This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
describe parallel testing in perlhack.pod
[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
299testing CPAN modules visit L<http://search.cpan.org/dist/CPAN-YACSmoke/>
300or L<http://search.cpan.org/dist/POE-Component-CPAN-YACSmoke/> or
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
368Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
369subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
370contains the core XS modules.
371
f7e1e956
MS
372=item Tests
373
374There are tests for nearly all the modules, built-ins and major bits
375of functionality. Test files all have a .t suffix. Module tests live
376in the F<lib/> and F<ext/> directories next to the module being
377tested. Others live in F<t/>. See L<Writing a test>
378
a422fd2d
SC
379=item Documentation
380
381Documentation maintenance includes looking after everything in the
382F<pod/> directory, (as well as contributing new documentation) and
383the documentation to the modules in core.
384
385=item Configure
386
387The configure process is the way we make Perl portable across the
388myriad of operating systems it supports. Responsibility for the
389configure, build and installation process, as well as the overall
390portability of the core code rests with the configure pumpkin - others
391help out with individual operating systems.
392
393The files involved are the operating system directories, (F<win32/>,
394F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
395and F<Makefile>, as well as the metaconfig files which generate
396F<Configure>. (metaconfig isn't included in the core distribution.)
397
398=item Interpreter
399
400And of course, there's the core of the Perl interpreter itself. Let's
401have a look at that in a little more detail.
402
403=back
404
405Before we leave looking at the layout, though, don't forget that
406F<MANIFEST> contains not only the file names in the Perl distribution,
407but short descriptions of what's in them, too. For an overview of the
408important files, try this:
409
410 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
411
412=head2 Elements of the interpreter
413
414The work of the interpreter has two main stages: compiling the code
415into the internal representation, or bytecode, and then executing it.
416L<perlguts/Compiled code> explains exactly how the compilation stage
417happens.
418
419Here is a short breakdown of perl's operation:
420
421=over 3
422
423=item Startup
424
425The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
426This is very high-level code, enough to fit on a single screen, and it
427resembles the code found in L<perlembed>; most of the real action takes
428place in F<perl.c>
429
9df8f87f
LB
430F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at
431make time, so you should make perl to follow this along.
432
a422fd2d 433First, F<perlmain.c> allocates some memory and constructs a Perl
9df8f87f 434interpreter, along these lines:
a422fd2d
SC
435
436 1 PERL_SYS_INIT3(&argc,&argv,&env);
437 2
438 3 if (!PL_do_undump) {
439 4 my_perl = perl_alloc();
440 5 if (!my_perl)
441 6 exit(1);
442 7 perl_construct(my_perl);
443 8 PL_perl_destruct_level = 0;
444 9 }
445
446Line 1 is a macro, and its definition is dependent on your operating
447system. Line 3 references C<PL_do_undump>, a global variable - all
448global variables in Perl start with C<PL_>. This tells you whether the
449current running program was created with the C<-u> flag to perl and then
450F<undump>, which means it's going to be false in any sane context.
451
452Line 4 calls a function in F<perl.c> to allocate memory for a Perl
453interpreter. It's quite a simple function, and the guts of it looks like
454this:
455
456 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
457
458Here you see an example of Perl's system abstraction, which we'll see
459later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
460own C<malloc> as defined in F<malloc.c> if you selected that option at
461configure time.
462
9df8f87f
LB
463Next, in line 7, we construct the interpreter using perl_construct,
464also in F<perl.c>; this sets up all the special variables that Perl
465needs, the stacks, and so on.
a422fd2d
SC
466
467Now we pass Perl the command line options, and tell it to go:
468
469 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
9df8f87f
LB
470 if (!exitstatus)
471 perl_run(my_perl);
472
473 exitstatus = perl_destruct(my_perl);
a422fd2d 474
9df8f87f 475 perl_free(my_perl);
a422fd2d
SC
476
477C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
478in F<perl.c>, which processes the command line options, sets up any
479statically linked XS modules, opens the program and calls C<yyparse> to
480parse it.
481
482=item Parsing
483
484The aim of this stage is to take the Perl source, and turn it into an op
485tree. We'll see what one of those looks like later. Strictly speaking,
486there's three things going on here.
487
488C<yyparse>, the parser, lives in F<perly.c>, although you're better off
489reading the original YACC input in F<perly.y>. (Yes, Virginia, there
490B<is> a YACC grammar for Perl!) The job of the parser is to take your
b432a672 491code and "understand" it, splitting it into sentences, deciding which
a422fd2d
SC
492operands go with which operators and so on.
493
494The parser is nobly assisted by the lexer, which chunks up your input
495into tokens, and decides what type of thing each token is: a variable
496name, an operator, a bareword, a subroutine, a core function, and so on.
497The main point of entry to the lexer is C<yylex>, and that and its
498associated routines can be found in F<toke.c>. Perl isn't much like
499other computer languages; it's highly context sensitive at times, it can
500be tricky to work out what sort of token something is, or where a token
501ends. As such, there's a lot of interplay between the tokeniser and the
502parser, which can get pretty frightening if you're not used to it.
503
504As the parser understands a Perl program, it builds up a tree of
505operations for the interpreter to perform during execution. The routines
506which construct and link together the various operations are to be found
507in F<op.c>, and will be examined later.
508
509=item Optimization
510
511Now the parsing stage is complete, and the finished tree represents
512the operations that the Perl interpreter needs to perform to execute our
513program. Next, Perl does a dry run over the tree looking for
514optimisations: constant expressions such as C<3 + 4> will be computed
515now, and the optimizer will also see if any multiple operations can be
516replaced with a single one. For instance, to fetch the variable C<$foo>,
517instead of grabbing the glob C<*foo> and looking at the scalar
518component, the optimizer fiddles the op tree to use a function which
519directly looks up the scalar in question. The main optimizer is C<peep>
520in F<op.c>, and many ops have their own optimizing functions.
521
522=item Running
523
524Now we're finally ready to go: we have compiled Perl byte code, and all
525that's left to do is run it. The actual execution is done by the
526C<runops_standard> function in F<run.c>; more specifically, it's done by
527these three innocent looking lines:
528
529 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
530 PERL_ASYNC_CHECK();
531 }
532
533You may be more comfortable with the Perl version of that:
534
535 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
536
537Well, maybe not. Anyway, each op contains a function pointer, which
538stipulates the function which will actually carry out the operation.
539This function will return the next op in the sequence - this allows for
540things like C<if> which choose the next op dynamically at run time.
541The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
542execution if required.
543
544The actual functions called are known as PP code, and they're spread
b432a672 545between four files: F<pp_hot.c> contains the "hot" code, which is most
a422fd2d
SC
546often used and highly optimized, F<pp_sys.c> contains all the
547system-specific functions, F<pp_ctl.c> contains the functions which
548implement control structures (C<if>, C<while> and the like) and F<pp.c>
549contains everything else. These are, if you like, the C code for Perl's
550built-in functions and operators.
551
dfc98234
DM
552Note that each C<pp_> function is expected to return a pointer to the next
553op. Calls to perl subs (and eval blocks) are handled within the same
554runops loop, and do not consume extra space on the C stack. For example,
555C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
556struct onto the context stack which contain the address of the op
557following the sub call or eval. They then return the first op of that sub
558or eval block, and so execution continues of that sub or block. Later, a
559C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
560retrieves the return op from it, and returns it.
561
562=item Exception handing
563
0503309d 564Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level
dfc98234 565C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
28a5cf3b 566way to capture the current PC and SP registers and later restore them; i.e.
dfc98234
DM
567a C<longjmp()> continues at the point in code where a previous C<setjmp()>
568was done, with anything further up on the C stack being lost. This is why
569code should always save values using C<SAVE_FOO> rather than in auto
570variables.
571
572The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
573C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
574C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
575C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
576
577At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
578C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
579loop or whatever, and handle possible exception returns. For a 2 return,
580final cleanup is performed, such as popping stacks and calling C<CHECK> or
581C<END> blocks. Amongst other things, this is how scope cleanup still
582occurs during an C<exit>.
583
584If a C<die> can find a C<CxEVAL> block on the context stack, then the
585stack is popped to that level and the return op in that block is assigned
586to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
587passes control back to the guard. In the case of C<perl_run> and
588C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
589loop. The is the normal way that C<die> or C<croak> is handled within an
590C<eval>.
591
592Sometimes ops are executed within an inner runops loop, such as tie, sort
593or overload code. In this case, something like
594
595 sub FETCH { eval { die } }
596
597would cause a longjmp right back to the guard in C<perl_run>, popping both
598runops loops, which is clearly incorrect. One way to avoid this is for the
599tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
600runops loop, but for efficiency reasons, perl in fact just sets a flag,
601using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
602C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
603which does a C<JMPENV_PUSH> and starts a new runops level to execute the
604code, rather than doing it on the current loop.
605
606As a further optimisation, on exit from the eval block in the C<FETCH>,
607execution of the code following the block is still carried on in the inner
608loop. When an exception is raised, C<docatch> compares the C<JMPENV>
609level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
610re-throws the exception. In this way any inner loops get popped.
611
612Here's an example.
