, and will be examined later.
529
530=item Optimization
531
532Now the parsing stage is complete, and the finished tree represents
533the operations that the Perl interpreter needs to perform to execute our
534program. Next, Perl does a dry run over the tree looking for
535optimisations: constant expressions such as C<3 + 4> will be computed
536now, and the optimizer will also see if any multiple operations can be
537replaced with a single one. For instance, to fetch the variable C<$foo>,
538instead of grabbing the glob C<*foo> and looking at the scalar
539component, the optimizer fiddles the op tree to use a function which
540directly looks up the scalar in question. The main optimizer is C<peep>
541in F<op.c>, and many ops have their own optimizing functions.
542
543=item Running
544
545Now we're finally ready to go: we have compiled Perl byte code, and all
546that's left to do is run it. The actual execution is done by the
547C<runops_standard> function in F<run.c>; more specifically, it's done by
548these three innocent looking lines:
549
550 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
551 PERL_ASYNC_CHECK();
552 }
553
554You may be more comfortable with the Perl version of that:
555
556 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
557
558Well, maybe not. Anyway, each op contains a function pointer, which
559stipulates the function which will actually carry out the operation.
560This function will return the next op in the sequence - this allows for
561things like C<if> which choose the next op dynamically at run time.
562The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
563execution if required.
564
565The actual functions called are known as PP code, and they're spread
b432a672 566between four files: F<pp_hot.c> contains the "hot" code, which is most
a422fd2d
SC
567often used and highly optimized, F<pp_sys.c> contains all the
568system-specific functions, F<pp_ctl.c> contains the functions which
569implement control structures (C<if>, C<while> and the like) and F<pp.c>
570contains everything else. These are, if you like, the C code for Perl's
571built-in functions and operators.
572
dfc98234
DM
573Note that each C<pp_> function is expected to return a pointer to the next
574op. Calls to perl subs (and eval blocks) are handled within the same
575runops loop, and do not consume extra space on the C stack. For example,
576C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
577struct onto the context stack which contain the address of the op
578following the sub call or eval. They then return the first op of that sub
579or eval block, and so execution continues of that sub or block. Later, a
580C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
581retrieves the return op from it, and returns it.
582
583=item Exception handing
584
0503309d 585Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level
dfc98234 586C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
28a5cf3b 587way to capture the current PC and SP registers and later restore them; i.e.
dfc98234
DM
588a C<longjmp()> continues at the point in code where a previous C<setjmp()>
589was done, with anything further up on the C stack being lost. This is why
590code should always save values using C<SAVE_FOO> rather than in auto
591variables.
592
593The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
594C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
595C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
596C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
597
598At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
599C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
600loop or whatever, and handle possible exception returns. For a 2 return,
601final cleanup is performed, such as popping stacks and calling C<CHECK> or
602C<END> blocks. Amongst other things, this is how scope cleanup still
603occurs during an C<exit>.
604
605If a C<die> can find a C<CxEVAL> block on the context stack, then the
606stack is popped to that level and the return op in that block is assigned
607to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
608passes control back to the guard. In the case of C<perl_run> and
609C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
610loop. The is the normal way that C<die> or C<croak> is handled within an
611C<eval>.
612
613Sometimes ops are executed within an inner runops loop, such as tie, sort
614or overload code. In this case, something like
615
616 sub FETCH { eval { die } }
617
618would cause a longjmp right back to the guard in C<perl_run>, popping both
619runops loops, which is clearly incorrect. One way to avoid this is for the
620tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
621runops loop, but for efficiency reasons, perl in fact just sets a flag,
622using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
623C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
624which does a C<JMPENV_PUSH> and starts a new runops level to execute the
625code, rather than doing it on the current loop.
626
627As a further optimisation, on exit from the eval block in the C<FETCH>,
628execution of the code following the block is still carried on in the inner
629loop. When an exception is raised, C<docatch> compares the C<JMPENV>
630level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
631re-throws the exception. In this way any inner loops get popped.
632
633Here's an example.
634
635 1: eval { tie @a, 'A' };
636 2: sub A::TIEARRAY {
637 3: eval { die };
638 4: die;
639 5: }
640
641To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
642enters a runops loop. This loop executes the eval and tie ops on line 1,
643with the eval pushing a C<CxEVAL> onto the context stack.
644
645The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
646to execute the body of C<TIEARRAY>. When it executes the entertry op on
647line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
648does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
649the die op. At this point the C call stack looks like this:
650
651 Perl_pp_die
652 Perl_runops # third loop
653 S_docatch_body
654 S_docatch
655 Perl_pp_entertry
656 Perl_runops # second loop
657 S_call_body
658 Perl_call_sv
659 Perl_pp_tie
660 Perl_runops # first loop
661 S_run_body
662 perl_run
663 main
664
665and the context and data stacks, as shown by C<-Dstv>, look like:
666
667 STACK 0: MAIN
668 CX 0: BLOCK =>
669 CX 1: EVAL => AV() PV("A"\0)
670 retop=leave
671 STACK 1: MAGIC
672 CX 0: SUB =>
673 retop=(null)
674 CX 1: EVAL => *
675 retop=nextstate
676
677The die pops the first C<CxEVAL> off the context stack, sets
678C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
679the top C<docatch>. This then starts another third-level runops level,
680which executes the nextstate, pushmark and die ops on line 4. At the point
681that the second C<pp_die> is called, the C call stack looks exactly like
682that above, even though we are no longer within an inner eval; this is
683because of the optimization mentioned earlier. However, the context stack
684now looks like this, ie with the top CxEVAL popped:
685
686 STACK 0: MAIN
687 CX 0: BLOCK =>
688 CX 1: EVAL => AV() PV("A"\0)
689 retop=leave
690 STACK 1: MAGIC
691 CX 0: SUB =>
692 retop=(null)
693
694The die on line 4 pops the context stack back down to the CxEVAL, leaving
695it as:
696
697 STACK 0: MAIN
698 CX 0: BLOCK =>
699
700As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
701C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
702
703 S_docatch
704 Perl_pp_entertry
705 Perl_runops # second loop
706 S_call_body
707 Perl_call_sv
708 Perl_pp_tie
709 Perl_runops # first loop
710 S_run_body
711 perl_run
712 main
713
714In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
715differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
716and the C stack unwinds to:
717
718 perl_run
719 main
720
721Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
722and execution continues.
723
a422fd2d
SC
724=back
725
726=head2 Internal Variable Types
727
728You should by now have had a look at L<perlguts>, which tells you about
729Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
730that now.
731
732These variables are used not only to represent Perl-space variables, but
733also any constants in the code, as well as some structures completely
734internal to Perl. The symbol table, for instance, is an ordinary Perl
735hash. Your code is represented by an SV as it's read into the parser;
736any program files you call are opened via ordinary Perl filehandles, and
737so on.
738
739The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
740Perl program. Let's see, for instance, how Perl treats the constant
741C<"hello">.
742
743 % perl -MDevel::Peek -e 'Dump("hello")'
744 1 SV = PV(0xa041450) at 0xa04ecbc
745 2 REFCNT = 1
746 3 FLAGS = (POK,READONLY,pPOK)
747 4 PV = 0xa0484e0 "hello"\0
748 5 CUR = 5
749 6 LEN = 6
750
751Reading C<Devel::Peek> output takes a bit of practise, so let's go
752through it line by line.
753
754Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
755memory. SVs themselves are very simple structures, but they contain a
756pointer to a more complex structure. In this case, it's a PV, a
757structure which holds a string value, at location C<0xa041450>. Line 2
758is the reference count; there are no other references to this data, so
759it's 1.
760
761Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
762read-only SV (because it's a constant) and the data is a PV internally.
763Next we've got the contents of the string, starting at location
764C<0xa0484e0>.
765
766Line 5 gives us the current length of the string - note that this does
767B<not> include the null terminator. Line 6 is not the length of the
768string, but the length of the currently allocated buffer; as the string
769grows, Perl automatically extends the available storage via a routine
770called C<SvGROW>.
771
772You can get at any of these quantities from C very easily; just add
773C<Sv> to the name of the field shown in the snippet, and you've got a
774macro which will return the value: C<SvCUR(sv)> returns the current
775length of the string, C<SvREFCOUNT(sv)> returns the reference count,
776C<SvPV(sv, len)> returns the string itself with its length, and so on.
777More macros to manipulate these properties can be found in L<perlguts>.
778
779Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
780
781 1 void
782 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
783 3 {
784 4 STRLEN tlen;
785 5 char *junk;
786
787 6 junk = SvPV_force(sv, tlen);
788 7 SvGROW(sv, tlen + len + 1);
789 8 if (ptr == junk)
790 9 ptr = SvPVX(sv);
791 10 Move(ptr,SvPVX(sv)+tlen,len,char);
792 11 SvCUR(sv) += len;
793 12 *SvEND(sv) = '\0';
794 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
795 14 SvTAINT(sv);
796 15 }
797
798This is a function which adds a string, C<ptr>, of length C<len> onto
799the end of the PV stored in C<sv>. The first thing we do in line 6 is
800make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
801macro to force a PV. As a side effect, C<tlen> gets set to the current
802value of the PV, and the PV itself is returned to C<junk>.
803
b1866b2d 804In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d
SC
805the old string, the new string and the null terminator. If C<LEN> isn't
806big enough, C<SvGROW> will reallocate space for us.
807
808Now, if C<junk> is the same as the string we're trying to add, we can
809grab the string directly from the SV; C<SvPVX> is the address of the PV
810in the SV.
811
812Line 10 does the actual catenation: the C<Move> macro moves a chunk of
813memory around: we move the string C<ptr> to the end of the PV - that's
814the start of the PV plus its current length. We're moving C<len> bytes
815of type C<char>. After doing so, we need to tell Perl we've extended the
816string, by altering C<CUR> to reflect the new length. C<SvEND> is a
817macro which gives us the end of the string, so that needs to be a
818C<"\0">.
819
820Line 13 manipulates the flags; since we've changed the PV, any IV or NV
821values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1e54db1a 822want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
a422fd2d
SC
823version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
824and turns on POK. The final C<SvTAINT> is a macro which launders tainted
825data if taint mode is turned on.
826
827AVs and HVs are more complicated, but SVs are by far the most common
828variable type being thrown around. Having seen something of how we
829manipulate these, let's go on and look at how the op tree is
830constructed.
831
832=head2 Op Trees
833
834First, what is the op tree, anyway? The op tree is the parsed
835representation of your program, as we saw in our section on parsing, and
836it's the sequence of operations that Perl goes through to execute your
837program, as we saw in L</Running>.
838
839An op is a fundamental operation that Perl can perform: all the built-in
840functions and operators are ops, and there are a series of ops which
841deal with concepts the interpreter needs internally - entering and
842leaving a block, ending a statement, fetching a variable, and so on.
843
844The op tree is connected in two ways: you can imagine that there are two
845"routes" through it, two orders in which you can traverse the tree.
846First, parse order reflects how the parser understood the code, and
847secondly, execution order tells perl what order to perform the
848operations in.
849
850The easiest way to examine the op tree is to stop Perl after it has
851finished parsing, and get it to dump out the tree. This is exactly what
7d7d5695
RGS
852the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
853and L<B::Debug|B::Debug> do.
a422fd2d
SC
854
855Let's have a look at how Perl sees C<$a = $b + $c>:
856
857 % perl -MO=Terse -e '$a=$b+$c'
858 1 LISTOP (0x8179888) leave
859 2 OP (0x81798b0) enter
860 3 COP (0x8179850) nextstate
861 4 BINOP (0x8179828) sassign
862 5 BINOP (0x8179800) add [1]
863 6 UNOP (0x81796e0) null [15]
864 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
865 8 UNOP (0x81797e0) null [15]
866 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
867 10 UNOP (0x816b4f0) null [15]
868 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
869
870Let's start in the middle, at line 4. This is a BINOP, a binary
871operator, which is at location C<0x8179828>. The specific operator in
872question is C<sassign> - scalar assignment - and you can find the code
873which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
874binary operator, it has two children: the add operator, providing the
875result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
876line 10.
877
878Line 10 is the null op: this does exactly nothing. What is that doing
879there? If you see the null op, it's a sign that something has been
880optimized away after parsing. As we mentioned in L</Optimization>,
881the optimization stage sometimes converts two operations into one, for
882example when fetching a scalar variable. When this happens, instead of
883rewriting the op tree and cleaning up the dangling pointers, it's easier
884just to replace the redundant operation with the null op. Originally,
885the tree would have looked like this:
886
887 10 SVOP (0x816b4f0) rv2sv [15]
888 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
889
890That is, fetch the C<a> entry from the main symbol table, and then look
891at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
892happens to do both these things.