613
614 1: eval { tie @a, 'A' };
615 2: sub A::TIEARRAY {
616 3: eval { die };
617 4: die;
618 5: }
619
620To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
621enters a runops loop. This loop executes the eval and tie ops on line 1,
622with the eval pushing a C<CxEVAL> onto the context stack.
623
624The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
625to execute the body of C<TIEARRAY>. When it executes the entertry op on
626line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
627does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
628the die op. At this point the C call stack looks like this:
629
630 Perl_pp_die
631 Perl_runops # third loop
632 S_docatch_body
633 S_docatch
634 Perl_pp_entertry
635 Perl_runops # second loop
636 S_call_body
637 Perl_call_sv
638 Perl_pp_tie
639 Perl_runops # first loop
640 S_run_body
641 perl_run
642 main
643
644and the context and data stacks, as shown by C<-Dstv>, look like:
645
646 STACK 0: MAIN
647 CX 0: BLOCK =>
648 CX 1: EVAL => AV() PV("A"\0)
649 retop=leave
650 STACK 1: MAGIC
651 CX 0: SUB =>
652 retop=(null)
653 CX 1: EVAL => *
654 retop=nextstate
655
656The die pops the first C<CxEVAL> off the context stack, sets
657C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
658the top C<docatch>. This then starts another third-level runops level,
659which executes the nextstate, pushmark and die ops on line 4. At the point
660that the second C<pp_die> is called, the C call stack looks exactly like
661that above, even though we are no longer within an inner eval; this is
662because of the optimization mentioned earlier. However, the context stack
663now looks like this, ie with the top CxEVAL popped:
664
665 STACK 0: MAIN
666 CX 0: BLOCK =>
667 CX 1: EVAL => AV() PV("A"\0)
668 retop=leave
669 STACK 1: MAGIC
670 CX 0: SUB =>
671 retop=(null)
672
673The die on line 4 pops the context stack back down to the CxEVAL, leaving
674it as:
675
676 STACK 0: MAIN
677 CX 0: BLOCK =>
678
679As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
680C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
681
682 S_docatch
683 Perl_pp_entertry
684 Perl_runops # second loop
685 S_call_body
686 Perl_call_sv
687 Perl_pp_tie
688 Perl_runops # first loop
689 S_run_body
690 perl_run
691 main
692
693In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
694differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
695and the C stack unwinds to:
696
697 perl_run
698 main
699
700Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
701and execution continues.
702
a422fd2d
SC
703=back
704
705=head2 Internal Variable Types
706
707You should by now have had a look at L<perlguts>, which tells you about
708Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
709that now.
710
711These variables are used not only to represent Perl-space variables, but
712also any constants in the code, as well as some structures completely
713internal to Perl. The symbol table, for instance, is an ordinary Perl
714hash. Your code is represented by an SV as it's read into the parser;
715any program files you call are opened via ordinary Perl filehandles, and
716so on.
717
718The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
719Perl program. Let's see, for instance, how Perl treats the constant
720C<"hello">.
721
722 % perl -MDevel::Peek -e 'Dump("hello")'
723 1 SV = PV(0xa041450) at 0xa04ecbc
724 2 REFCNT = 1
725 3 FLAGS = (POK,READONLY,pPOK)
726 4 PV = 0xa0484e0 "hello"\0
727 5 CUR = 5
728 6 LEN = 6
729
730Reading C<Devel::Peek> output takes a bit of practise, so let's go
731through it line by line.
732
733Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
734memory. SVs themselves are very simple structures, but they contain a
735pointer to a more complex structure. In this case, it's a PV, a
736structure which holds a string value, at location C<0xa041450>. Line 2
737is the reference count; there are no other references to this data, so
738it's 1.
739
740Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
741read-only SV (because it's a constant) and the data is a PV internally.
742Next we've got the contents of the string, starting at location
743C<0xa0484e0>.
744
745Line 5 gives us the current length of the string - note that this does
746B<not> include the null terminator. Line 6 is not the length of the
747string, but the length of the currently allocated buffer; as the string
748grows, Perl automatically extends the available storage via a routine
749called C<SvGROW>.
750
751You can get at any of these quantities from C very easily; just add
752C<Sv> to the name of the field shown in the snippet, and you've got a
753macro which will return the value: C<SvCUR(sv)> returns the current
754length of the string, C<SvREFCOUNT(sv)> returns the reference count,
755C<SvPV(sv, len)> returns the string itself with its length, and so on.
756More macros to manipulate these properties can be found in L<perlguts>.
757
758Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
759
760 1 void
761 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
762 3 {
763 4 STRLEN tlen;
764 5 char *junk;
765
766 6 junk = SvPV_force(sv, tlen);
767 7 SvGROW(sv, tlen + len + 1);
768 8 if (ptr == junk)
769 9 ptr = SvPVX(sv);
770 10 Move(ptr,SvPVX(sv)+tlen,len,char);
771 11 SvCUR(sv) += len;
772 12 *SvEND(sv) = '\0';
773 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
774 14 SvTAINT(sv);
775 15 }
776
777This is a function which adds a string, C<ptr>, of length C<len> onto
778the end of the PV stored in C<sv>. The first thing we do in line 6 is
779make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
780macro to force a PV. As a side effect, C<tlen> gets set to the current
781value of the PV, and the PV itself is returned to C<junk>.
782
b1866b2d 783In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d
SC
784the old string, the new string and the null terminator. If C<LEN> isn't
785big enough, C<SvGROW> will reallocate space for us.
786
787Now, if C<junk> is the same as the string we're trying to add, we can
788grab the string directly from the SV; C<SvPVX> is the address of the PV
789in the SV.
790
791Line 10 does the actual catenation: the C<Move> macro moves a chunk of
792memory around: we move the string C<ptr> to the end of the PV - that's
793the start of the PV plus its current length. We're moving C<len> bytes
794of type C<char>. After doing so, we need to tell Perl we've extended the
795string, by altering C<CUR> to reflect the new length. C<SvEND> is a
796macro which gives us the end of the string, so that needs to be a
797C<"\0">.
798
799Line 13 manipulates the flags; since we've changed the PV, any IV or NV
800values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1e54db1a 801want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
a422fd2d
SC
802version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
803and turns on POK. The final C<SvTAINT> is a macro which launders tainted
804data if taint mode is turned on.
805
806AVs and HVs are more complicated, but SVs are by far the most common
807variable type being thrown around. Having seen something of how we
808manipulate these, let's go on and look at how the op tree is
809constructed.
810
811=head2 Op Trees
812
813First, what is the op tree, anyway? The op tree is the parsed
814representation of your program, as we saw in our section on parsing, and
815it's the sequence of operations that Perl goes through to execute your
816program, as we saw in L</Running>.
817
818An op is a fundamental operation that Perl can perform: all the built-in
819functions and operators are ops, and there are a series of ops which
820deal with concepts the interpreter needs internally - entering and
821leaving a block, ending a statement, fetching a variable, and so on.
822
823The op tree is connected in two ways: you can imagine that there are two
824"routes" through it, two orders in which you can traverse the tree.
825First, parse order reflects how the parser understood the code, and
826secondly, execution order tells perl what order to perform the
827operations in.
828
829The easiest way to examine the op tree is to stop Perl after it has
830finished parsing, and get it to dump out the tree. This is exactly what
7d7d5695
RGS
831the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
832and L<B::Debug|B::Debug> do.
a422fd2d
SC
833
834Let's have a look at how Perl sees C<$a = $b + $c>:
835
836 % perl -MO=Terse -e '$a=$b+$c'
837 1 LISTOP (0x8179888) leave
838 2 OP (0x81798b0) enter
839 3 COP (0x8179850) nextstate
840 4 BINOP (0x8179828) sassign
841 5 BINOP (0x8179800) add [1]
842 6 UNOP (0x81796e0) null [15]
843 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
844 8 UNOP (0x81797e0) null [15]
845 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
846 10 UNOP (0x816b4f0) null [15]
847 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
848
849Let's start in the middle, at line 4. This is a BINOP, a binary
850operator, which is at location C<0x8179828>. The specific operator in
851question is C<sassign> - scalar assignment - and you can find the code
852which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
853binary operator, it has two children: the add operator, providing the
854result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
855line 10.
856
857Line 10 is the null op: this does exactly nothing. What is that doing
858there? If you see the null op, it's a sign that something has been
859optimized away after parsing. As we mentioned in L</Optimization>,
860the optimization stage sometimes converts two operations into one, for
861example when fetching a scalar variable. When this happens, instead of
862rewriting the op tree and cleaning up the dangling pointers, it's easier
863just to replace the redundant operation with the null op. Originally,
864the tree would have looked like this:
865
866 10 SVOP (0x816b4f0) rv2sv [15]
867 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
868
869That is, fetch the C<a> entry from the main symbol table, and then look
870at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
871happens to do both these things.
872
873The right hand side, starting at line 5 is similar to what we've just
874seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
875two C<gvsv>s.
876
877Now, what's this about?
878
879 1 LISTOP (0x8179888) leave
880 2 OP (0x81798b0) enter
881 3 COP (0x8179850) nextstate
882
883C<enter> and C<leave> are scoping ops, and their job is to perform any
884housekeeping every time you enter and leave a block: lexical variables
885are tidied up, unreferenced variables are destroyed, and so on. Every
886program will have those first three lines: C<leave> is a list, and its
887children are all the statements in the block. Statements are delimited
888by C<nextstate>, so a block is a collection of C<nextstate> ops, with
889the ops to be performed for each statement being the children of
890C<nextstate>. C<enter> is a single op which functions as a marker.
891
892That's how Perl parsed the program, from top to bottom:
893
894 Program
895 |
896 Statement
897 |
898 =
899 / \
900 / \
901 $a +
902 / \
903 $b $c
904
905However, it's impossible to B<perform> the operations in this order:
906you have to find the values of C<$b> and C<$c> before you add them
907together, for instance. So, the other thread that runs through the op
908tree is the execution order: each op has a field C<op_next> which points
909to the next op to be run, so following these pointers tells us how perl
910executes the code. We can traverse the tree in this order using
911the C<exec> option to C<B::Terse>:
912
913 % perl -MO=Terse,exec -e '$a=$b+$c'
914 1 OP (0x8179928) enter
915 2 COP (0x81798c8) nextstate
916 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
917 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
918 5 BINOP (0x8179878) add [1]
919 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
920 7 BINOP (0x81798a0) sassign
921 8 LISTOP (0x8179900) leave
922
923This probably makes more sense for a human: enter a block, start a
924statement. Get the values of C<$b> and C<$c>, and add them together.
925Find C<$a>, and assign one to the other. Then leave.
926
927The way Perl builds up these op trees in the parsing process can be
928unravelled by examining F<perly.y>, the YACC grammar. Let's take the
929piece we need to construct the tree for C<$a = $b + $c>
930
931 1 term : term ASSIGNOP term
932 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
933 3 | term ADDOP term
934 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
935
936If you're not used to reading BNF grammars, this is how it works: You're
937fed certain things by the tokeniser, which generally end up in upper
938case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
939code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
b432a672 940"terminal symbols", because you can't get any simpler than them.
a422fd2d
SC
941
942The grammar, lines one and three of the snippet above, tells you how to
b432a672 943build up more complex forms. These complex forms, "non-terminal symbols"
a422fd2d
SC
944are generally placed in lower case. C<term> here is a non-terminal
945symbol, representing a single expression.
946
947The grammar gives you the following rule: you can make the thing on the
948left of the colon if you see all the things on the right in sequence.
949This is called a "reduction", and the aim of parsing is to completely
950reduce the input. There are several different ways you can perform a
951reduction, separated by vertical bars: so, C<term> followed by C<=>
952followed by C<term> makes a C<term>, and C<term> followed by C<+>
953followed by C<term> can also make a C<term>.
954
955So, if you see two terms with an C<=> or C<+>, between them, you can
956turn them into a single expression. When you do this, you execute the
957code in the block on the next line: if you see C<=>, you'll do the code
958in line 2. If you see C<+>, you'll do the code in line 4. It's this code
959which contributes to the op tree.
960
961 | term ADDOP term
962 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
963
964What this does is creates a new binary op, and feeds it a number of
965variables. The variables refer to the tokens: C<$1> is the first token in
966the input, C<$2> the second, and so on - think regular expression
967backreferences. C<$$> is the op returned from this reduction. So, we
968call C<newBINOP> to create a new binary operator. The first parameter to
969C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
970operator, so we want the type to be C<ADDOP>. We could specify this
971directly, but it's right there as the second token in the input, so we
b432a672
AL
972use C<$2>. The second parameter is the op's flags: 0 means "nothing
973special". Then the things to add: the left and right hand side of our
a422fd2d
SC
974expression, in scalar context.
975
976=head2 Stacks
977
978When perl executes something like C<addop>, how does it pass on its
979results to the next op? The answer is, through the use of stacks. Perl
980has a number of stacks to store things it's currently working on, and
981we'll look at the three most important ones here.
982
983=over 3
984
985=item Argument stack
986
987Arguments are passed to PP code and returned from PP code using the
988argument stack, C<ST>. The typical way to handle arguments is to pop
989them off the stack, deal with them how you wish, and then push the result
990back onto the stack. This is how, for instance, the cosine operator
991works:
992
993 NV value;
994 value = POPn;
995 value = Perl_cos(value);
996 XPUSHn(value);
997
998We'll see a more tricky example of this when we consider Perl's macros
999below. C<POPn> gives you the NV (floating point value) of the top SV on
1000the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1001the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1002should be extended if necessary - it can't be necessary here, because we
1003know there's room for one more item on the stack, since we've just
1004removed one! The C<XPUSH*> macros at least guarantee safety.
1005
1006Alternatively, you can fiddle with the stack directly: C<SP> gives you
1007the first element in your portion of the stack, and C<TOP*> gives you
1008the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1009negation of an integer:
1010
1011 SETi(-TOPi);
1012
1013Just set the integer value of the top stack entry to its negation.
1014
1015Argument stack manipulation in the core is exactly the same as it is in
1016XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1017description of the macros used in stack manipulation.
1018
1019=item Mark stack
1020
b432a672 1021I say "your portion of the stack" above because PP code doesn't
a422fd2d
SC
1022necessarily get the whole stack to itself: if your function calls
1023another function, you'll only want to expose the arguments aimed for the
1024called function, and not (necessarily) let it get at your own data. The
b432a672 1025way we do this is to have a "virtual" bottom-of-stack, exposed to each
a422fd2d
SC
1026function. The mark stack keeps bookmarks to locations in the argument
1027stack usable by each function. For instance, when dealing with a tied
b432a672 1028variable, (internally, something with "P" magic) Perl has to call
a422fd2d
SC
1029methods for accesses to the tied variables. However, we need to separate
1030the arguments exposed to the method to the argument exposed to the
ed233832
DM
1031original function - the store or fetch or whatever it may be. Here's
1032roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
a422fd2d
SC
1033
1034 1 PUSHMARK(SP);
1035 2 EXTEND(SP,2);
1036 3 PUSHs(SvTIED_obj((SV*)av, mg));
1037 4 PUSHs(val);
1038 5 PUTBACK;
1039 6 ENTER;
1040 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1041 8 LEAVE;
13a2d996 1042
a422fd2d
SC
1043Let's examine the whole implementation, for practice:
1044
1045 1 PUSHMARK(SP);
1046
1047Push the current state of the stack pointer onto the mark stack. This is
1048so that when we've finished adding items to the argument stack, Perl
1049knows how many things we've added recently.
1050
1051 2 EXTEND(SP,2);
1052 3 PUSHs(SvTIED_obj((SV*)av, mg));
1053 4 PUSHs(val);
1054
1055We're going to add two more items onto the argument stack: when you have
1056a tied array, the C<PUSH> subroutine receives the object and the value
1057to be pushed, and that's exactly what we have here - the tied object,
1058retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1059
1060 5 PUTBACK;
1061
e89a6d4e
JD
1062Next we tell Perl to update the global stack pointer from our internal
1063variable: C<dSP> only gave us a local copy, not a reference to the global.
a422fd2d
SC
1064
1065 6 ENTER;
1066 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1067 8 LEAVE;
1068
1069C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1070variables are tidied up, everything that has been localised gets
1071its previous value returned, and so on. Think of them as the C<{> and
1072C<}> of a Perl block.
1073
1074To actually do the magic method call, we have to call a subroutine in
1075Perl space: C<call_method> takes care of that, and it's described in
1076L<perlcall>. We call the C<PUSH> method in scalar context, and we're
e89a6d4e
JD
1077going to discard its return value. The call_method() function
1078removes the top element of the mark stack, so there is nothing for
1079the caller to clean up.
a422fd2d 1080
a422fd2d
SC
1081=item Save stack
1082
1083C doesn't have a concept of local scope, so perl provides one. We've
1084seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1085stack implements the C equivalent of, for example:
1086
1087 {
1088 local $foo = 42;
1089 ...
1090 }
1091
1092See L<perlguts/Localising Changes> for how to use the save stack.
1093
1094=back
1095
1096=head2 Millions of Macros
1097
1098One thing you'll notice about the Perl source is that it's full of
1099macros. Some have called the pervasive use of macros the hardest thing
1100to understand, others find it adds to clarity. Let's take an example,
1101the code which implements the addition operator:
1102
1103 1 PP(pp_add)
1104 2 {
39644a26 1105 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1106 4 {
1107 5 dPOPTOPnnrl_ul;
1108 6 SETn( left + right );
1109 7 RETURN;
1110 8 }
1111 9 }
1112
1113Every line here (apart from the braces, of course) contains a macro. The
1114first line sets up the function declaration as Perl expects for PP code;
1115line 3 sets up variable declarations for the argument stack and the
1116target, the return value of the operation. Finally, it tries to see if
1117the addition operation is overloaded; if so, the appropriate subroutine
1118is called.
1119
1120Line 5 is another variable declaration - all variable declarations start
1121with C<d> - which pops from the top of the argument stack two NVs (hence
1122C<nn>) and puts them into the variables C<right> and C<left>, hence the
1123C<rl>. These are the two operands to the addition operator. Next, we
1124call C<SETn> to set the NV of the return value to the result of adding
1125the two values. This done, we return - the C<RETURN> macro makes sure
1126that our return value is properly handled, and we pass the next operator
1127to run back to the main run loop.