893
894The right hand side, starting at line 5 is similar to what we've just
895seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
896two C<gvsv>s.
897
898Now, what's this about?
899
900 1 LISTOP (0x8179888) leave
901 2 OP (0x81798b0) enter
902 3 COP (0x8179850) nextstate
903
904C<enter> and C<leave> are scoping ops, and their job is to perform any
905housekeeping every time you enter and leave a block: lexical variables
906are tidied up, unreferenced variables are destroyed, and so on. Every
907program will have those first three lines: C<leave> is a list, and its
908children are all the statements in the block. Statements are delimited
909by C<nextstate>, so a block is a collection of C<nextstate> ops, with
910the ops to be performed for each statement being the children of
911C<nextstate>. C<enter> is a single op which functions as a marker.
912
913That's how Perl parsed the program, from top to bottom:
914
915 Program
916 |
917 Statement
918 |
919 =
920 / \
921 / \
922 $a +
923 / \
924 $b $c
925
926However, it's impossible to B<perform> the operations in this order:
927you have to find the values of C<$b> and C<$c> before you add them
928together, for instance. So, the other thread that runs through the op
929tree is the execution order: each op has a field C<op_next> which points
930to the next op to be run, so following these pointers tells us how perl
931executes the code. We can traverse the tree in this order using
932the C<exec> option to C<B::Terse>:
933
934 % perl -MO=Terse,exec -e '$a=$b+$c'
935 1 OP (0x8179928) enter
936 2 COP (0x81798c8) nextstate
937 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
938 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
939 5 BINOP (0x8179878) add [1]
940 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
941 7 BINOP (0x81798a0) sassign
942 8 LISTOP (0x8179900) leave
943
944This probably makes more sense for a human: enter a block, start a
945statement. Get the values of C<$b> and C<$c>, and add them together.
946Find C<$a>, and assign one to the other. Then leave.
947
948The way Perl builds up these op trees in the parsing process can be
949unravelled by examining F<perly.y>, the YACC grammar. Let's take the
950piece we need to construct the tree for C<$a = $b + $c>
951
952 1 term : term ASSIGNOP term
953 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
954 3 | term ADDOP term
955 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
956
957If you're not used to reading BNF grammars, this is how it works: You're
958fed certain things by the tokeniser, which generally end up in upper
959case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
960code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
b432a672 961"terminal symbols", because you can't get any simpler than them.
a422fd2d
SC
962
963The grammar, lines one and three of the snippet above, tells you how to
b432a672 964build up more complex forms. These complex forms, "non-terminal symbols"
a422fd2d
SC
965are generally placed in lower case. C<term> here is a non-terminal
966symbol, representing a single expression.
967
968The grammar gives you the following rule: you can make the thing on the
969left of the colon if you see all the things on the right in sequence.
970This is called a "reduction", and the aim of parsing is to completely
971reduce the input. There are several different ways you can perform a
972reduction, separated by vertical bars: so, C<term> followed by C<=>
973followed by C<term> makes a C<term>, and C<term> followed by C<+>
974followed by C<term> can also make a C<term>.
975
976So, if you see two terms with an C<=> or C<+>, between them, you can
977turn them into a single expression. When you do this, you execute the
978code in the block on the next line: if you see C<=>, you'll do the code
979in line 2. If you see C<+>, you'll do the code in line 4. It's this code
980which contributes to the op tree.
981
982 | term ADDOP term
983 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
984
985What this does is creates a new binary op, and feeds it a number of
986variables. The variables refer to the tokens: C<$1> is the first token in
987the input, C<$2> the second, and so on - think regular expression
988backreferences. C<$$> is the op returned from this reduction. So, we
989call C<newBINOP> to create a new binary operator. The first parameter to
990C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
991operator, so we want the type to be C<ADDOP>. We could specify this
992directly, but it's right there as the second token in the input, so we
b432a672
AL
993use C<$2>. The second parameter is the op's flags: 0 means "nothing
994special". Then the things to add: the left and right hand side of our
a422fd2d
SC
995expression, in scalar context.
996
997=head2 Stacks
998
999When perl executes something like C<addop>, how does it pass on its
1000results to the next op? The answer is, through the use of stacks. Perl
1001has a number of stacks to store things it's currently working on, and
1002we'll look at the three most important ones here.
1003
1004=over 3
1005
1006=item Argument stack
1007
1008Arguments are passed to PP code and returned from PP code using the
1009argument stack, C<ST>. The typical way to handle arguments is to pop
1010them off the stack, deal with them how you wish, and then push the result
1011back onto the stack. This is how, for instance, the cosine operator
1012works:
1013
1014 NV value;
1015 value = POPn;
1016 value = Perl_cos(value);
1017 XPUSHn(value);
1018
1019We'll see a more tricky example of this when we consider Perl's macros
1020below. C<POPn> gives you the NV (floating point value) of the top SV on
1021the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1022the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1023should be extended if necessary - it can't be necessary here, because we
1024know there's room for one more item on the stack, since we've just
1025removed one! The C<XPUSH*> macros at least guarantee safety.
1026
1027Alternatively, you can fiddle with the stack directly: C<SP> gives you
1028the first element in your portion of the stack, and C<TOP*> gives you
1029the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1030negation of an integer:
1031
1032 SETi(-TOPi);
1033
1034Just set the integer value of the top stack entry to its negation.
1035
1036Argument stack manipulation in the core is exactly the same as it is in
1037XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1038description of the macros used in stack manipulation.
1039
1040=item Mark stack
1041
b432a672 1042I say "your portion of the stack" above because PP code doesn't
a422fd2d
SC
1043necessarily get the whole stack to itself: if your function calls
1044another function, you'll only want to expose the arguments aimed for the
1045called function, and not (necessarily) let it get at your own data. The
b432a672 1046way we do this is to have a "virtual" bottom-of-stack, exposed to each
a422fd2d
SC
1047function. The mark stack keeps bookmarks to locations in the argument
1048stack usable by each function. For instance, when dealing with a tied
b432a672 1049variable, (internally, something with "P" magic) Perl has to call
a422fd2d
SC
1050methods for accesses to the tied variables. However, we need to separate
1051the arguments exposed to the method to the argument exposed to the
ed233832
DM
1052original function - the store or fetch or whatever it may be. Here's
1053roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
a422fd2d
SC
1054
1055 1 PUSHMARK(SP);
1056 2 EXTEND(SP,2);
1057 3 PUSHs(SvTIED_obj((SV*)av, mg));
1058 4 PUSHs(val);
1059 5 PUTBACK;
1060 6 ENTER;
1061 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1062 8 LEAVE;
13a2d996 1063
a422fd2d
SC
1064Let's examine the whole implementation, for practice:
1065
1066 1 PUSHMARK(SP);
1067
1068Push the current state of the stack pointer onto the mark stack. This is
1069so that when we've finished adding items to the argument stack, Perl
1070knows how many things we've added recently.
1071
1072 2 EXTEND(SP,2);
1073 3 PUSHs(SvTIED_obj((SV*)av, mg));
1074 4 PUSHs(val);
1075
1076We're going to add two more items onto the argument stack: when you have
1077a tied array, the C<PUSH> subroutine receives the object and the value
1078to be pushed, and that's exactly what we have here - the tied object,
1079retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1080
1081 5 PUTBACK;
1082
e89a6d4e
JD
1083Next we tell Perl to update the global stack pointer from our internal
1084variable: C<dSP> only gave us a local copy, not a reference to the global.
a422fd2d
SC
1085
1086 6 ENTER;
1087 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1088 8 LEAVE;
1089
1090C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1091variables are tidied up, everything that has been localised gets
1092its previous value returned, and so on. Think of them as the C<{> and
1093C<}> of a Perl block.
1094
1095To actually do the magic method call, we have to call a subroutine in
1096Perl space: C<call_method> takes care of that, and it's described in
1097L<perlcall>. We call the C<PUSH> method in scalar context, and we're
e89a6d4e
JD
1098going to discard its return value. The call_method() function
1099removes the top element of the mark stack, so there is nothing for
1100the caller to clean up.
a422fd2d 1101
a422fd2d
SC
1102=item Save stack
1103
1104C doesn't have a concept of local scope, so perl provides one. We've
1105seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1106stack implements the C equivalent of, for example:
1107
1108 {
1109 local $foo = 42;
1110 ...
1111 }
1112
1113See L<perlguts/Localising Changes> for how to use the save stack.
1114
1115=back
1116
1117=head2 Millions of Macros
1118
1119One thing you'll notice about the Perl source is that it's full of
1120macros. Some have called the pervasive use of macros the hardest thing
1121to understand, others find it adds to clarity. Let's take an example,
1122the code which implements the addition operator:
1123
1124 1 PP(pp_add)
1125 2 {
39644a26 1126 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1127 4 {
1128 5 dPOPTOPnnrl_ul;
1129 6 SETn( left + right );
1130 7 RETURN;
1131 8 }
1132 9 }
1133
1134Every line here (apart from the braces, of course) contains a macro. The
1135first line sets up the function declaration as Perl expects for PP code;
1136line 3 sets up variable declarations for the argument stack and the
1137target, the return value of the operation. Finally, it tries to see if
1138the addition operation is overloaded; if so, the appropriate subroutine
1139is called.
1140
1141Line 5 is another variable declaration - all variable declarations start
1142with C<d> - which pops from the top of the argument stack two NVs (hence
1143C<nn>) and puts them into the variables C<right> and C<left>, hence the
1144C<rl>. These are the two operands to the addition operator. Next, we
1145call C<SETn> to set the NV of the return value to the result of adding
1146the two values. This done, we return - the C<RETURN> macro makes sure
1147that our return value is properly handled, and we pass the next operator
1148to run back to the main run loop.
1149
1150Most of these macros are explained in L<perlapi>, and some of the more
1151important ones are explained in L<perlxs> as well. Pay special attention
1152to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1153the C<[pad]THX_?> macros.
1154
52d59bef
JH
1155=head2 The .i Targets
1156
1157You can expand the macros in a F<foo.c> file by saying
1158
1159 make foo.i
1160
1161which will expand the macros using cpp. Don't be scared by the results.
1162
955fec6b
JH
1163=head1 SOURCE CODE STATIC ANALYSIS
1164
1165Various tools exist for analysing C source code B<statically>, as
1166opposed to B<dynamically>, that is, without executing the code.
1167It is possible to detect resource leaks, undefined behaviour, type
1168mismatches, portability problems, code paths that would cause illegal
1169memory accesses, and other similar problems by just parsing the C code
1170and looking at the resulting graph, what does it tell about the
1171execution and data flows. As a matter of fact, this is exactly
1172how C compilers know to give warnings about dubious code.
1173
1174=head2 lint, splint
1175
1176The good old C code quality inspector, C<lint>, is available in
1177several platforms, but please be aware that there are several
1178different implementations of it by different vendors, which means that
1179the flags are not identical across different platforms.
1180
1181There is a lint variant called C<splint> (Secure Programming Lint)
1182available from http://www.splint.org/ that should compile on any
1183Unix-like platform.
1184
1185There are C<lint> and <splint> targets in Makefile, but you may have
1186to diddle with the flags (see above).
1187
1188=head2 Coverity
1189
1190Coverity (http://www.coverity.com/) is a product similar to lint and
1191as a testbed for their product they periodically check several open
1192source projects, and they give out accounts to open source developers
1193to the defect databases.
1194
1195=head2 cpd (cut-and-paste detector)
1196
1197The cpd tool detects cut-and-paste coding. If one instance of the
1198cut-and-pasted code changes, all the other spots should probably be
1199changed, too. Therefore such code should probably be turned into a
1200subroutine or a macro.
1201
1202cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1203(http://pmd.sourceforge.net/). pmd was originally written for static
1204analysis of Java code, but later the cpd part of it was extended to
1205parse also C and C++.
1206
a52aaefa
RGS
1207Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1208pmd-X.Y.jar from it, and then run that on source code thusly:
955fec6b
JH
1209
1210 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1211
1212You may run into memory limits, in which case you should use the -Xmx option:
1213
1214 java -Xmx512M ...
1215
1216=head2 gcc warnings
1217
1218Though much can be written about the inconsistency and coverage
1219problems of gcc warnings (like C<-Wall> not meaning "all the
1220warnings", or some common portability problems not being covered by
1221C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1222collection of warnings, and so forth), gcc is still a useful tool in
1223keeping our coding nose clean.
1224
1225The C<-Wall> is by default on.