1128
1129Most of these macros are explained in L<perlapi>, and some of the more
1130important ones are explained in L<perlxs> as well. Pay special attention
1131to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1132the C<[pad]THX_?> macros.
1133
52d59bef
JH
1134=head2 The .i Targets
1135
1136You can expand the macros in a F<foo.c> file by saying
1137
1138 make foo.i
1139
1140which will expand the macros using cpp. Don't be scared by the results.
1141
955fec6b
JH
1142=head1 SOURCE CODE STATIC ANALYSIS
1143
1144Various tools exist for analysing C source code B<statically>, as
1145opposed to B<dynamically>, that is, without executing the code.
1146It is possible to detect resource leaks, undefined behaviour, type
1147mismatches, portability problems, code paths that would cause illegal
1148memory accesses, and other similar problems by just parsing the C code
1149and looking at the resulting graph, what does it tell about the
1150execution and data flows. As a matter of fact, this is exactly
1151how C compilers know to give warnings about dubious code.
1152
1153=head2 lint, splint
1154
1155The good old C code quality inspector, C<lint>, is available in
1156several platforms, but please be aware that there are several
1157different implementations of it by different vendors, which means that
1158the flags are not identical across different platforms.
1159
1160There is a lint variant called C<splint> (Secure Programming Lint)
1161available from http://www.splint.org/ that should compile on any
1162Unix-like platform.
1163
1164There are C<lint> and <splint> targets in Makefile, but you may have
1165to diddle with the flags (see above).
1166
1167=head2 Coverity
1168
1169Coverity (http://www.coverity.com/) is a product similar to lint and
1170as a testbed for their product they periodically check several open
1171source projects, and they give out accounts to open source developers
1172to the defect databases.
1173
1174=head2 cpd (cut-and-paste detector)
1175
1176The cpd tool detects cut-and-paste coding. If one instance of the
1177cut-and-pasted code changes, all the other spots should probably be
1178changed, too. Therefore such code should probably be turned into a
1179subroutine or a macro.
1180
1181cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1182(http://pmd.sourceforge.net/). pmd was originally written for static
1183analysis of Java code, but later the cpd part of it was extended to
1184parse also C and C++.
1185
a52aaefa
RGS
1186Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1187pmd-X.Y.jar from it, and then run that on source code thusly:
955fec6b
JH
1188
1189 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1190
1191You may run into memory limits, in which case you should use the -Xmx option:
1192
1193 java -Xmx512M ...
1194
1195=head2 gcc warnings
1196
1197Though much can be written about the inconsistency and coverage
1198problems of gcc warnings (like C<-Wall> not meaning "all the
1199warnings", or some common portability problems not being covered by
1200C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1201collection of warnings, and so forth), gcc is still a useful tool in
1202keeping our coding nose clean.
1203
1204The C<-Wall> is by default on.
1205
a8e98a71
JH
1206The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1207always, but unfortunately they are not safe on all platforms, they can
1208for example cause fatal conflicts with the system headers (Solaris
1209being a prime example). If Configure C<-Dgccansipedantic> is used,
1210the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1211where they are known to be safe.
955fec6b
JH
1212
1213Starting from Perl 5.9.4 the following extra flags are added:
1214
1215=over 4
1216
1217=item *
1218
1219C<-Wendif-labels>
1220
1221=item *
1222
1223C<-Wextra>
1224
1225=item *
1226
1227C<-Wdeclaration-after-statement>
1228
1229=back
1230
1231The following flags would be nice to have but they would first need
0503309d 1232their own Augean stablemaster:
955fec6b
JH
1233
1234=over 4
1235
1236=item *
1237
1238C<-Wpointer-arith>
1239
1240=item *
1241
1242C<-Wshadow>
1243
1244=item *
1245
1246C<-Wstrict-prototypes>
1247
955fec6b
JH
1248=back
1249
1250The C<-Wtraditional> is another example of the annoying tendency of
1251gcc to bundle a lot of warnings under one switch -- it would be
1252impossible to deploy in practice because it would complain a lot -- but
1253it does contain some warnings that would be beneficial to have available
1254on their own, such as the warning about string constants inside macros
1255containing the macro arguments: this behaved differently pre-ANSI
1256than it does in ANSI, and some C compilers are still in transition,
1257AIX being an example.
1258
1259=head2 Warnings of other C compilers
1260
1261Other C compilers (yes, there B<are> other C compilers than gcc) often
1262have their "strict ANSI" or "strict ANSI with some portability extensions"
1263modes on, like for example the Sun Workshop has its C<-Xa> mode on
1264(though implicitly), or the DEC (these days, HP...) has its C<-std1>
1265mode on.
1266
1267=head2 DEBUGGING
1268
1269You can compile a special debugging version of Perl, which allows you
1270to use the C<-D> option of Perl to tell more about what Perl is doing.
1271But sometimes there is no alternative than to dive in with a debugger,
1272either to see the stack trace of a core dump (very useful in a bug
1273report), or trying to figure out what went wrong before the core dump
1274happened, or how did we end up having wrong or unexpected results.
1275
a422fd2d
SC
1276=head2 Poking at Perl
1277
1278To really poke around with Perl, you'll probably want to build Perl for
1279debugging, like this:
1280
1281 ./Configure -d -D optimize=-g
1282 make
1283
1284C<-g> is a flag to the C compiler to have it produce debugging
955fec6b
JH
1285information which will allow us to step through a running program,
1286and to see in which C function we are at (without the debugging
1287information we might see only the numerical addresses of the functions,
1288which is not very helpful).
1289
a422fd2d
SC
1290F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1291enables all the internal debugging code in Perl. There are a whole bunch
1292of things you can debug with this: L<perlrun> lists them all, and the
1293best way to find out about them is to play about with them. The most
1294useful options are probably
1295
1296 l Context (loop) stack processing
1297 t Trace execution
1298 o Method and overloading resolution
1299 c String/numeric conversions
1300
1301Some of the functionality of the debugging code can be achieved using XS
1302modules.
13a2d996 1303
a422fd2d
SC
1304 -Dr => use re 'debug'
1305 -Dx => use O 'Debug'
1306
1307=head2 Using a source-level debugger
1308
1309If the debugging output of C<-D> doesn't help you, it's time to step
1310through perl's execution with a source-level debugger.
1311
1312=over 3
1313
1314=item *
1315
955fec6b
JH
1316We'll use C<gdb> for our examples here; the principles will apply to
1317any debugger (many vendors call their debugger C<dbx>), but check the
1318manual of the one you're using.
a422fd2d
SC
1319
1320=back
1321
1322To fire up the debugger, type
1323
1324 gdb ./perl
1325
955fec6b
JH
1326Or if you have a core dump:
1327
1328 gdb ./perl core
1329
a422fd2d
SC
1330You'll want to do that in your Perl source tree so the debugger can read
1331the source code. You should see the copyright message, followed by the
1332prompt.
1333
1334 (gdb)
1335
1336C<help> will get you into the documentation, but here are the most
1337useful commands:
1338
1339=over 3
1340
1341=item run [args]
1342
1343Run the program with the given arguments.
1344
1345=item break function_name
1346
1347=item break source.c:xxx
1348
1349Tells the debugger that we'll want to pause execution when we reach
cea6626f 1350either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d
SC
1351line in the named source file.
1352
1353=item step
1354
1355Steps through the program a line at a time.
1356
1357=item next
1358
1359Steps through the program a line at a time, without descending into
1360functions.
1361
1362=item continue
1363
1364Run until the next breakpoint.
1365
1366=item finish
1367
1368Run until the end of the current function, then stop again.
1369
13a2d996 1370=item 'enter'
a422fd2d
SC
1371
1372Just pressing Enter will do the most recent operation again - it's a
1373blessing when stepping through miles of source code.
1374
1375=item print
1376
1377Execute the given C code and print its results. B<WARNING>: Perl makes
52d59bef
JH
1378heavy use of macros, and F<gdb> does not necessarily support macros
1379(see later L</"gdb macro support">). You'll have to substitute them
1380yourself, or to invoke cpp on the source code files
1381(see L</"The .i Targets">)
1382So, for instance, you can't say
a422fd2d
SC
1383
1384 print SvPV_nolen(sv)
1385
1386but you have to say
1387
1388 print Perl_sv_2pv_nolen(sv)
1389
ffc145e8
RK
1390=back
1391
a422fd2d
SC
1392You may find it helpful to have a "macro dictionary", which you can
1393produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
07aa3531 1394recursively apply those macros for you.
52d59bef
JH
1395
1396=head2 gdb macro support
a422fd2d 1397
52d59bef 1398Recent versions of F<gdb> have fairly good macro support, but
ea031e66
RGS
1399in order to use it you'll need to compile perl with macro definitions
1400included in the debugging information. Using F<gcc> version 3.1, this
1401means configuring with C<-Doptimize=-g3>. Other compilers might use a
1402different switch (if they support debugging macros at all).
1403
a422fd2d
SC
1404=head2 Dumping Perl Data Structures
1405
1406One way to get around this macro hell is to use the dumping functions in
1407F<dump.c>; these work a little like an internal
1408L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1409that you can't get at from Perl. Let's take an example. We'll use the
07aa3531 1410C<$a = $b + $c> we used before, but give it a bit of context:
a422fd2d
SC
1411C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1412
1413What about C<pp_add>, the function we examined earlier to implement the
1414C<+> operator:
1415
1416 (gdb) break Perl_pp_add
1417 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1418
cea6626f 1419Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d
SC
1420With the breakpoint in place, we can run our program:
1421
1422 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1423
1424Lots of junk will go past as gdb reads in the relevant source files and
1425libraries, and then:
1426
1427 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1428 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1429 (gdb) step
1430 311 dPOPTOPnnrl_ul;
1431 (gdb)
1432
1433We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1434arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1435slightly expand it:
1436
1437 #define dPOPTOPnnrl_ul NV right = POPn; \
1438 SV *leftsv = TOPs; \
1439 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1440
1441C<POPn> takes the SV from the top of the stack and obtains its NV either
1442directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1443C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1444C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
07aa3531 1445C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
a422fd2d
SC
1446
1447Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1448convert it. If we step again, we'll find ourselves there:
1449
1450 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1451 1669 if (!sv)
1452 (gdb)
1453
1454We can now use C<Perl_sv_dump> to investigate the SV:
1455
1456 SV = PV(0xa057cc0) at 0xa0675d0
1457 REFCNT = 1
1458 FLAGS = (POK,pPOK)
1459 PV = 0xa06a510 "6XXXX"\0
1460 CUR = 5
1461 LEN = 6
1462 $1 = void
1463
1464We know we're going to get C<6> from this, so let's finish the
1465subroutine:
1466
1467 (gdb) finish
1468 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1469 0x462669 in Perl_pp_add () at pp_hot.c:311
1470 311 dPOPTOPnnrl_ul;
1471
1472We can also dump out this op: the current op is always stored in
1473C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1474similar output to L<B::Debug|B::Debug>.
1475
1476 {
1477 13 TYPE = add ===> 14
1478 TARG = 1
1479 FLAGS = (SCALAR,KIDS)
1480 {
1481 TYPE = null ===> (12)
1482 (was rv2sv)
1483 FLAGS = (SCALAR,KIDS)
1484 {
1485 11 TYPE = gvsv ===> 12
1486 FLAGS = (SCALAR)
1487 GV = main::b
1488 }
1489 }
1490
10f58044 1491# finish this later #
a422fd2d
SC
1492
1493=head2 Patching
1494
1495All right, we've now had a look at how to navigate the Perl sources and
1496some things you'll need to know when fiddling with them. Let's now get
1497on and create a simple patch. Here's something Larry suggested: if a
07aa3531 1498C<U> is the first active format during a C<pack>, (for example,
a422fd2d 1499C<pack "U3C8", @stuff>) then the resulting string should be treated as
1e54db1a 1500UTF-8 encoded.
a422fd2d
SC
1501
1502How do we prepare to fix this up? First we locate the code in question -
1503the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1504files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1505altering this file, let's copy it to F<pp.c~>.
1506
a6ec74c1
JH
1507[Well, it was in F<pp.c> when this tutorial was written. It has now been
1508split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1509
a422fd2d
SC
1510Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1511loop over the pattern, taking each format character in turn into
1512C<datum_type>. Then for each possible format character, we swallow up
1513the other arguments in the pattern (a field width, an asterisk, and so
1514on) and convert the next chunk input into the specified format, adding
1515it onto the output SV C<cat>.
1516
1517How do we know if the C<U> is the first format in the C<pat>? Well, if
1518we have a pointer to the start of C<pat> then, if we see a C<U> we can
1519test whether we're still at the start of the string. So, here's where
1520C<pat> is set up:
1521
1522 STRLEN fromlen;
1523 register char *pat = SvPVx(*++MARK, fromlen);
1524 register char *patend = pat + fromlen;
1525 register I32 len;
1526 I32 datumtype;
1527 SV *fromstr;
1528
1529We'll have another string pointer in there:
1530
1531 STRLEN fromlen;
1532 register char *pat = SvPVx(*++MARK, fromlen);
1533 register char *patend = pat + fromlen;
1534 + char *patcopy;
1535 register I32 len;
1536 I32 datumtype;
1537 SV *fromstr;
1538
1539And just before we start the loop, we'll set C<patcopy> to be the start
1540of C<pat>:
1541
1542 items = SP - MARK;
1543 MARK++;
1544 sv_setpvn(cat, "", 0);
1545 + patcopy = pat;
1546 while (pat < patend) {
1547
1548Now if we see a C<U> which was at the start of the string, we turn on
1e54db1a 1549the C<UTF8> flag for the output SV, C<cat>:
a422fd2d
SC
1550
1551 + if (datumtype == 'U' && pat==patcopy+1)
1552 + SvUTF8_on(cat);
1553 if (datumtype == '#') {
1554 while (pat < patend && *pat != '\n')
1555 pat++;
1556
1557Remember that it has to be C<patcopy+1> because the first character of
1558the string is the C<U> which has been swallowed into C<datumtype!>
1559
1560Oops, we forgot one thing: what if there are spaces at the start of the
1561pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1562character, even though it's not the first thing in the pattern. In this
1563case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1564
1565 if (isSPACE(datumtype))
1566 continue;
1567
1568needs to become
1569
1570 if (isSPACE(datumtype)) {
1571 patcopy++;
1572 continue;
1573 }
1574
1575OK. That's the C part done. Now we must do two additional things before
1576this patch is ready to go: we've changed the behaviour of Perl, and so
1577we must document that change. We must also provide some more regression
1578tests to make sure our patch works and doesn't create a bug somewhere
1579else along the line.
1580
b23b8711
MS
1581The regression tests for each operator live in F<t/op/>, and so we
1582make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1583tests to the end. First, we'll test that the C<U> does indeed create
07aa3531 1584Unicode strings.
b23b8711
MS
1585
1586t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1587use the one from t/test.pl.
b23b8711 1588
35c336e6
MS
1589 require './test.pl';
1590 plan( tests => 159 );
b23b8711
MS
1591
1592so instead of this:
a422fd2d
SC
1593
1594 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1595 print "ok $test\n"; $test++;
1596
35c336e6
MS
1597we can write the more sensible (see L<Test::More> for a full
1598explanation of is() and other testing functions).
b23b8711 1599
07aa3531 1600 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
38a44b82 1601 "U* produces Unicode" );
b23b8711 1602
a422fd2d
SC
1603Now we'll test that we got that space-at-the-beginning business right:
1604
35c336e6 1605 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1606 " with spaces at the beginning" );
a422fd2d
SC
1607
1608And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1609the first active format:
1610
35c336e6 1611 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
38a44b82 1612 "U* not first isn't Unicode" );
a422fd2d 1613
35c336e6
MS
1614Mustn't forget to change the number of tests which appears at the top,
1615or else the automated tester will get confused. This will either look
1616like this:
a422fd2d 1617
35c336e6
MS
1618 print "1..156\n";
1619
1620or this:
1621
1622 plan( tests => 156 );
a422fd2d
SC
1623
1624We now compile up Perl, and run it through the test suite. Our new
1625tests pass, hooray!
1626
1627Finally, the documentation. The job is never done until the paperwork is
1628over, so let's describe the change we've just made. The relevant place
1629is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1630this text in the description of C<pack>:
1631
1632 =item *
1633
1634 If the pattern begins with a C<U>, the resulting string will be treated
1e54db1a
JH
1635 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1636 with an initial C<U0>, and the bytes that follow will be interpreted as
1637 Unicode characters. If you don't want this to happen, you can begin your
1638 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
a422fd2d
SC
1639 string, and then follow this with a C<U*> somewhere in your pattern.
1640
1641All done. Now let's create the patch. F<Porting/patching.pod> tells us
1642that if we're making major changes, we should copy the entire directory
1643to somewhere safe before we begin fiddling, and then do
13a2d996 1644
a422fd2d
SC
1645 diff -ruN old new > patch
1646
1647However, we know which files we've changed, and we can simply do this:
1648
1649 diff -u pp.c~ pp.c > patch
1650 diff -u t/op/pack.t~ t/op/pack.t >> patch
1651 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1652
1653We end up with a patch looking a little like this:
1654
1655 --- pp.c~ Fri Jun 02 04:34:10 2000
1656 +++ pp.c Fri Jun 16 11:37:25 2000
1657 @@ -4375,6 +4375,7 @@
1658 register I32 items;
1659 STRLEN fromlen;
1660 register char *pat = SvPVx(*++MARK, fromlen);
1661 + char *patcopy;
1662 register char *patend = pat + fromlen;
1663 register I32 len;
1664 I32 datumtype;
1665 @@ -4405,6 +4406,7 @@
1666 ...
1667
1668And finally, we submit it, with our rationale, to perl5-porters. Job
1669done!
1670
f7e1e956
MS
1671=head2 Patching a core module
1672
1673This works just like patching anything else, with an extra
1674consideration. Many core modules also live on CPAN. If this is so,
1675patch the CPAN version instead of the core and send the patch off to
1676the module maintainer (with a copy to p5p). This will help the module
1677maintainer keep the CPAN version in sync with the core version without
1678constantly scanning p5p.