1226
a8e98a71
JH
1227The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1228always, but unfortunately they are not safe on all platforms, they can
1229for example cause fatal conflicts with the system headers (Solaris
1230being a prime example). If Configure C<-Dgccansipedantic> is used,
1231the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1232where they are known to be safe.
955fec6b
JH
1233
1234Starting from Perl 5.9.4 the following extra flags are added:
1235
1236=over 4
1237
1238=item *
1239
1240C<-Wendif-labels>
1241
1242=item *
1243
1244C<-Wextra>
1245
1246=item *
1247
1248C<-Wdeclaration-after-statement>
1249
1250=back
1251
1252The following flags would be nice to have but they would first need
0503309d 1253their own Augean stablemaster:
955fec6b
JH
1254
1255=over 4
1256
1257=item *
1258
1259C<-Wpointer-arith>
1260
1261=item *
1262
1263C<-Wshadow>
1264
1265=item *
1266
1267C<-Wstrict-prototypes>
1268
955fec6b
JH
1269=back
1270
1271The C<-Wtraditional> is another example of the annoying tendency of
1272gcc to bundle a lot of warnings under one switch -- it would be
1273impossible to deploy in practice because it would complain a lot -- but
1274it does contain some warnings that would be beneficial to have available
1275on their own, such as the warning about string constants inside macros
1276containing the macro arguments: this behaved differently pre-ANSI
1277than it does in ANSI, and some C compilers are still in transition,
1278AIX being an example.
1279
1280=head2 Warnings of other C compilers
1281
1282Other C compilers (yes, there B<are> other C compilers than gcc) often
1283have their "strict ANSI" or "strict ANSI with some portability extensions"
1284modes on, like for example the Sun Workshop has its C<-Xa> mode on
1285(though implicitly), or the DEC (these days, HP...) has its C<-std1>
1286mode on.
1287
1288=head2 DEBUGGING
1289
1290You can compile a special debugging version of Perl, which allows you
1291to use the C<-D> option of Perl to tell more about what Perl is doing.
1292But sometimes there is no alternative than to dive in with a debugger,
1293either to see the stack trace of a core dump (very useful in a bug
1294report), or trying to figure out what went wrong before the core dump
1295happened, or how did we end up having wrong or unexpected results.
1296
a422fd2d
SC
1297=head2 Poking at Perl
1298
1299To really poke around with Perl, you'll probably want to build Perl for
1300debugging, like this:
1301
1302 ./Configure -d -D optimize=-g
1303 make
1304
1305C<-g> is a flag to the C compiler to have it produce debugging
955fec6b
JH
1306information which will allow us to step through a running program,
1307and to see in which C function we are at (without the debugging
1308information we might see only the numerical addresses of the functions,
1309which is not very helpful).
1310
a422fd2d
SC
1311F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1312enables all the internal debugging code in Perl. There are a whole bunch
1313of things you can debug with this: L<perlrun> lists them all, and the
1314best way to find out about them is to play about with them. The most
1315useful options are probably
1316
1317 l Context (loop) stack processing
1318 t Trace execution
1319 o Method and overloading resolution
1320 c String/numeric conversions
1321
1322Some of the functionality of the debugging code can be achieved using XS
1323modules.
13a2d996 1324
a422fd2d
SC
1325 -Dr => use re 'debug'
1326 -Dx => use O 'Debug'
1327
1328=head2 Using a source-level debugger
1329
1330If the debugging output of C<-D> doesn't help you, it's time to step
1331through perl's execution with a source-level debugger.
1332
1333=over 3
1334
1335=item *
1336
955fec6b
JH
1337We'll use C<gdb> for our examples here; the principles will apply to
1338any debugger (many vendors call their debugger C<dbx>), but check the
1339manual of the one you're using.
a422fd2d
SC
1340
1341=back
1342
1343To fire up the debugger, type
1344
1345 gdb ./perl
1346
955fec6b
JH
1347Or if you have a core dump:
1348
1349 gdb ./perl core
1350
a422fd2d
SC
1351You'll want to do that in your Perl source tree so the debugger can read
1352the source code. You should see the copyright message, followed by the
1353prompt.
1354
1355 (gdb)
1356
1357C<help> will get you into the documentation, but here are the most
1358useful commands:
1359
1360=over 3
1361
1362=item run [args]
1363
1364Run the program with the given arguments.
1365
1366=item break function_name
1367
1368=item break source.c:xxx
1369
1370Tells the debugger that we'll want to pause execution when we reach
cea6626f 1371either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d
SC
1372line in the named source file.
1373
1374=item step
1375
1376Steps through the program a line at a time.
1377
1378=item next
1379
1380Steps through the program a line at a time, without descending into
1381functions.
1382
1383=item continue
1384
1385Run until the next breakpoint.
1386
1387=item finish
1388
1389Run until the end of the current function, then stop again.
1390
13a2d996 1391=item 'enter'
a422fd2d
SC
1392
1393Just pressing Enter will do the most recent operation again - it's a
1394blessing when stepping through miles of source code.
1395
1396=item print
1397
1398Execute the given C code and print its results. B<WARNING>: Perl makes
52d59bef
JH
1399heavy use of macros, and F<gdb> does not necessarily support macros
1400(see later L</"gdb macro support">). You'll have to substitute them
1401yourself, or to invoke cpp on the source code files
1402(see L</"The .i Targets">)
1403So, for instance, you can't say
a422fd2d
SC
1404
1405 print SvPV_nolen(sv)
1406
1407but you have to say
1408
1409 print Perl_sv_2pv_nolen(sv)
1410
ffc145e8
RK
1411=back
1412
a422fd2d
SC
1413You may find it helpful to have a "macro dictionary", which you can
1414produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
07aa3531 1415recursively apply those macros for you.
52d59bef
JH
1416
1417=head2 gdb macro support
a422fd2d 1418
52d59bef 1419Recent versions of F<gdb> have fairly good macro support, but
ea031e66
RGS
1420in order to use it you'll need to compile perl with macro definitions
1421included in the debugging information. Using F<gcc> version 3.1, this
1422means configuring with C<-Doptimize=-g3>. Other compilers might use a
1423different switch (if they support debugging macros at all).
1424
a422fd2d
SC
1425=head2 Dumping Perl Data Structures
1426
1427One way to get around this macro hell is to use the dumping functions in
1428F<dump.c>; these work a little like an internal
1429L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1430that you can't get at from Perl. Let's take an example. We'll use the
07aa3531 1431C<$a = $b + $c> we used before, but give it a bit of context:
a422fd2d
SC
1432C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1433
1434What about C<pp_add>, the function we examined earlier to implement the
1435C<+> operator:
1436
1437 (gdb) break Perl_pp_add
1438 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1439
cea6626f 1440Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d
SC
1441With the breakpoint in place, we can run our program:
1442
1443 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1444
1445Lots of junk will go past as gdb reads in the relevant source files and
1446libraries, and then:
1447
1448 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1449 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1450 (gdb) step
1451 311 dPOPTOPnnrl_ul;
1452 (gdb)
1453
1454We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1455arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1456slightly expand it:
1457
1458 #define dPOPTOPnnrl_ul NV right = POPn; \
1459 SV *leftsv = TOPs; \
1460 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1461
1462C<POPn> takes the SV from the top of the stack and obtains its NV either
1463directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1464C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1465C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
07aa3531 1466C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
a422fd2d
SC
1467
1468Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1469convert it. If we step again, we'll find ourselves there:
1470
1471 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1472 1669 if (!sv)
1473 (gdb)
1474
1475We can now use C<Perl_sv_dump> to investigate the SV:
1476
1477 SV = PV(0xa057cc0) at 0xa0675d0
1478 REFCNT = 1
1479 FLAGS = (POK,pPOK)
1480 PV = 0xa06a510 "6XXXX"\0
1481 CUR = 5
1482 LEN = 6
1483 $1 = void
1484
1485We know we're going to get C<6> from this, so let's finish the
1486subroutine:
1487
1488 (gdb) finish
1489 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1490 0x462669 in Perl_pp_add () at pp_hot.c:311
1491 311 dPOPTOPnnrl_ul;
1492
1493We can also dump out this op: the current op is always stored in
1494C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1495similar output to L<B::Debug|B::Debug>.
1496
1497 {
1498 13 TYPE = add ===> 14
1499 TARG = 1
1500 FLAGS = (SCALAR,KIDS)
1501 {
1502 TYPE = null ===> (12)
1503 (was rv2sv)
1504 FLAGS = (SCALAR,KIDS)
1505 {
1506 11 TYPE = gvsv ===> 12
1507 FLAGS = (SCALAR)
1508 GV = main::b
1509 }
1510 }
1511
10f58044 1512# finish this later #
a422fd2d
SC
1513
1514=head2 Patching
1515
1516All right, we've now had a look at how to navigate the Perl sources and
1517some things you'll need to know when fiddling with them. Let's now get
1518on and create a simple patch. Here's something Larry suggested: if a
07aa3531 1519C<U> is the first active format during a C<pack>, (for example,
a422fd2d 1520C<pack "U3C8", @stuff>) then the resulting string should be treated as
1e54db1a 1521UTF-8 encoded.
a422fd2d 1522
168a53cc
DR
1523If you are working with a git clone of the Perl repository, you will want to
1524create a branch for your changes. This will make creating a proper patch much
1525simpler. See the L<perlrepository> for details on how to do this.
1526
a422fd2d
SC
1527How do we prepare to fix this up? First we locate the code in question -
1528the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1529files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1530altering this file, let's copy it to F<pp.c~>.
1531
a6ec74c1
JH
1532[Well, it was in F<pp.c> when this tutorial was written. It has now been
1533split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1534
a422fd2d
SC
1535Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1536loop over the pattern, taking each format character in turn into
1537C<datum_type>. Then for each possible format character, we swallow up
1538the other arguments in the pattern (a field width, an asterisk, and so
1539on) and convert the next chunk input into the specified format, adding
1540it onto the output SV C<cat>.
1541
1542How do we know if the C<U> is the first format in the C<pat>? Well, if
1543we have a pointer to the start of C<pat> then, if we see a C<U> we can
1544test whether we're still at the start of the string. So, here's where
1545C<pat> is set up:
1546
1547 STRLEN fromlen;
1548 register char *pat = SvPVx(*++MARK, fromlen);
1549 register char *patend = pat + fromlen;
1550 register I32 len;
1551 I32 datumtype;
1552 SV *fromstr;
1553
1554We'll have another string pointer in there:
1555
1556 STRLEN fromlen;
1557 register char *pat = SvPVx(*++MARK, fromlen);
1558 register char *patend = pat + fromlen;
1559 + char *patcopy;
1560 register I32 len;
1561 I32 datumtype;
1562 SV *fromstr;
1563
1564And just before we start the loop, we'll set C<patcopy> to be the start
1565of C<pat>:
1566
1567 items = SP - MARK;
1568 MARK++;
1569 sv_setpvn(cat, "", 0);
1570 + patcopy = pat;
1571 while (pat < patend) {
1572
1573Now if we see a C<U> which was at the start of the string, we turn on
1e54db1a 1574the C<UTF8> flag for the output SV, C<cat>:
a422fd2d
SC
1575
1576 + if (datumtype == 'U' && pat==patcopy+1)
1577 + SvUTF8_on(cat);
1578 if (datumtype == '#') {
1579 while (pat < patend && *pat != '\n')
1580 pat++;
1581
1582Remember that it has to be C<patcopy+1> because the first character of
1583the string is the C<U> which has been swallowed into C<datumtype!>
1584
1585Oops, we forgot one thing: what if there are spaces at the start of the
1586pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1587character, even though it's not the first thing in the pattern. In this
1588case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1589
1590 if (isSPACE(datumtype))
1591 continue;
1592
1593needs to become
1594
1595 if (isSPACE(datumtype)) {
1596 patcopy++;
1597 continue;
1598 }
1599
1600OK. That's the C part done. Now we must do two additional things before
1601this patch is ready to go: we've changed the behaviour of Perl, and so
1602we must document that change. We must also provide some more regression
1603tests to make sure our patch works and doesn't create a bug somewhere
1604else along the line.