1679
db300100
RGS
1680The list of maintainers of core modules is usefully documented in
1681F<Porting/Maintainers.pl>.
1682
acbe17fc
JP
1683=head2 Adding a new function to the core
1684
1685If, as part of a patch to fix a bug, or just because you have an
1686especially good idea, you decide to add a new function to the core,
1687discuss your ideas on p5p well before you start work. It may be that
1688someone else has already attempted to do what you are considering and
1689can give lots of good advice or even provide you with bits of code
1690that they already started (but never finished).
1691
1692You have to follow all of the advice given above for patching. It is
1693extremely important to test any addition thoroughly and add new tests
1694to explore all boundary conditions that your new function is expected
1695to handle. If your new function is used only by one module (e.g. toke),
1696then it should probably be named S_your_function (for static); on the
210b36aa 1697other hand, if you expect it to accessible from other functions in
acbe17fc
JP
1698Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1699for more details.
1700
1701The location of any new code is also an important consideration. Don't
1702just create a new top level .c file and put your code there; you would
1703have to make changes to Configure (so the Makefile is created properly),
1704as well as possibly lots of include files. This is strictly pumpking
1705business.
1706
1707It is better to add your function to one of the existing top level
1708source code files, but your choice is complicated by the nature of
1709the Perl distribution. Only the files that are marked as compiled
1710static are located in the perl executable. Everything else is located
1711in the shared library (or DLL if you are running under WIN32). So,
1712for example, if a function was only used by functions located in
1713toke.c, then your code can go in toke.c. If, however, you want to call
1714the function from universal.c, then you should put your code in another
1715location, for example util.c.
1716
1717In addition to writing your c-code, you will need to create an
1718appropriate entry in embed.pl describing your function, then run
1719'make regen_headers' to create the entries in the numerous header
1720files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1721for information on the various options that you can set in embed.pl.
1722You will forget to do this a few (or many) times and you will get
1723warnings during the compilation phase. Make sure that you mention
1724this when you post your patch to P5P; the pumpking needs to know this.
1725
1726When you write your new code, please be conscious of existing code
884bad00 1727conventions used in the perl source files. See L<perlstyle> for
acbe17fc
JP
1728details. Although most of the guidelines discussed seem to focus on
1729Perl code, rather than c, they all apply (except when they don't ;).
1730See also I<Porting/patching.pod> file in the Perl source distribution
1731for lots of details about both formatting and submitting patches of
1732your changes.
1733
1734Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1735Test on as many platforms as you can find. Test as many perl
1736Configure options as you can (e.g. MULTIPLICITY). If you have
1737profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 1738below for how to use them to further test your code. Remember that
acbe17fc
JP
1739most of the people on P5P are doing this on their own time and
1740don't have the time to debug your code.
f7e1e956
MS
1741
1742=head2 Writing a test
1743
1744Every module and built-in function has an associated test file (or
1745should...). If you add or change functionality, you have to write a
1746test. If you fix a bug, you have to write a test so that bug never
1747comes back. If you alter the docs, it would be nice to test what the
1748new documentation says.
1749
1750In short, if you submit a patch you probably also have to patch the
1751tests.
1752
1753For modules, the test file is right next to the module itself.
1754F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1755so there are some snags (and it would be wonderful for you to brush
1756them out), but it basically works that way. Everything else lives in
1757F<t/>.
1758
d5f28025
JV
1759If you add a new test directory under F<t/>, it is imperative that you
1760add that directory to F<t/HARNESS> and F<t/TEST>.
1761
f7e1e956
MS
1762=over 3
1763
1764=item F<t/base/>
1765
1766Testing of the absolute basic functionality of Perl. Things like
1767C<if>, basic file reads and writes, simple regexes, etc. These are
1768run first in the test suite and if any of them fail, something is
1769I<really> broken.
1770
1771=item F<t/cmd/>
1772
1773These test the basic control structures, C<if/else>, C<while>,
35c336e6 1774subroutines, etc.
f7e1e956
MS
1775
1776=item F<t/comp/>
1777
1778Tests basic issues of how Perl parses and compiles itself.
1779
1780=item F<t/io/>
1781
1782Tests for built-in IO functions, including command line arguments.
1783
1784=item F<t/lib/>
1785
1786The old home for the module tests, you shouldn't put anything new in
1787here. There are still some bits and pieces hanging around in here
1788that need to be moved. Perhaps you could move them? Thanks!
1789
3c295041
RGS
1790=item F<t/mro/>
1791
0503309d 1792Tests for perl's method resolution order implementations
3c295041
RGS
1793(see L<mro>).
1794
f7e1e956
MS
1795=item F<t/op/>
1796
1797Tests for perl's built in functions that don't fit into any of the
1798other directories.
1799
1800=item F<t/pod/>
1801
1802Tests for POD directives. There are still some tests for the Pod
1803modules hanging around in here that need to be moved out into F<lib/>.
1804
a4499558
YO
1805=item F<t/re/>
1806
1807Tests for regex related functions or behaviour. (These used to live
1808in t/op).
1809
f7e1e956
MS
1810=item F<t/run/>
1811
1812Testing features of how perl actually runs, including exit codes and
1813handling of PERL* environment variables.
1814
244d9cb7
RGS
1815=item F<t/uni/>
1816
1817Tests for the core support of Unicode.
1818
1819=item F<t/win32/>
1820
1821Windows-specific tests.
1822
1823=item F<t/x2p>
1824
1825A test suite for the s2p converter.
1826
f7e1e956
MS
1827=back
1828
1829The core uses the same testing style as the rest of Perl, a simple
1830"ok/not ok" run through Test::Harness, but there are a few special
1831considerations.
1832
35c336e6
MS
1833There are three ways to write a test in the core. Test::More,
1834t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1835decision of which to use depends on what part of the test suite you're
1836working on. This is a measure to prevent a high-level failure (such
1837as Config.pm breaking) from causing basic functionality tests to fail.
1838
07aa3531 1839=over 4
35c336e6
MS
1840
1841=item t/base t/comp
1842
1843Since we don't know if require works, or even subroutines, use ad hoc
1844tests for these two. Step carefully to avoid using the feature being
1845tested.
1846
1847=item t/cmd t/run t/io t/op
1848
1849Now that basic require() and subroutines are tested, you can use the
1850t/test.pl library which emulates the important features of Test::More
1851while using a minimum of core features.
1852
1853You can also conditionally use certain libraries like Config, but be
1854sure to skip the test gracefully if it's not there.
1855
1856=item t/lib ext lib
1857
1858Now that the core of Perl is tested, Test::More can be used. You can
1859also use the full suite of core modules in the tests.
1860
1861=back
f7e1e956
MS
1862
1863When you say "make test" Perl uses the F<t/TEST> program to run the
07aa3531
JC
1864test suite (except under Win32 where it uses F<t/harness> instead.)
1865All tests are run from the F<t/> directory, B<not> the directory
1866which contains the test. This causes some problems with the tests
7205a85d 1867in F<lib/>, so here's some opportunity for some patching.
f7e1e956
MS
1868
1869You must be triply conscious of cross-platform concerns. This usually
1870boils down to using File::Spec and avoiding things like C<fork()> and
1871C<system()> unless absolutely necessary.
1872
e018f8be
JH
1873=head2 Special Make Test Targets
1874
1875There are various special make targets that can be used to test Perl
1876slightly differently than the standard "test" target. Not all them
1877are expected to give a 100% success rate. Many of them have several
7205a85d
YO
1878aliases, and many of them are not available on certain operating
1879systems.
e018f8be
JH
1880
1881=over 4
1882
1883=item coretest
1884
7d7d5695 1885Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
e018f8be 1886
7205a85d
YO
1887(Not available on Win32)
1888
e018f8be
JH
1889=item test.deparse
1890
b26492ee
RGS
1891Run all the tests through B::Deparse. Not all tests will succeed.
1892
7205a85d
YO
1893(Not available on Win32)
1894
b26492ee
RGS
1895=item test.taintwarn
1896
1897Run all tests with the B<-t> command-line switch. Not all tests
1898are expected to succeed (until they're specifically fixed, of course).
e018f8be 1899
7205a85d
YO
1900(Not available on Win32)
1901
e018f8be
JH
1902=item minitest
1903
1904Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
8cebccf4 1905F<t/op>, F<t/uni> and F<t/mro> tests.
e018f8be 1906
7a834142
JH
1907=item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1908
1909(Only in Linux) Run all the tests using the memory leak + naughty
1910memory access tool "valgrind". The log files will be named
1911F<testname.valgrind>.
1912
e018f8be
JH
1913=item test.third check.third utest.third ucheck.third
1914
1915(Only in Tru64) Run all the tests using the memory leak + naughty
1916memory access tool "Third Degree". The log files will be named
60a57c1c 1917F<perl.3log.testname>.
e018f8be
JH
1918
1919=item test.torture torturetest
1920
1921Run all the usual tests and some extra tests. As of Perl 5.8.0 the
244d9cb7 1922only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
e018f8be
JH
1923
1924You can also run the torture test with F<t/harness> by giving
1925C<-torture> argument to F<t/harness>.