1605
b23b8711
MS
1606The regression tests for each operator live in F<t/op/>, and so we
1607make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1608tests to the end. First, we'll test that the C<U> does indeed create
07aa3531 1609Unicode strings.
b23b8711
MS
1610
1611t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1612use the one from t/test.pl.
b23b8711 1613
35c336e6
MS
1614 require './test.pl';
1615 plan( tests => 159 );
b23b8711
MS
1616
1617so instead of this:
a422fd2d
SC
1618
1619 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1620 print "ok $test\n"; $test++;
1621
35c336e6
MS
1622we can write the more sensible (see L<Test::More> for a full
1623explanation of is() and other testing functions).
b23b8711 1624
07aa3531 1625 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
38a44b82 1626 "U* produces Unicode" );
b23b8711 1627
a422fd2d
SC
1628Now we'll test that we got that space-at-the-beginning business right:
1629
35c336e6 1630 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1631 " with spaces at the beginning" );
a422fd2d
SC
1632
1633And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1634the first active format:
1635
35c336e6 1636 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
38a44b82 1637 "U* not first isn't Unicode" );
a422fd2d 1638
35c336e6
MS
1639Mustn't forget to change the number of tests which appears at the top,
1640or else the automated tester will get confused. This will either look
1641like this:
a422fd2d 1642
35c336e6
MS
1643 print "1..156\n";
1644
1645or this:
1646
1647 plan( tests => 156 );
a422fd2d
SC
1648
1649We now compile up Perl, and run it through the test suite. Our new
1650tests pass, hooray!
1651
1652Finally, the documentation. The job is never done until the paperwork is
1653over, so let's describe the change we've just made. The relevant place
1654is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1655this text in the description of C<pack>:
1656
1657 =item *
1658
1659 If the pattern begins with a C<U>, the resulting string will be treated
1e54db1a
JH
1660 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1661 with an initial C<U0>, and the bytes that follow will be interpreted as
1662 Unicode characters. If you don't want this to happen, you can begin your
1663 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
a422fd2d
SC
1664 string, and then follow this with a C<U*> somewhere in your pattern.
1665
f7e1e956
MS
1666=head2 Patching a core module
1667
1668This works just like patching anything else, with an extra
1669consideration. Many core modules also live on CPAN. If this is so,
1670patch the CPAN version instead of the core and send the patch off to
1671the module maintainer (with a copy to p5p). This will help the module
1672maintainer keep the CPAN version in sync with the core version without
1673constantly scanning p5p.
1674
db300100
RGS
1675The list of maintainers of core modules is usefully documented in
1676F<Porting/Maintainers.pl>.
1677
acbe17fc
JP
1678=head2 Adding a new function to the core
1679
1680If, as part of a patch to fix a bug, or just because you have an
1681especially good idea, you decide to add a new function to the core,
1682discuss your ideas on p5p well before you start work. It may be that
1683someone else has already attempted to do what you are considering and
1684can give lots of good advice or even provide you with bits of code
1685that they already started (but never finished).
1686
1687You have to follow all of the advice given above for patching. It is
1688extremely important to test any addition thoroughly and add new tests
1689to explore all boundary conditions that your new function is expected
1690to handle. If your new function is used only by one module (e.g. toke),
1691then it should probably be named S_your_function (for static); on the
210b36aa 1692other hand, if you expect it to accessible from other functions in
acbe17fc
JP
1693Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1694for more details.
1695
1696The location of any new code is also an important consideration. Don't
1697just create a new top level .c file and put your code there; you would
1698have to make changes to Configure (so the Makefile is created properly),
1699as well as possibly lots of include files. This is strictly pumpking
1700business.
1701
1702It is better to add your function to one of the existing top level
1703source code files, but your choice is complicated by the nature of
1704the Perl distribution. Only the files that are marked as compiled
1705static are located in the perl executable. Everything else is located
1706in the shared library (or DLL if you are running under WIN32). So,
1707for example, if a function was only used by functions located in
1708toke.c, then your code can go in toke.c. If, however, you want to call
1709the function from universal.c, then you should put your code in another
1710location, for example util.c.
1711
1712In addition to writing your c-code, you will need to create an
1713appropriate entry in embed.pl describing your function, then run
1714'make regen_headers' to create the entries in the numerous header
1715files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1716for information on the various options that you can set in embed.pl.
1717You will forget to do this a few (or many) times and you will get
1718warnings during the compilation phase. Make sure that you mention
1719this when you post your patch to P5P; the pumpking needs to know this.
1720
1721When you write your new code, please be conscious of existing code
884bad00 1722conventions used in the perl source files. See L<perlstyle> for
acbe17fc
JP
1723details. Although most of the guidelines discussed seem to focus on
1724Perl code, rather than c, they all apply (except when they don't ;).
552d2b87
DR
1725Also see I<perlrepository> for lots of details about both formatting and
1726submitting patches of your changes.
acbe17fc
JP
1727
1728Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1729Test on as many platforms as you can find. Test as many perl
1730Configure options as you can (e.g. MULTIPLICITY). If you have
1731profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 1732below for how to use them to further test your code. Remember that
acbe17fc
JP
1733most of the people on P5P are doing this on their own time and
1734don't have the time to debug your code.
f7e1e956
MS
1735
1736=head2 Writing a test
1737
1738Every module and built-in function has an associated test file (or
1739should...). If you add or change functionality, you have to write a
1740test. If you fix a bug, you have to write a test so that bug never
1741comes back. If you alter the docs, it would be nice to test what the
1742new documentation says.
1743
1744In short, if you submit a patch you probably also have to patch the
1745tests.
1746
1747For modules, the test file is right next to the module itself.
1748F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1749so there are some snags (and it would be wonderful for you to brush
1750them out), but it basically works that way. Everything else lives in
1751F<t/>.
1752
d5f28025
JV
1753If you add a new test directory under F<t/>, it is imperative that you
1754add that directory to F<t/HARNESS> and F<t/TEST>.
1755
f7e1e956
MS
1756=over 3
1757
1758=item F<t/base/>
1759
1760Testing of the absolute basic functionality of Perl. Things like
1761C<if>, basic file reads and writes, simple regexes, etc. These are
1762run first in the test suite and if any of them fail, something is
1763I<really> broken.
1764
1765=item F<t/cmd/>
1766
1767These test the basic control structures, C<if/else>, C<while>,
35c336e6 1768subroutines, etc.
f7e1e956
MS
1769
1770=item F<t/comp/>
1771
1772Tests basic issues of how Perl parses and compiles itself.
1773
1774=item F<t/io/>
1775
1776Tests for built-in IO functions, including command line arguments.
1777
1778=item F<t/lib/>
1779
1780The old home for the module tests, you shouldn't put anything new in
1781here. There are still some bits and pieces hanging around in here
1782that need to be moved. Perhaps you could move them? Thanks!
1783
3c295041
RGS
1784=item F<t/mro/>
1785
0503309d 1786Tests for perl's method resolution order implementations
3c295041
RGS
1787(see L<mro>).
1788
f7e1e956
MS
1789=item F<t/op/>
1790
1791Tests for perl's built in functions that don't fit into any of the
1792other directories.
1793
a4499558
YO
1794=item F<t/re/>
1795
1796Tests for regex related functions or behaviour. (These used to live
1797in t/op).
1798
f7e1e956
MS
1799=item F<t/run/>
1800
1801Testing features of how perl actually runs, including exit codes and
1802handling of PERL* environment variables.
1803
244d9cb7
RGS
1804=item F<t/uni/>
1805
1806Tests for the core support of Unicode.
1807
1808=item F<t/win32/>
1809
1810Windows-specific tests.
1811
1812=item F<t/x2p>
1813
1814A test suite for the s2p converter.
1815
f7e1e956
MS
1816=back
1817
1818The core uses the same testing style as the rest of Perl, a simple
1819"ok/not ok" run through Test::Harness, but there are a few special
1820considerations.
1821
35c336e6
MS
1822There are three ways to write a test in the core. Test::More,
1823t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1824decision of which to use depends on what part of the test suite you're
1825working on. This is a measure to prevent a high-level failure (such
1826as Config.pm breaking) from causing basic functionality tests to fail.
1827
07aa3531 1828=over 4
35c336e6
MS
1829
1830=item t/base t/comp
1831
1832Since we don't know if require works, or even subroutines, use ad hoc
1833tests for these two. Step carefully to avoid using the feature being
1834tested.
1835
1836=item t/cmd t/run t/io t/op
1837
1838Now that basic require() and subroutines are tested, you can use the
1839t/test.pl library which emulates the important features of Test::More
1840while using a minimum of core features.
1841
1842You can also conditionally use certain libraries like Config, but be
1843sure to skip the test gracefully if it's not there.
1844
1845=item t/lib ext lib
1846
1847Now that the core of Perl is tested, Test::More can be used. You can
1848also use the full suite of core modules in the tests.
1849
1850=back
f7e1e956
MS
1851
1852When you say "make test" Perl uses the F<t/TEST> program to run the
07aa3531
JC
1853test suite (except under Win32 where it uses F<t/harness> instead.)
1854All tests are run from the F<t/> directory, B<not> the directory
1855which contains the test. This causes some problems with the tests
7205a85d 1856in F<lib/>, so here's some opportunity for some patching.
f7e1e956
MS
1857
1858You must be triply conscious of cross-platform concerns. This usually
1859boils down to using File::Spec and avoiding things like C<fork()> and
1860C<system()> unless absolutely necessary.
1861
e018f8be
JH
1862=head2 Special Make Test Targets
1863
1864There are various special make targets that can be used to test Perl
1865slightly differently than the standard "test" target. Not all them
1866are expected to give a 100% success rate. Many of them have several
7205a85d
YO
1867aliases, and many of them are not available on certain operating
1868systems.
e018f8be
JH
1869
1870=over 4
1871
1872=item coretest
1873
7d7d5695 1874Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
e018f8be 1875
7205a85d
YO
1876(Not available on Win32)
1877
e018f8be
JH
1878=item test.deparse
1879
b26492ee
RGS
1880Run all the tests through B::Deparse. Not all tests will succeed.
1881
7205a85d
YO
1882(Not available on Win32)
1883
b26492ee
RGS
1884=item test.taintwarn
1885
1886Run all tests with the B<-t> command-line switch. Not all tests
1887are expected to succeed (until they're specifically fixed, of course).
e018f8be 1888
7205a85d
YO
1889(Not available on Win32)
1890
e018f8be
JH
1891=item minitest
1892
1893Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
8cebccf4 1894F<t/op>, F<t/uni> and F<t/mro> tests.
e018f8be 1895
7a834142
JH
1896=item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1897
1898(Only in Linux) Run all the tests using the memory leak + naughty
1899memory access tool "valgrind". The log files will be named
1900F<testname.valgrind>.
1901
e018f8be
JH
1902=item test.third check.third utest.third ucheck.third
1903
1904(Only in Tru64) Run all the tests using the memory leak + naughty
1905memory access tool "Third Degree". The log files will be named
60a57c1c 1906F<perl.3log.testname>.
e018f8be
JH
1907
1908=item test.torture torturetest
1909
1910Run all the usual tests and some extra tests. As of Perl 5.8.0 the
244d9cb7 1911only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
e018f8be
JH
1912
1913You can also run the torture test with F<t/harness> by giving
1914C<-torture> argument to F<t/harness>.
1915
1916=item utest ucheck test.utf8 check.utf8
1917
1918Run all the tests with -Mutf8. Not all tests will succeed.
1919
7205a85d
YO
1920(Not available on Win32)
1921
cc0710ff
RGS
1922=item minitest.utf16 test.utf16
1923
1924Runs the tests with UTF-16 encoded scripts, encoded with different
1925versions of this encoding.
1926
1927C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
1928C<-utf16> arguments to F<t/TEST>.
1929
7205a85d
YO
1930(Not available on Win32)
1931
244d9cb7
RGS
1932=item test_harness
1933
1934Run the test suite with the F<t/harness> controlling program, instead of
1935F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1936L<Test::Harness> module, thus using this test target supposes that perl
1937mostly works. The main advantage for our purposes is that it prints a
00bf5cd9
RGS
1938detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1939doesn't redirect stderr to stdout.
244d9cb7 1940
7205a85d
YO
1941Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
1942there is no special "test_harness" target.
1943
1944Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
1945environment variables to control the behaviour of F<t/harness>. This means
1946you can say
1947
1948 nmake test TEST_FILES="op/*.t"
1949 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
1950
a75f557c
JV
1951=item Parallel tests
1952
1953The core distribution can now run its regression tests in parallel on
1954Unix-like platforms. Instead of running C<make test>, set C<TEST_JOBS> in
1955your environment to the number of tests to run in parallel, and run
1956C<make test_harness>. On a Bourne-like shell, this can be done as
1957
1958 TEST_JOBS=3 make test_harness # Run 3 tests in parallel
1959
1960An environment variable is used, rather than parallel make itself, because
1961L<TAP::Harness> needs to be able to schedule individual non-conflicting test
1962scripts itself, and there is no standard interface to C<make> utilities to
1963interact with their job schedulers.
1964
1965Note that currently some test scripts may fail when run in parallel (most
1966notably C<ext/IO/t/io_dir.t>). If necessary run just the failing scripts
1967again sequentially and see if the failures go away.
7205a85d
YO
1968=item test-notty test_notty
1969
1970Sets PERL_SKIP_TTY_TEST to true before running normal test.
1971
244d9cb7
RGS
1972=back
1973
1974=head2 Running tests by hand
1975
1976You can run part of the test suite by hand by using one the following
1977commands from the F<t/> directory :
1978
1979 ./perl -I../lib TEST list-of-.t-files
1980
1981or
1982
1983 ./perl -I../lib harness list-of-.t-files
1984
1985(if you don't specify test scripts, the whole test suite will be run.)
1986
7205a85d
YO
1987=head3 Using t/harness for testing
1988
1989If you use C<harness> for testing you have several command line options
1990available to you. The arguments are as follows, and are in the order
1991that they must appear if used together.
1992
1993 harness -v -torture -re=pattern LIST OF FILES TO TEST
1994 harness -v -torture -re LIST OF PATTERNS TO MATCH
1995
1996If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
07aa3531 1997the manifest. The file list may include shell wildcards which will be
7205a85d
YO
1998expanded out.
1999
2000=over 4
2001
2002=item -v
2003
07aa3531 2004Run the tests under verbose mode so you can see what tests were run,
8550bf48 2005and debug output.
7205a85d
YO
2006
2007=item -torture
2008
2009Run the torture tests as well as the normal set.
2010
2011=item -re=PATTERN
2012
2013Filter the file list so that all the test files run match PATTERN.
2014Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2015in that it allows the file list to be provided as well.
2016
2017=item -re LIST OF PATTERNS
2018
07aa3531 2019Filter the file list so that all the test files run match
7205a85d
YO
2020/(LIST|OF|PATTERNS)/. Note that with this form the patterns
2021are joined by '|' and you cannot supply a list of files, instead
2022the test files are obtained from the MANIFEST.
2023
2024=back
2025
244d9cb7
RGS
2026You can run an individual test by a command similar to
2027
2028 ./perl -I../lib patho/to/foo.t
2029
2030except that the harnesses set up some environment variables that may
2031affect the execution of the test :
2032
07aa3531 2033=over 4
244d9cb7
RGS
2034
2035=item PERL_CORE=1
2036
2037indicates that we're running this test part of the perl core test suite.
2038This is useful for modules that have a dual life on CPAN.
2039
2040=item PERL_DESTRUCT_LEVEL=2
2041
2042is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2043
2044=item PERL
2045
2046(used only by F<t/TEST>) if set, overrides the path to the perl executable
2047that should be used to run the tests (the default being F<./perl>).
2048
2049=item PERL_SKIP_TTY_TEST
2050
2051if set, tells to skip the tests that need a terminal. It's actually set
2052automatically by the Makefile, but can also be forced artificially by
2053running 'make test_notty'.
2054
e018f8be 2055=back
f7e1e956 2056
7cd58830
RGS
2057=head3 Other environment variables that may influence tests
2058
2059=over 4
2060
2061=item PERL_TEST_Net_Ping
2062
2063Setting this variable runs all the Net::Ping modules tests,
2064otherwise some tests that interact with the outside world are skipped.
2065See L<perl58delta>.
2066
2067=item PERL_TEST_NOVREXX
2068
2069Setting this variable skips the vrexx.t tests for OS2::REXX.
2070
2071=item PERL_TEST_NUMCONVERTS
2072
2073This sets a variable in op/numconvert.t.
2074
2075=back
2076
2077See also the documentation for the Test and Test::Harness modules,
2078for more environment variables that affect testing.
2079
d7889f52
JH
2080=head2 Common problems when patching Perl source code
2081
2082Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2083some cases we have to take pre-ANSI requirements into consideration.
2084You don't care about some particular platform having broken Perl?
2085I hear there is still a strong demand for J2EE programmers.
2086
2087=head2 Perl environment problems
2088
2089=over 4
2090
2091=item *
2092
2093Not compiling with threading
2094
2095Compiling with threading (-Duseithreads) completely rewrites
2096the function prototypes of Perl. You better try your changes
0bec6c03 2097with that. Related to this is the difference between "Perl_-less"
d7889f52
JH
2098and "Perl_-ly" APIs, for example:
2099
2100 Perl_sv_setiv(aTHX_ ...);
2101 sv_setiv(...);
2102
ee9468a2
RGS
2103The first one explicitly passes in the context, which is needed for e.g.
2104threaded builds. The second one does that implicitly; do not get them
def4ed7d
JH
2105mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2106(or a dVAR) as the first thing in the function.
d7889f52
JH
2107
2108See L<perlguts/"How multiple interpreters and concurrency are supported">
2109for further discussion about context.
2110
2111=item *
2112
2113Not compiling with -DDEBUGGING
2114
2115The DEBUGGING define exposes more code to the compiler,
0bec6c03 2116therefore more ways for things to go wrong. You should try it.
d7889f52
JH
2117
2118=item *
2119
ee9468a2
RGS
2120Introducing (non-read-only) globals
2121
2122Do not introduce any modifiable globals, truly global or file static.
bc028b6b
JH
2123They are bad form and complicate multithreading and other forms of
2124concurrency. The right way is to introduce them as new interpreter
2125variables, see F<intrpvar.h> (at the very end for binary compatibility).
ee9468a2
RGS
2126
2127Introducing read-only (const) globals is okay, as long as you verify
2128with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2129BSD-style output) that the data you added really is read-only.
2130(If it is, it shouldn't show up in the output of that command.)
2131
2132If you want to have static strings, make them constant:
2133
2134 static const char etc[] = "...";
2135
bc028b6b 2136If you want to have arrays of constant strings, note carefully
ee9468a2
RGS
2137the right combination of C<const>s:
2138
2139 static const char * const yippee[] =
2140 {"hi", "ho", "silver"};
2141
bc028b6b
JH
2142There is a way to completely hide any modifiable globals (they are all
2143moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2144It is not normally used, but can be used for testing, read more
def4ed7d 2145about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
bc028b6b 2146
ee9468a2
RGS
2147=item *
2148
d7889f52
JH
2149Not exporting your new function
2150
2151Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2152function that is part of the public API (the shared Perl library)
2153to be explicitly marked as exported. See the discussion about
2154F<embed.pl> in L<perlguts>.
2155
2156=item *
2157
2158Exporting your new function
2159
2160The new shiny result of either genuine new functionality or your
2161arduous refactoring is now ready and correctly exported. So what
def4ed7d 2162could possibly go wrong?
d7889f52
JH
2163
2164Maybe simply that your function did not need to be exported in the
2165first place. Perl has a long and not so glorious history of exporting
2166functions that it should not have.
2167
2168If the function is used only inside one source code file, make it
2169static. See the discussion about F<embed.pl> in L<perlguts>.
2170
2171If the function is used across several files, but intended only for
2172Perl's internal use (and this should be the common case), do not
2173export it to the public API. See the discussion about F<embed.pl>
2174in L<perlguts>.
2175
2176=back
2177
5b38b9cd 2178=head2 Portability problems
d7889f52
JH
2179
2180The following are common causes of compilation and/or execution
2181failures, not common to Perl as such. The C FAQ is good bedtime
0bec6c03
JH
2182reading. Please test your changes with as many C compilers and
2183platforms as possible -- we will, anyway, and it's nice to save
2184oneself from public embarrassment.
2185
9aaf14db
RGS
2186If using gcc, you can add the C<-std=c89> option which will hopefully
2187catch most of these unportabilities. (However it might also catch
2188incompatibilities in your system's header files.)
d1307786 2189
a8e98a71
JH
2190Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2191C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2192
def4ed7d
JH
2193If using the C<gcc -Wall> note that not all the possible warnings
2194(like C<-Wunitialized>) are given unless you also compile with C<-O>.
2195
2196Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2197code files (the ones at the top level of the source code distribution,
2198but not e.g. the extensions under ext/) are automatically compiled
2199with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2200and a selection of C<-W> flags (see cflags.SH).
27565cb6 2201
0bec6c03 2202Also study L<perlport> carefully to avoid any bad assumptions
def4ed7d 2203about the operating system, filesystems, and so forth.
0bec6c03 2204
606fd33d 2205You may once in a while try a "make microperl" to see whether we
63796a85 2206can still compile Perl with just the bare minimum of interfaces.
606fd33d 2207(See README.micro.)
ee9468a2 2208
0bec6c03 2209Do not assume an operating system indicates a certain compiler.
d7889f52
JH
2210
2211=over 4
2212
2213=item *
2214
2215Casting pointers to integers or casting integers to pointers
2216
2217 void castaway(U8* p)
2218 {
2219 IV i = p;
2220
2221or
2222
2223 void castaway(U8* p)
2224 {
2225 IV i = (IV)p;
2226
ee9468a2 2227Both are bad, and broken, and unportable. Use the PTR2IV()
d7889f52
JH
2228macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2229INT2PTR(), and NUM2PTR().)
2230
2231=item *
2232
0bec6c03
JH
2233Casting between data function pointers and data pointers
2234
d7889f52
JH
2235Technically speaking casting between function pointers and data
2236pointers is unportable and undefined, but practically speaking
2237it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
0bec6c03 2238macros. Sometimes you can also play games with unions.
d7889f52
JH
2239
2240=item *
2241
2242Assuming sizeof(int) == sizeof(long)
2243
2244There are platforms where longs are 64 bits, and platforms where ints
2245are 64 bits, and while we are out to shock you, even platforms where
2246shorts are 64 bits. This is all legal according to the C standard.
2247(In other words, "long long" is not a portable way to specify 64 bits,
2248and "long long" is not even guaranteed to be any wider than "long".)
63796a85
JH
2249
2250Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2251Avoid things like I32 because they are B<not> guaranteed to be
2252I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2253guaranteed to be B<int> or B<long>. If you really explicitly need
225464-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
d7889f52
JH
2255
2256=item *
2257
2258Assuming one can dereference any type of pointer for any type of data
2259
2260 char *p = ...;
def4ed7d 2261 long pony = *p; /* BAD */
d7889f52
JH
2262
2263Many platforms, quite rightly so, will give you a core dump instead
2264of a pony if the p happens not be correctly aligned.
2265
2266=item *
2267
2268Lvalue casts
2269
def4ed7d 2270 (int)*p = ...; /* BAD */
d7889f52
JH
2271
2272Simply not portable. Get your lvalue to be of the right type,
27565cb6
JH
2273or maybe use temporary variables, or dirty tricks with unions.
2274
2275=item *
2276
606fd33d
JH
2277Assume B<anything> about structs (especially the ones you
2278don't control, like the ones coming from the system headers)
27565cb6
JH
2279
2280=over 8
2281
2282=item *
2283
2284That a certain field exists in a struct
2285
2286=item *
2287
902821cc 2288That no other fields exist besides the ones you know of
27565cb6
JH
2289
2290=item *
2291
606fd33d 2292That a field is of certain signedness, sizeof, or type
27565cb6
JH
2293
2294=item *
2295
2296That the fields are in a certain order
2297
606fd33d
JH
2298=over 8
2299
27565cb6
JH
2300=item *
2301
606fd33d
JH
2302While C guarantees the ordering specified in the struct definition,
2303between different platforms the definitions might differ
2304
2305=back
27565cb6
JH
2306
2307=item *
2308
606fd33d
JH
2309That the sizeof(struct) or the alignments are the same everywhere
2310
2311=over 8
27565cb6
JH
2312
2313=item *
2314
606fd33d
JH
2315There might be padding bytes between the fields to align the fields -
2316the bytes can be anything
2317
2318=item *
2319
2320Structs are required to be aligned to the maximum alignment required
2321by the fields - which for native types is for usually equivalent to
2322sizeof() of the field
2323
2324=back
27565cb6
JH
2325
2326=back
d7889f52
JH
2327
2328=item *
2329
2bbc8d55
SP
2330Assuming the character set is ASCIIish
2331
2332Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
2333This is transparent for the most part, but because the character sets
2334differ, you shouldn't use numeric (decimal, octal, nor hex) constants
2335to refer to characters. You can safely say 'A', but not 0x41.
2336You can safely say '\n', but not \012.
2337If a character doesn't have a trivial input form, you can
2338create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that
2339it resolves to different values depending on the character set being used.
2340(There are three different EBCDIC character sets defined in C<utfebcdic.h>,
2341so it might be best to insert the #define three times in that file.)
2342
2343Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case
2344alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'.
2345But '0' - '9' is an unbroken range in both systems. Don't assume anything
2346about other ranges.