1926
1927=item utest ucheck test.utf8 check.utf8
1928
1929Run all the tests with -Mutf8. Not all tests will succeed.
1930
7205a85d
YO
1931(Not available on Win32)
1932
cc0710ff
RGS
1933=item minitest.utf16 test.utf16
1934
1935Runs the tests with UTF-16 encoded scripts, encoded with different
1936versions of this encoding.
1937
1938C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
1939C<-utf16> arguments to F<t/TEST>.
1940
7205a85d
YO
1941(Not available on Win32)
1942
244d9cb7
RGS
1943=item test_harness
1944
1945Run the test suite with the F<t/harness> controlling program, instead of
1946F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1947L<Test::Harness> module, thus using this test target supposes that perl
1948mostly works. The main advantage for our purposes is that it prints a
00bf5cd9
RGS
1949detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1950doesn't redirect stderr to stdout.
244d9cb7 1951
7205a85d
YO
1952Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
1953there is no special "test_harness" target.
1954
1955Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
1956environment variables to control the behaviour of F<t/harness>. This means
1957you can say
1958
1959 nmake test TEST_FILES="op/*.t"
1960 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
1961
a75f557c
JV
1962=item Parallel tests
1963
1964The core distribution can now run its regression tests in parallel on
1965Unix-like platforms. Instead of running C<make test>, set C<TEST_JOBS> in
1966your environment to the number of tests to run in parallel, and run
1967C<make test_harness>. On a Bourne-like shell, this can be done as
1968
1969 TEST_JOBS=3 make test_harness # Run 3 tests in parallel
1970
1971An environment variable is used, rather than parallel make itself, because
1972L<TAP::Harness> needs to be able to schedule individual non-conflicting test
1973scripts itself, and there is no standard interface to C<make> utilities to
1974interact with their job schedulers.
1975
1976Note that currently some test scripts may fail when run in parallel (most
1977notably C<ext/IO/t/io_dir.t>). If necessary run just the failing scripts
1978again sequentially and see if the failures go away.
7205a85d
YO
1979=item test-notty test_notty
1980
1981Sets PERL_SKIP_TTY_TEST to true before running normal test.
1982
244d9cb7
RGS
1983=back
1984
1985=head2 Running tests by hand
1986
1987You can run part of the test suite by hand by using one the following
1988commands from the F<t/> directory :
1989
1990 ./perl -I../lib TEST list-of-.t-files
1991
1992or
1993
1994 ./perl -I../lib harness list-of-.t-files
1995
1996(if you don't specify test scripts, the whole test suite will be run.)
1997
7205a85d
YO
1998=head3 Using t/harness for testing
1999
2000If you use C<harness> for testing you have several command line options
2001available to you. The arguments are as follows, and are in the order
2002that they must appear if used together.
2003
2004 harness -v -torture -re=pattern LIST OF FILES TO TEST
2005 harness -v -torture -re LIST OF PATTERNS TO MATCH
2006
2007If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
07aa3531 2008the manifest. The file list may include shell wildcards which will be
7205a85d
YO
2009expanded out.
2010
2011=over 4
2012
2013=item -v
2014
07aa3531 2015Run the tests under verbose mode so you can see what tests were run,
8550bf48 2016and debug output.
7205a85d
YO
2017
2018=item -torture
2019
2020Run the torture tests as well as the normal set.
2021
2022=item -re=PATTERN
2023
2024Filter the file list so that all the test files run match PATTERN.
2025Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2026in that it allows the file list to be provided as well.
2027
2028=item -re LIST OF PATTERNS
2029
07aa3531 2030Filter the file list so that all the test files run match
7205a85d
YO
2031/(LIST|OF|PATTERNS)/. Note that with this form the patterns
2032are joined by '|' and you cannot supply a list of files, instead
2033the test files are obtained from the MANIFEST.
2034
2035=back
2036
244d9cb7
RGS
2037You can run an individual test by a command similar to
2038
2039 ./perl -I../lib patho/to/foo.t
2040
2041except that the harnesses set up some environment variables that may
2042affect the execution of the test :
2043
07aa3531 2044=over 4
244d9cb7
RGS
2045
2046=item PERL_CORE=1
2047
2048indicates that we're running this test part of the perl core test suite.
2049This is useful for modules that have a dual life on CPAN.
2050
2051=item PERL_DESTRUCT_LEVEL=2
2052
2053is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2054
2055=item PERL
2056
2057(used only by F<t/TEST>) if set, overrides the path to the perl executable
2058that should be used to run the tests (the default being F<./perl>).
2059
2060=item PERL_SKIP_TTY_TEST
2061
2062if set, tells to skip the tests that need a terminal. It's actually set
2063automatically by the Makefile, but can also be forced artificially by
2064running 'make test_notty'.
2065
e018f8be 2066=back
f7e1e956 2067
7cd58830
RGS
2068=head3 Other environment variables that may influence tests
2069
2070=over 4
2071
2072=item PERL_TEST_Net_Ping
2073
2074Setting this variable runs all the Net::Ping modules tests,
2075otherwise some tests that interact with the outside world are skipped.
2076See L<perl58delta>.
2077
2078=item PERL_TEST_NOVREXX
2079
2080Setting this variable skips the vrexx.t tests for OS2::REXX.
2081
2082=item PERL_TEST_NUMCONVERTS
2083
2084This sets a variable in op/numconvert.t.
2085
2086=back
2087
2088See also the documentation for the Test and Test::Harness modules,
2089for more environment variables that affect testing.
2090
d7889f52
JH
2091=head2 Common problems when patching Perl source code
2092
2093Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2094some cases we have to take pre-ANSI requirements into consideration.
2095You don't care about some particular platform having broken Perl?
2096I hear there is still a strong demand for J2EE programmers.
2097
2098=head2 Perl environment problems
2099
2100=over 4
2101
2102=item *
2103
2104Not compiling with threading
2105
2106Compiling with threading (-Duseithreads) completely rewrites
2107the function prototypes of Perl. You better try your changes
0bec6c03 2108with that. Related to this is the difference between "Perl_-less"
d7889f52
JH
2109and "Perl_-ly" APIs, for example:
2110
2111 Perl_sv_setiv(aTHX_ ...);
2112 sv_setiv(...);
2113
ee9468a2
RGS
2114The first one explicitly passes in the context, which is needed for e.g.
2115threaded builds. The second one does that implicitly; do not get them
def4ed7d
JH
2116mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2117(or a dVAR) as the first thing in the function.
d7889f52
JH
2118
2119See L<perlguts/"How multiple interpreters and concurrency are supported">
2120for further discussion about context.
2121
2122=item *
2123
2124Not compiling with -DDEBUGGING
2125
2126The DEBUGGING define exposes more code to the compiler,
0bec6c03 2127therefore more ways for things to go wrong. You should try it.
d7889f52
JH
2128
2129=item *
2130
ee9468a2
RGS
2131Introducing (non-read-only) globals
2132
2133Do not introduce any modifiable globals, truly global or file static.
bc028b6b
JH
2134They are bad form and complicate multithreading and other forms of
2135concurrency. The right way is to introduce them as new interpreter
2136variables, see F<intrpvar.h> (at the very end for binary compatibility).
ee9468a2
RGS
2137
2138Introducing read-only (const) globals is okay, as long as you verify
2139with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2140BSD-style output) that the data you added really is read-only.
2141(If it is, it shouldn't show up in the output of that command.)
2142
2143If you want to have static strings, make them constant:
2144
2145 static const char etc[] = "...";
2146
bc028b6b 2147If you want to have arrays of constant strings, note carefully
ee9468a2
RGS
2148the right combination of C<const>s:
2149
2150 static const char * const yippee[] =
2151 {"hi", "ho", "silver"};
2152
bc028b6b
JH
2153There is a way to completely hide any modifiable globals (they are all
2154moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2155It is not normally used, but can be used for testing, read more
def4ed7d 2156about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
bc028b6b 2157
ee9468a2
RGS
2158=item *
2159
d7889f52
JH
2160Not exporting your new function
2161
2162Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2163function that is part of the public API (the shared Perl library)
2164to be explicitly marked as exported. See the discussion about
2165F<embed.pl> in L<perlguts>.
2166
2167=item *
2168
2169Exporting your new function
2170
2171The new shiny result of either genuine new functionality or your
2172arduous refactoring is now ready and correctly exported. So what
def4ed7d 2173could possibly go wrong?
d7889f52
JH
2174
2175Maybe simply that your function did not need to be exported in the
2176first place. Perl has a long and not so glorious history of exporting
2177functions that it should not have.
2178
2179If the function is used only inside one source code file, make it
2180static. See the discussion about F<embed.pl> in L<perlguts>.
2181
2182If the function is used across several files, but intended only for
2183Perl's internal use (and this should be the common case), do not
2184export it to the public API. See the discussion about F<embed.pl>
2185in L<perlguts>.
2186
2187=back
2188
5b38b9cd 2189=head2 Portability problems
d7889f52
JH
2190
2191The following are common causes of compilation and/or execution
2192failures, not common to Perl as such. The C FAQ is good bedtime
0bec6c03
JH
2193reading. Please test your changes with as many C compilers and
2194platforms as possible -- we will, anyway, and it's nice to save
2195oneself from public embarrassment.