2347
2348Many of the comments in the existing code ignore the possibility of EBCDIC,
2349and may be wrong therefore, even if the code works.
2350This is actually a tribute to the successful transparent insertion of being
fe749c9a 2351able to handle EBCDIC without having to change pre-existing code.
2bbc8d55
SP
2352
2353UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode
2354code points as sequences of bytes. Macros
2355with the same names (but different definitions)
2356in C<utf8.h> and C<utfebcdic.h>
fe749c9a
KW
2357are used to allow the calling code to think that there is only one such
2358encoding.
2359This is almost always referred to as C<utf8>, but it means the EBCDIC version
2360as well. Again, comments in the code may well be wrong even if the code itself
2361is right.
2bbc8d55
SP
2362For example, the concept of C<invariant characters> differs between ASCII and
2363EBCDIC.
2364On ASCII platforms, only characters that do not have the high-order
2365bit set (i.e. whose ordinals are strict ASCII, 0 - 127)
2366are invariant, and the documentation and comments in the code
2367may assume that,
2368often referring to something like, say, C<hibit>.
2369The situation differs and is not so simple on EBCDIC machines, but as long as
2370the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it
2371works, even if the comments are wrong.
2372
2373=item *
2374
2375Assuming the character set is just ASCII
2376
2377ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
2378characters have different meanings depending on the locale. Absent a locale,
2379currently these extra characters are generally considered to be unassigned,
2380and this has presented some problems.
2381This is scheduled to be changed in 5.12 so that these characters will
2382be considered to be Latin-1 (ISO-8859-1).
2383
2384=item *
2385
0bec6c03
JH
2386Mixing #define and #ifdef
2387
2388 #define BURGLE(x) ... \
def4ed7d 2389 #ifdef BURGLE_OLD_STYLE /* BAD */
0bec6c03
JH
2390 ... do it the old way ... \
2391 #else
2392 ... do it the new way ... \
2393 #endif
2394
ee9468a2
RGS
2395You cannot portably "stack" cpp directives. For example in the above
2396you need two separate BURGLE() #defines, one for each #ifdef branch.
2397
2398=item *
2399
2bbc8d55 2400Adding non-comment stuff after #endif or #else
ee9468a2
RGS
2401
2402 #ifdef SNOSH
2403 ...
def4ed7d 2404 #else !SNOSH /* BAD */
ee9468a2 2405 ...
def4ed7d 2406 #endif SNOSH /* BAD */
ee9468a2 2407
def4ed7d
JH
2408The #endif and #else cannot portably have anything non-comment after
2409them. If you want to document what is going (which is a good idea
2410especially if the branches are long), use (C) comments:
ee9468a2
RGS
2411
2412 #ifdef SNOSH
2413 ...
2414 #else /* !SNOSH */
2415 ...
2416 #endif /* SNOSH */
2417
2418The gcc option C<-Wendif-labels> warns about the bad variant
2419(by default on starting from Perl 5.9.4).
0bec6c03
JH
2420
2421=item *
2422
27565cb6
JH
2423Having a comma after the last element of an enum list
2424
2425 enum color {
2426 CERULEAN,
2427 CHARTREUSE,
def4ed7d 2428 CINNABAR, /* BAD */
27565cb6
JH
2429 };
2430
2431is not portable. Leave out the last comma.
2432
2433Also note that whether enums are implicitly morphable to ints
2434varies between compilers, you might need to (int).
2435
2436=item *
2437
d7889f52
JH
2438Using //-comments
2439
def4ed7d 2440 // This function bamfoodles the zorklator. /* BAD */
d7889f52
JH
2441
2442That is C99 or C++. Perl is C89. Using the //-comments is silently
0bec6c03
JH
2443allowed by many C compilers but cranking up the ANSI C89 strictness
2444(which we like to do) causes the compilation to fail.
d7889f52
JH
2445
2446=item *
2447
2448Mixing declarations and code
2449
2450 void zorklator()
2451 {
2452 int n = 3;
def4ed7d 2453 set_zorkmids(n); /* BAD */
d7889f52
JH
2454 int q = 4;
2455
0bec6c03
JH
2456That is C99 or C++. Some C compilers allow that, but you shouldn't.
2457
63796a85
JH
2458The gcc option C<-Wdeclaration-after-statements> scans for such problems
2459(by default on starting from Perl 5.9.4).
2460
0bec6c03
JH
2461=item *
2462
2463Introducing variables inside for()
2464
def4ed7d 2465 for(int i = ...; ...; ...) { /* BAD */
0bec6c03
JH
2466
2467That is C99 or C++. While it would indeed be awfully nice to have that
2468also in C89, to limit the scope of the loop variable, alas, we cannot.
d7889f52
JH
2469
2470=item *
2471
2472Mixing signed char pointers with unsigned char pointers
2473
2474 int foo(char *s) { ... }
2475 ...
2476 unsigned char *t = ...; /* Or U8* t = ... */
def4ed7d 2477 foo(t); /* BAD */
d7889f52
JH
2478
2479While this is legal practice, it is certainly dubious, and downright
2480fatal in at least one platform: for example VMS cc considers this a
def4ed7d
JH
2481fatal error. One cause for people often making this mistake is that a
2482"naked char" and therefore dereferencing a "naked char pointer" have
2483an undefined signedness: it depends on the compiler and the flags of
2484the compiler and the underlying platform whether the result is signed
2485or unsigned. For this very same reason using a 'char' as an array
2486index is bad.
d7889f52
JH
2487
2488=item *
2489
2490Macros that have string constants and their arguments as substrings of
2491the string constants
2492
def4ed7d 2493 #define FOO(n) printf("number = %d\n", n) /* BAD */
d7889f52
JH
2494 FOO(10);
2495
2496Pre-ANSI semantics for that was equivalent to
2497
2498 printf("10umber = %d\10");
2499
0bec6c03
JH
2500which is probably not what you were expecting. Unfortunately at least
2501one reasonably common and modern C compiler does "real backward
63796a85 2502compatibility" here, in AIX that is what still happens even though the
0bec6c03
JH
2503rest of the AIX compiler is very happily C89.
2504
2505=item *
2506
ee9468a2
RGS
2507Using printf formats for non-basic C types
2508
2509 IV i = ...;
def4ed7d 2510 printf("i = %d\n", i); /* BAD */
ee9468a2
RGS
2511
2512While this might by accident work in some platform (where IV happens
2513to be an C<int>), in general it cannot. IV might be something larger.
2514Even worse the situation is with more specific types (defined by Perl's
2515configuration step in F<config.h>):
2516
2517 Uid_t who = ...;
def4ed7d 2518 printf("who = %d\n", who); /* BAD */
ee9468a2
RGS
2519
2520The problem here is that Uid_t might be not only not C<int>-wide
2521but it might also be unsigned, in which case large uids would be
2522printed as negative values.
2523
2524There is no simple solution to this because of printf()'s limited
2525intelligence, but for many types the right format is available as
2526with either 'f' or '_f' suffix, for example:
2527
2528 IVdf /* IV in decimal */
2529 UVxf /* UV is hexadecimal */
2530
2531 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2532
2533 Uid_t_f /* Uid_t in decimal */
2534
2535 printf("who = %"Uid_t_f"\n", who);
2536
63796a85
JH
2537Or you can try casting to a "wide enough" type:
2538
2539 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2540
2541Also remember that the C<%p> format really does require a void pointer:
2542
2543 U8* p = ...;
2544 printf("p = %p\n", (void*)p);
2545
ee9468a2
RGS
2546The gcc option C<-Wformat> scans for such problems.
2547
2548=item *
2549
0bec6c03
JH
2550Blindly using variadic macros
2551
63796a85
JH
2552gcc has had them for a while with its own syntax, and C99 brought
2553them with a standardized syntax. Don't use the former, and use
2554the latter only if the HAS_C99_VARIADIC_MACROS is defined.
0bec6c03
JH
2555
2556=item *
2557
2558Blindly passing va_list
2559
2560Not all platforms support passing va_list to further varargs (stdarg)
2561functions. The right thing to do is to copy the va_list using the
2562Perl_va_copy() if the NEED_VA_COPY is defined.
d7889f52 2563
ee9468a2
RGS
2564=item *
2565
e5afc1ae 2566Using gcc statement expressions
63796a85 2567
def4ed7d 2568 val = ({...;...;...}); /* BAD */
63796a85 2569
def4ed7d
JH
2570While a nice extension, it's not portable. The Perl code does
2571admittedly use them if available to gain some extra speed
2572(essentially as a funky form of inlining), but you shouldn't.
63796a85
JH
2573
2574=item *
2575
2bbc8d55 2576Binding together several statements in a macro
63796a85
JH
2577
2578Use the macros STMT_START and STMT_END.
2579
2580 STMT_START {
2581 ...
2582 } STMT_END
2583
2584=item *
2585
ee9468a2
RGS
2586Testing for operating systems or versions when should be testing for features
2587
def4ed7d 2588 #ifdef __FOONIX__ /* BAD */
ee9468a2
RGS
2589 foo = quux();
2590 #endif
2591
2592Unless you know with 100% certainty that quux() is only ever available
2593for the "Foonix" operating system B<and> that is available B<and>
2594correctly working for B<all> past, present, B<and> future versions of
2595"Foonix", the above is very wrong. This is more correct (though still
2596not perfect, because the below is a compile-time check):
2597
2598 #ifdef HAS_QUUX
2599 foo = quux();
2600 #endif
2601
def4ed7d 2602How does the HAS_QUUX become defined where it needs to be? Well, if
353c6505 2603Foonix happens to be UNIXy enough to be able to run the Configure
ee9468a2
RGS
2604script, and Configure has been taught about detecting and testing
2605quux(), the HAS_QUUX will be correctly defined. In other platforms,
2606the corresponding configuration step will hopefully do the same.
2607
2608In a pinch, if you cannot wait for Configure to be educated,
2609or if you have a good hunch of where quux() might be available,
2610you can temporarily try the following:
2611
2612 #if (defined(__FOONIX__) || defined(__BARNIX__))
2613 # define HAS_QUUX
2614 #endif
2615
2616 ...
2617
2618 #ifdef HAS_QUUX
2619 foo = quux();
2620 #endif
2621
2622But in any case, try to keep the features and operating systems separate.
2623
d7889f52
JH
2624=back
2625
ad7244db
JH
2626=head2 Problematic System Interfaces
2627
2628=over 4
2629
2630=item *
2631
353c6505 2632malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
ad7244db
JH
2633allocate at least one byte. (In general you should rarely need to
2634work at this low level, but instead use the various malloc wrappers.)
2635
2636=item *
2637
2638snprintf() - the return type is unportable. Use my_snprintf() instead.
2639
2640=back
2641
d7889f52
JH
2642=head2 Security problems
2643
2644Last but not least, here are various tips for safer coding.
2645
2646=over 4
2647
2648=item *
2649
2650Do not use gets()
2651
2652Or we will publicly ridicule you. Seriously.
2653
2654=item *
2655
d1307786 2656Do not use strcpy() or strcat() or strncpy() or strncat()
d7889f52 2657
d1307786
JH
2658Use my_strlcpy() and my_strlcat() instead: they either use the native
2659implementation, or Perl's own implementation (borrowed from the public
2660domain implementation of INN).
d7889f52
JH
2661
2662=item *
2663
2664Do not use sprintf() or vsprintf()
2665
0bec6c03 2666If you really want just plain byte strings, use my_snprintf()
64d9b66b 2667and my_vsnprintf() instead, which will try to use snprintf() and
0bec6c03
JH
2668vsnprintf() if those safer APIs are available. If you want something
2669fancier than a plain byte string, use SVs and Perl_sv_catpvf().
d7889f52
JH
2670
2671=back
2672
902b9dbf
MLF
2673=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2674
2675Sometimes it helps to use external tools while debugging and
2676testing Perl. This section tries to guide you through using
2677some common testing and debugging tools with Perl. This is
2678meant as a guide to interfacing these tools with Perl, not
2679as any kind of guide to the use of the tools themselves.
2680
a958818a
JH
2681B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2682Third Degree greatly slows down the execution: seconds become minutes,
2683minutes become hours. For example as of Perl 5.8.1, the
2684ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2685e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2686than six hours, even on a snappy computer-- the said test must be
2687doing something that is quite unfriendly for memory debuggers. If you
2688don't feel like waiting, that you can simply kill away the perl
2689process.
2690
2691B<NOTE 2>: To minimize the number of memory leak false alarms (see
2692L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2693environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2694and harness scripts do that automatically. But if you are running
2695some of the tests manually-- for csh-like shells:
2696
2697 setenv PERL_DESTRUCT_LEVEL 2
2698
2699and for Bourne-type shells:
2700
2701 PERL_DESTRUCT_LEVEL=2
2702 export PERL_DESTRUCT_LEVEL
2703
2704or in UNIXy environments you can also use the C<env> command:
2705
2706 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
a1b65709 2707
37c0adeb
JH
2708B<NOTE 3>: There are known memory leaks when there are compile-time
2709errors within eval or require, seeing C<S_doeval> in the call stack
2710is a good sign of these. Fixing these leaks is non-trivial,
2711unfortunately, but they must be fixed eventually.
2712
f50e5b73
MH
2713B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2714unless Perl is built with the Configure option
2715C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2716
902b9dbf
MLF
2717=head2 Rational Software's Purify
2718
2719Purify is a commercial tool that is helpful in identifying
2720memory overruns, wild pointers, memory leaks and other such
2721badness. Perl must be compiled in a specific way for
2722optimal testing with Purify. Purify is available under
2723Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2724
902b9dbf
MLF
2725=head2 Purify on Unix
2726
2727On Unix, Purify creates a new Perl binary. To get the most
2728benefit out of Purify, you should create the perl to Purify
2729using:
2730
2731 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2732 -Uusemymalloc -Dusemultiplicity
2733
2734where these arguments mean:
2735
2736=over 4
2737
2738=item -Accflags=-DPURIFY
2739
2740Disables Perl's arena memory allocation functions, as well as
2741forcing use of memory allocation functions derived from the
2742system malloc.
2743
2744=item -Doptimize='-g'
2745
2746Adds debugging information so that you see the exact source
2747statements where the problem occurs. Without this flag, all
2748you will see is the source filename of where the error occurred.
2749
2750=item -Uusemymalloc
2751
2752Disable Perl's malloc so that Purify can more closely monitor
2753allocations and leaks. Using Perl's malloc will make Purify
2754report most leaks in the "potential" leaks category.
2755
2756=item -Dusemultiplicity
2757
2758Enabling the multiplicity option allows perl to clean up
2759thoroughly when the interpreter shuts down, which reduces the
2760number of bogus leak reports from Purify.
2761
2762=back
2763
2764Once you've compiled a perl suitable for Purify'ing, then you
2765can just:
2766
07aa3531 2767 make pureperl
902b9dbf
MLF
2768
2769which creates a binary named 'pureperl' that has been Purify'ed.
2770This binary is used in place of the standard 'perl' binary
2771when you want to debug Perl memory problems.
2772
2773As an example, to show any memory leaks produced during the
2774standard Perl testset you would create and run the Purify'ed
2775perl as:
2776
2777 make pureperl
2778 cd t
07aa3531 2779 ../pureperl -I../lib harness
902b9dbf
MLF
2780
2781which would run Perl on test.pl and report any memory problems.
2782
2783Purify outputs messages in "Viewer" windows by default. If
2784you don't have a windowing environment or if you simply
2785want the Purify output to unobtrusively go to a log file
2786instead of to the interactive window, use these following
2787options to output to the log file "perl.log":
2788
2789 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2790 -log-file=perl.log -append-logfile=yes"
2791
2792If you plan to use the "Viewer" windows, then you only need this option:
2793
2794 setenv PURIFYOPTIONS "-chain-length=25"
2795
c406981e
JH
2796In Bourne-type shells:
2797
98631ff8
JL
2798 PURIFYOPTIONS="..."
2799 export PURIFYOPTIONS
c406981e
JH
2800
2801or if you have the "env" utility:
2802
98631ff8 2803 env PURIFYOPTIONS="..." ../pureperl ...
c406981e 2804
902b9dbf
MLF
2805=head2 Purify on NT
2806
2807Purify on Windows NT instruments the Perl binary 'perl.exe'
2808on the fly. There are several options in the makefile you
2809should change to get the most use out of Purify:
2810
2811=over 4
2812
2813=item DEFINES
2814
2815You should add -DPURIFY to the DEFINES line so the DEFINES
2816line looks something like:
2817
07aa3531 2818 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
902b9dbf
MLF
2819
2820to disable Perl's arena memory allocation functions, as
2821well as to force use of memory allocation functions derived
2822from the system malloc.
2823
2824=item USE_MULTI = define
2825
2826Enabling the multiplicity option allows perl to clean up
2827thoroughly when the interpreter shuts down, which reduces the
2828number of bogus leak reports from Purify.
2829
2830=item #PERL_MALLOC = define
2831
2832Disable Perl's malloc so that Purify can more closely monitor
2833allocations and leaks. Using Perl's malloc will make Purify
2834report most leaks in the "potential" leaks category.
2835
2836=item CFG = Debug
2837
2838Adds debugging information so that you see the exact source
2839statements where the problem occurs. Without this flag, all
2840you will see is the source filename of where the error occurred.
2841
2842=back
2843
2844As an example, to show any memory leaks produced during the
2845standard Perl testset you would create and run Purify as:
2846
2847 cd win32
2848 make
2849 cd ../t
07aa3531 2850 purify ../perl -I../lib harness
902b9dbf
MLF
2851
2852which would instrument Perl in memory, run Perl on test.pl,
2853then finally report any memory problems.
2854
7a834142
JH
2855=head2 valgrind
2856
2857The excellent valgrind tool can be used to find out both memory leaks
9df8f87f
LB
2858and illegal memory accesses. As of version 3.3.0, Valgrind only
2859supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind"
2860target can be used to run the tests under valgrind. Found errors
2861and memory leaks are logged in files named F<testfile.valgrind>.
07aa3531
JC
2862
2863Valgrind also provides a cachegrind tool, invoked on perl as:
2864
038c294a 2865 VG_OPTS=--tool=cachegrind make test.valgrind
d44161bf
MHM
2866
2867As system libraries (most notably glibc) are also triggering errors,
2868valgrind allows to suppress such errors using suppression files. The
2869default suppression file that comes with valgrind already catches a lot
2870of them. Some additional suppressions are defined in F<t/perl.supp>.
7a834142
JH
2871
2872To get valgrind and for more information see
2873
2874 http://developer.kde.org/~sewardj/
2875
f134cc4e 2876=head2 Compaq's/Digital's/HP's Third Degree
09187cb1
JH
2877
2878Third Degree is a tool for memory leak detection and memory access checks.
2879It is one of the many tools in the ATOM toolkit. The toolkit is only
2880available on Tru64 (formerly known as Digital UNIX formerly known as
2881DEC OSF/1).
2882
2883When building Perl, you must first run Configure with -Doptimize=-g
2884and -Uusemymalloc flags, after that you can use the make targets
51a35ef1
JH
2885"perl.third" and "test.third". (What is required is that Perl must be
2886compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 2887
64cea5fd 2888The short story is that with "atom" you can instrument the Perl
83f0ef60 2889executable to create a new executable called F<perl.third>. When the
4ae3d70a 2890instrumented executable is run, it creates a log of dubious memory
83f0ef60 2891traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a
JH
2892third for more information. The most extensive Third Degree
2893documentation is available in the Compaq "Tru64 UNIX Programmer's
2894Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 2895
9c54ecba 2896The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
64cea5fd
JH
2897subdirectory. There is a problem with these files: Third Degree is so
2898effective that it finds problems also in the system libraries.
9c54ecba
JH
2899Therefore you should used the Porting/thirdclean script to cleanup
2900the F<*.3log> files.
64cea5fd
JH
2901
2902There are also leaks that for given certain definition of a leak,
2903aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2904
2905=head2 PERL_DESTRUCT_LEVEL
2906
a958818a
JH
2907If you want to run any of the tests yourself manually using e.g.
2908valgrind, or the pureperl or perl.third executables, please note that
2909by default perl B<does not> explicitly cleanup all the memory it has
2910allocated (such as global memory arenas) but instead lets the exit()
2911of the whole program "take care" of such allocations, also known as
2912"global destruction of objects".
64cea5fd
JH
2913
2914There is a way to tell perl to do complete cleanup: set the
2915environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2916The t/TEST wrapper does set this to 2, and this is what you
2917need to do too, if you don't want to see the "global leaks":
1f56d61a 2918For example, for "third-degreed" Perl:
64cea5fd 2919
1f56d61a 2920 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
09187cb1 2921
414f2397
RGS
2922(Note: the mod_perl apache module uses also this environment variable
2923for its own purposes and extended its semantics. Refer to the mod_perl
287a822c
RGS
2924documentation for more information. Also, spawned threads do the
2925equivalent of setting this variable to the value 1.)
5a6c59ef
DM
2926
2927If, at the end of a run you get the message I<N scalars leaked>, you can
fd0854ff
DM
2928recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2929of all those leaked SVs to be dumped along with details as to where each
2930SV was originally allocated. This information is also displayed by
2931Devel::Peek. Note that the extra details recorded with each SV increases
2932memory usage, so it shouldn't be used in production environments. It also
2933converts C<new_SV()> from a macro into a real function, so you can use
2934your favourite debugger to discover where those pesky SVs were allocated.
414f2397 2935
d7a2c63c
MHM
2936If you see that you're leaking memory at runtime, but neither valgrind
2937nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
2938leaking SVs that are still reachable and will be properly cleaned up
2939during destruction of the interpreter. In such cases, using the C<-Dm>
2940switch can point you to the source of the leak. If the executable was
2941built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations
2942in addition to memory allocations. Each SV allocation has a distinct
2943serial number that will be written on creation and destruction of the SV.
2944So if you're executing the leaking code in a loop, you need to look for
2945SVs that are created, but never destroyed between each cycle. If such an
2946SV is found, set a conditional breakpoint within C<new_SV()> and make it
2947break only when C<PL_sv_serial> is equal to the serial number of the
2948leaking SV. Then you will catch the interpreter in exactly the state
2949where the leaking SV is allocated, which is sufficient in many cases to
2950find the source of the leak.
2951
2952As C<-Dm> is using the PerlIO layer for output, it will by itself
2953allocate quite a bunch of SVs, which are hidden to avoid recursion.
2954You can bypass the PerlIO layer if you use the SV logging provided
2955by C<-DPERL_MEM_LOG> instead.
2956
46c6c7e2
JH
2957=head2 PERL_MEM_LOG
2958
10a879f5
JC
2959If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go
2960through logging functions, which is handy for breakpoint setting.
2961
2962Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging
2e5b5004 2963functions read $ENV{PERL_MEM_LOG} to determine whether to log the
10a879f5
JC
2964event, and if so how:
2965
2e5b5004
RGS
2966 $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops
2967 $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops
2968 $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log
2969 $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2)
10a879f5
JC
2970
2971Memory logging is somewhat similar to C<-Dm> but is independent of
2972C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(),
2e5b5004 2973and Safefree() are logged with the caller's source code file and line
10a879f5
JC
2974number (and C function name, if supported by the C compiler). In
2975contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging
2976is similar.
2977
2978Since the logging doesn't use PerlIO, all SV allocations are logged
2979and no extra SV allocations are introduced by enabling the logging.
2980If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for
2981each SV allocation is also logged.
d7a2c63c 2982
51a35ef1
JH
2983=head2 Profiling
2984
2985Depending on your platform there are various of profiling Perl.
2986
2987There are two commonly used techniques of profiling executables:
10f58044 2988I<statistical time-sampling> and I<basic-block counting>.
51a35ef1
JH
2989
2990The first method takes periodically samples of the CPU program
2991counter, and since the program counter can be correlated with the code
2992generated for functions, we get a statistical view of in which
2993functions the program is spending its time. The caveats are that very
2994small/fast functions have lower probability of showing up in the
2995profile, and that periodically interrupting the program (this is
2996usually done rather frequently, in the scale of milliseconds) imposes
2997an additional overhead that may skew the results. The first problem
2998can be alleviated by running the code for longer (in general this is a
2999good idea for profiling), the second problem is usually kept in guard
3000by the profiling tools themselves.
3001
10f58044 3002The second method divides up the generated code into I<basic blocks>.
51a35ef1
JH
3003Basic blocks are sections of code that are entered only in the
3004beginning and exited only at the end. For example, a conditional jump
3005starts a basic block. Basic block profiling usually works by
10f58044 3006I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1
JH
3007book-keeping code to the generated code. During the execution of the
3008code the basic block counters are then updated appropriately. The
3009caveat is that the added extra code can skew the results: again, the
3010profiling tools usually try to factor their own effects out of the
3011results.
3012
83f0ef60
JH
3013=head2 Gprof Profiling
3014
51a35ef1
JH
3015gprof is a profiling tool available in many UNIX platforms,
3016it uses F<statistical time-sampling>.
83f0ef60
JH
3017
3018You can build a profiled version of perl called "perl.gprof" by
51a35ef1
JH
3019invoking the make target "perl.gprof" (What is required is that Perl
3020must be compiled using the C<-pg> flag, you may need to re-Configure).
3021Running the profiled version of Perl will create an output file called
3022F<gmon.out> is created which contains the profiling data collected
3023during the execution.
83f0ef60
JH
3024
3025The gprof tool can then display the collected data in various ways.
3026Usually gprof understands the following options:
3027
3028=over 4
3029
3030=item -a
3031
3032Suppress statically defined functions from the profile.
3033
3034=item -b
3035
3036Suppress the verbose descriptions in the profile.
3037
3038=item -e routine
3039
3040Exclude the given routine and its descendants from the profile.
3041
3042=item -f routine
3043
3044Display only the given routine and its descendants in the profile.
3045
3046=item -s
3047
3048Generate a summary file called F<gmon.sum> which then may be given
3049to subsequent gprof runs to accumulate data over several runs.
3050
3051=item -z
3052
3053Display routines that have zero usage.
3054
3055=back
3056
3057For more detailed explanation of the available commands and output
3058formats, see your own local documentation of gprof.
3059
038c294a 3060quick hint:
07aa3531 3061
289d61c2
JL
3062 $ sh Configure -des -Dusedevel -Doptimize='-pg' && make perl.gprof
3063 $ ./perl.gprof someprog # creates gmon.out in current directory
3064 $ gprof ./perl.gprof > out
07aa3531
JC
3065 $ view out
3066
51a35ef1
JH
3067=head2 GCC gcov Profiling
3068
10f58044 3069Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1
JH
3070for the GNU CC.
3071
3072You can build a profiled version of perl called F<perl.gcov> by
3073invoking the make target "perl.gcov" (what is required that Perl must
3074be compiled using gcc with the flags C<-fprofile-arcs
3075-ftest-coverage>, you may need to re-Configure).
3076
3077Running the profiled version of Perl will cause profile output to be
3078generated. For each source file an accompanying ".da" file will be
3079created.
3080
3081To display the results you use the "gcov" utility (which should
3082be installed if you have gcc 3.0 or newer installed). F<gcov> is
3083run on source code files, like this
3084
3085 gcov sv.c
3086
3087which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3088contain the source code annotated with relative frequencies of
3089execution indicated by "#" markers.
3090
3091Useful options of F<gcov> include C<-b> which will summarise the
3092basic block, branch, and function call coverage, and C<-c> which
3093instead of relative frequencies will use the actual counts. For
3094more information on the use of F<gcov> and basic block profiling
3095with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3096
3097 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3098
3099and its section titled "8. gcov: a Test Coverage Program"
3100
3101 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3102
07aa3531
JC
3103quick hint:
3104
3105 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3106 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3107 $ rm -f regexec.c.gcov regexec.gcda
3108 $ ./perl.gcov
3109 $ gcov regexec.c
3110 $ view regexec.c.gcov
3111
4ae3d70a
JH
3112=head2 Pixie Profiling
3113
51a35ef1
JH
3114Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3115UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 3116I<basic-block counting>.
4ae3d70a 3117
83f0ef60 3118You can build a profiled version of perl called F<perl.pixie> by
51a35ef1
JH
3119invoking the make target "perl.pixie" (what is required is that Perl
3120must be compiled using the C<-g> flag, you may need to re-Configure).
3121
3122In Tru64 a file called F<perl.Addrs> will also be silently created,
3123this file contains the addresses of the basic blocks. Running the
3124profiled version of Perl will create a new file called "perl.Counts"
3125which contains the counts for the basic block for that particular
3126program execution.
4ae3d70a 3127
51a35ef1 3128To display the results you use the F<prof> utility. The exact
4ae3d70a
JH
3129incantation depends on your operating system, "prof perl.Counts" in
3130IRIX, and "prof -pixie -all -L. perl" in Tru64.
3131
6c41479b
JH
3132In IRIX the following prof options are available:
3133
3134=over 4
3135
3136=item -h
3137
3138Reports the most heavily used lines in descending order of use.
6e36760b 3139Useful for finding the hotspot lines.
6c41479b
JH
3140
3141=item -l
3142
3143Groups lines by procedure, with procedures sorted in descending order of use.
3144Within a procedure, lines are listed in source order.
6e36760b 3145Useful for finding the hotspots of procedures.
6c41479b
JH
3146
3147=back
3148
3149In Tru64 the following options are available:
3150
3151=over 4
3152
3958b146 3153=item -p[rocedures]
6c41479b 3154
3958b146 3155Procedures sorted in descending order by the number of cycles executed
6e36760b 3156in each procedure. Useful for finding the hotspot procedures.
6c41479b
JH
3157(This is the default option.)
3158
24000d2f 3159=item -h[eavy]
6c41479b 3160
6e36760b
JH
3161Lines sorted in descending order by the number of cycles executed in
3162each line. Useful for finding the hotspot lines.
6c41479b 3163
24000d2f 3164=item -i[nvocations]
6c41479b 3165
6e36760b
JH
3166The called procedures are sorted in descending order by number of calls
3167made to the procedures. Useful for finding the most used procedures.
6c41479b 3168
24000d2f 3169=item -l[ines]
6c41479b
JH
3170
3171Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 3172Useful for finding the hotspots of procedures.
6c41479b
JH
3173
3174=item -testcoverage
3175
3176The compiler emitted code for these lines, but the code was unexecuted.
3177
24000d2f 3178=item -z[ero]
6c41479b
JH
3179
3180Unexecuted procedures.
3181
aa500c9e 3182=back
6c41479b
JH
3183
3184For further information, see your system's manual pages for pixie and prof.
4ae3d70a 3185
b8ddf6b3
SB
3186=head2 Miscellaneous tricks
3187
3188=over 4
3189
3190=item *
3191
cc177e1a 3192Those debugging perl with the DDD frontend over gdb may find the
b8ddf6b3
SB
3193following useful:
3194
3195You can extend the data conversion shortcuts menu, so for example you
3196can display an SV's IV value with one click, without doing any typing.
3197To do that simply edit ~/.ddd/init file and add after:
3198
3199 ! Display shortcuts.
3200 Ddd*gdbDisplayShortcuts: \
3201 /t () // Convert to Bin\n\
3202 /d () // Convert to Dec\n\
3203 /x () // Convert to Hex\n\
3204 /o () // Convert to Oct(\n\
3205
3206the following two lines:
3207
3208 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3209 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3210
3211so now you can do ivx and pvx lookups or you can plug there the
3212sv_peek "conversion":
3213
3214 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3215
3216(The my_perl is for threaded builds.)
3217Just remember that every line, but the last one, should end with \n\
3218
3219Alternatively edit the init file interactively via:
32203rd mouse button -> New Display -> Edit Menu
3221
3222Note: you can define up to 20 conversion shortcuts in the gdb
3223section.
3224
9965345d
JH
3225=item *
3226
7e337ee0
JH
3227If you see in a debugger a memory area mysteriously full of 0xABABABAB
3228or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3229see L<perlclib>.
9965345d 3230
f1fac472
NC
3231=item *
3232
3233Under ithreads the optree is read only. If you want to enforce this, to check
3234for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3235enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3236that allocates op memory via C<mmap>, and sets it read-only at run time.
3237Any write access to an op results in a C<SIGBUS> and abort.
3238
3239This code is intended for development only, and may not be portable even to
3240all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3241all ops read only. Specifically it
3242
3243=over
3244
3245=item 1
3246
3247Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3248later via C<require> or C<eval> will be re-write
3249
3250=item 2
3251
3252Turns an entire slab of ops read-write if the refcount of any op in the slab
3253needs to be decreased.
3254
3255=item 3
3256
3257Turns an entire slab of ops read-write if any op from the slab is freed.
3258
b8ddf6b3
SB
3259=back
3260
f1fac472
NC
3261It's not possible to turn the slabs to read-only after an action requiring
3262read-write access, as either can happen during op tree building time, so
3263there may still be legitimate write access.
3264
3265However, as an 80% solution it is still effective, as currently it catches
3266a write access during the generation of F<Config.pm>, which means that we
3267can't yet build F<perl> with this enabled.
3268
3269=back
3270
3271
955fec6b 3272=head1 CONCLUSION
a422fd2d 3273
955fec6b
JH
3274We've had a brief look around the Perl source, how to maintain quality
3275of the source code, an overview of the stages F<perl> goes through
3276when it's running your code, how to use debuggers to poke at the Perl
3277guts, and finally how to analyse the execution of Perl. We took a very
3278simple problem and demonstrated how to solve it fully - with
3279documentation, regression tests, and finally a patch for submission to
3280p5p. Finally, we talked about how to use external tools to debug and
3281test Perl.
a422fd2d
SC
3282
3283I'd now suggest you read over those references again, and then, as soon
3284as possible, get your hands dirty. The best way to learn is by doing,
07aa3531 3285so:
a422fd2d
SC
3286
3287=over 3
3288
3289=item *
3290
3291Subscribe to perl5-porters, follow the patches and try and understand
3292them; don't be afraid to ask if there's a portion you're not clear on -
3293who knows, you may unearth a bug in the patch...
3294
3295=item *
3296
3297Keep up to date with the bleeding edge Perl distributions and get
3298familiar with the changes. Try and get an idea of what areas people are
3299working on and the changes they're making.
3300
3301=item *
3302
3e148164 3303Do read the README associated with your operating system, e.g. README.aix
a1f349fd
MB
3304on the IBM AIX OS. Don't hesitate to supply patches to that README if
3305you find anything missing or changed over a new OS release.
3306
3307=item *
3308
a422fd2d
SC
3309Find an area of Perl that seems interesting to you, and see if you can
3310work out how it works. Scan through the source, and step over it in the
3311debugger. Play, poke, investigate, fiddle! You'll probably get to
3312understand not just your chosen area but a much wider range of F<perl>'s
3313activity as well, and probably sooner than you'd think.
3314
3315=back
3316
3317=over 3
3318
3319=item I<The Road goes ever on and on, down from the door where it began.>
3320
3321=back
3322
64d9b66b 3323If you can do these things, you've started on the long road to Perl porting.
a422fd2d
SC
3324Thanks for wanting to help make Perl better - and happy hacking!
3325
4ac71550
TC
3326=head2 Metaphoric Quotations
3327
3328If you recognized the quote about the Road above, you're in luck.
3329
3330Most software projects begin each file with a literal description of each
3331file's purpose. Perl instead begins each with a literary allusion to that
3332file's purpose.
3333
3334Like chapters in many books, all top-level Perl source files (along with a
3335few others here and there) begin with an epigramic inscription that alludes,
3336indirectly and metaphorically, to the material you're about to read.
3337
3338Quotations are taken from writings of J.R.R Tolkien pertaining to his
3339Legendarium, almost always from I<The Lord of the Rings>. Chapters and
3340page numbers are given using the following editions:
3341
3342=over 4
3343
3344=item *
3345
3346I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary
3347edition of 2007 was used, published in the UK by Harper Collins Publishers
3348and in the US by the Houghton Mifflin Company.
3349
3350=item *
3351
3352I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover,
335350th-anniversary edition of 2004 was used, published in the UK by Harper
3354Collins Publishers and in the US by the Houghton Mifflin Company.
3355
3356=item *
3357
3358I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his
3359son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes
3360in Christopher's mammoth I<History of Middle Earth>. Page numbers derive
3361from the hardcover edition, first published in 1983 by George Allen &
3362Unwin; no page numbers changed for the special 3-volume omnibus edition of
33632002 or the various trade-paper editions, all again now by Harper Collins
3364or Houghton Mifflin.
3365
3366=back
3367
3368Other JRRT books fair game for quotes would thus include I<The Adventures of
3369Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of
3370the Children of Hurin>, all but the first posthumously assembled by CJRT.
3371But I<The Lord of the Rings> itself is perfectly fine and probably best to
3372quote from, provided you can find a suitable quote there.
3373
3374So if you were to supply a new, complete, top-level source file to add to
3375Perl, you should conform to this peculiar practice by yourself selecting an
3376appropriate quotation from Tolkien, retaining the original spelling and
3377punctuation and using the same format the rest of the quotes are in.
3378Indirect and oblique is just fine; remember, it's a metaphor, so being meta
3379is, after all, what it's for.
3380
e8cd7eae
GS
3381=head1 AUTHOR
3382
3383This document was written by Nathan Torkington, and is maintained by
3384the perl5-porters mailing list.
4ac71550 3385
b16c2e4a
RGS
3386=head1 SEE ALSO
3387
3388L<perlrepository>