2196
9aaf14db
RGS
2197If using gcc, you can add the C<-std=c89> option which will hopefully
2198catch most of these unportabilities. (However it might also catch
2199incompatibilities in your system's header files.)
d1307786 2200
a8e98a71
JH
2201Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2202C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2203
def4ed7d
JH
2204If using the C<gcc -Wall> note that not all the possible warnings
2205(like C<-Wunitialized>) are given unless you also compile with C<-O>.
2206
2207Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2208code files (the ones at the top level of the source code distribution,
2209but not e.g. the extensions under ext/) are automatically compiled
2210with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2211and a selection of C<-W> flags (see cflags.SH).
27565cb6 2212
0bec6c03 2213Also study L<perlport> carefully to avoid any bad assumptions
def4ed7d 2214about the operating system, filesystems, and so forth.
0bec6c03 2215
606fd33d 2216You may once in a while try a "make microperl" to see whether we
63796a85 2217can still compile Perl with just the bare minimum of interfaces.
606fd33d 2218(See README.micro.)
ee9468a2 2219
0bec6c03 2220Do not assume an operating system indicates a certain compiler.
d7889f52
JH
2221
2222=over 4
2223
2224=item *
2225
2226Casting pointers to integers or casting integers to pointers
2227
2228 void castaway(U8* p)
2229 {
2230 IV i = p;
2231
2232or
2233
2234 void castaway(U8* p)
2235 {
2236 IV i = (IV)p;
2237
ee9468a2 2238Both are bad, and broken, and unportable. Use the PTR2IV()
d7889f52
JH
2239macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2240INT2PTR(), and NUM2PTR().)
2241
2242=item *
2243
0bec6c03
JH
2244Casting between data function pointers and data pointers
2245
d7889f52
JH
2246Technically speaking casting between function pointers and data
2247pointers is unportable and undefined, but practically speaking
2248it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
0bec6c03 2249macros. Sometimes you can also play games with unions.
d7889f52
JH
2250
2251=item *
2252
2253Assuming sizeof(int) == sizeof(long)
2254
2255There are platforms where longs are 64 bits, and platforms where ints
2256are 64 bits, and while we are out to shock you, even platforms where
2257shorts are 64 bits. This is all legal according to the C standard.
2258(In other words, "long long" is not a portable way to specify 64 bits,
2259and "long long" is not even guaranteed to be any wider than "long".)
63796a85
JH
2260
2261Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2262Avoid things like I32 because they are B<not> guaranteed to be
2263I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2264guaranteed to be B<int> or B<long>. If you really explicitly need
226564-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
d7889f52
JH
2266
2267=item *
2268
2269Assuming one can dereference any type of pointer for any type of data
2270
2271 char *p = ...;
def4ed7d 2272 long pony = *p; /* BAD */
d7889f52
JH
2273
2274Many platforms, quite rightly so, will give you a core dump instead
2275of a pony if the p happens not be correctly aligned.
2276
2277=item *
2278
2279Lvalue casts
2280
def4ed7d 2281 (int)*p = ...; /* BAD */
d7889f52
JH
2282
2283Simply not portable. Get your lvalue to be of the right type,
27565cb6
JH
2284or maybe use temporary variables, or dirty tricks with unions.
2285
2286=item *
2287
606fd33d
JH
2288Assume B<anything> about structs (especially the ones you
2289don't control, like the ones coming from the system headers)
27565cb6
JH
2290
2291=over 8
2292
2293=item *
2294
2295That a certain field exists in a struct
2296
2297=item *
2298
902821cc 2299That no other fields exist besides the ones you know of
27565cb6
JH
2300
2301=item *
2302
606fd33d 2303That a field is of certain signedness, sizeof, or type
27565cb6
JH
2304
2305=item *
2306
2307That the fields are in a certain order
2308
606fd33d
JH
2309=over 8
2310
27565cb6
JH
2311=item *
2312
606fd33d
JH
2313While C guarantees the ordering specified in the struct definition,
2314between different platforms the definitions might differ
2315
2316=back
27565cb6
JH
2317
2318=item *
2319
606fd33d
JH
2320That the sizeof(struct) or the alignments are the same everywhere
2321
2322=over 8
27565cb6
JH
2323
2324=item *
2325
606fd33d
JH
2326There might be padding bytes between the fields to align the fields -
2327the bytes can be anything
2328
2329=item *
2330
2331Structs are required to be aligned to the maximum alignment required
2332by the fields - which for native types is for usually equivalent to
2333sizeof() of the field
2334
2335=back
27565cb6
JH
2336
2337=back
d7889f52
JH
2338
2339=item *
2340
2bbc8d55
SP
2341Assuming the character set is ASCIIish
2342
2343Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
2344This is transparent for the most part, but because the character sets
2345differ, you shouldn't use numeric (decimal, octal, nor hex) constants
2346to refer to characters. You can safely say 'A', but not 0x41.
2347You can safely say '\n', but not \012.
2348If a character doesn't have a trivial input form, you can
2349create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that
2350it resolves to different values depending on the character set being used.
2351(There are three different EBCDIC character sets defined in C<utfebcdic.h>,
2352so it might be best to insert the #define three times in that file.)
2353
2354Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case
2355alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'.
2356But '0' - '9' is an unbroken range in both systems. Don't assume anything
2357about other ranges.
2358
2359Many of the comments in the existing code ignore the possibility of EBCDIC,
2360and may be wrong therefore, even if the code works.
2361This is actually a tribute to the successful transparent insertion of being
fe749c9a 2362able to handle EBCDIC without having to change pre-existing code.
2bbc8d55
SP
2363
2364UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode
2365code points as sequences of bytes. Macros
2366with the same names (but different definitions)
2367in C<utf8.h> and C<utfebcdic.h>
fe749c9a
KW
2368are used to allow the calling code to think that there is only one such
2369encoding.
2370This is almost always referred to as C<utf8>, but it means the EBCDIC version
2371as well. Again, comments in the code may well be wrong even if the code itself
2372is right.
2bbc8d55
SP
2373For example, the concept of C<invariant characters> differs between ASCII and
2374EBCDIC.
2375On ASCII platforms, only characters that do not have the high-order
2376bit set (i.e. whose ordinals are strict ASCII, 0 - 127)
2377are invariant, and the documentation and comments in the code
2378may assume that,
2379often referring to something like, say, C<hibit>.
2380The situation differs and is not so simple on EBCDIC machines, but as long as
2381the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it
2382works, even if the comments are wrong.
2383
2384=item *
2385
2386Assuming the character set is just ASCII
2387
2388ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
2389characters have different meanings depending on the locale. Absent a locale,
2390currently these extra characters are generally considered to be unassigned,
2391and this has presented some problems.
2392This is scheduled to be changed in 5.12 so that these characters will
2393be considered to be Latin-1 (ISO-8859-1).
2394
2395=item *
2396
0bec6c03
JH
2397Mixing #define and #ifdef
2398
2399 #define BURGLE(x) ... \
def4ed7d 2400 #ifdef BURGLE_OLD_STYLE /* BAD */
0bec6c03
JH
2401 ... do it the old way ... \
2402 #else
2403 ... do it the new way ... \
2404 #endif
2405
ee9468a2
RGS
2406You cannot portably "stack" cpp directives. For example in the above
2407you need two separate BURGLE() #defines, one for each #ifdef branch.
2408
2409=item *
2410
2bbc8d55 2411Adding non-comment stuff after #endif or #else
ee9468a2
RGS
2412
2413 #ifdef SNOSH
2414 ...
def4ed7d 2415 #else !SNOSH /* BAD */
ee9468a2 2416 ...
def4ed7d 2417 #endif SNOSH /* BAD */
ee9468a2 2418
def4ed7d
JH
2419The #endif and #else cannot portably have anything non-comment after
2420them. If you want to document what is going (which is a good idea
2421especially if the branches are long), use (C) comments:
ee9468a2
RGS
2422
2423 #ifdef SNOSH
2424 ...
2425 #else /* !SNOSH */
2426 ...
2427 #endif /* SNOSH */
2428
2429The gcc option C<-Wendif-labels> warns about the bad variant
2430(by default on starting from Perl 5.9.4).
0bec6c03
JH
2431
2432=item *
2433
27565cb6
JH
2434Having a comma after the last element of an enum list
2435
2436 enum color {
2437 CERULEAN,
2438 CHARTREUSE,
def4ed7d 2439 CINNABAR, /* BAD */
27565cb6
JH
2440 };
2441
2442is not portable. Leave out the last comma.
2443
2444Also note that whether enums are implicitly morphable to ints
2445varies between compilers, you might need to (int).
2446
2447=item *
2448
d7889f52
JH
2449Using //-comments
2450
def4ed7d 2451 // This function bamfoodles the zorklator. /* BAD */
d7889f52
JH
2452
2453That is C99 or C++. Perl is C89. Using the //-comments is silently
0bec6c03
JH
2454allowed by many C compilers but cranking up the ANSI C89 strictness
2455(which we like to do) causes the compilation to fail.
d7889f52
JH
2456
2457=item *
2458
2459Mixing declarations and code
2460
2461 void zorklator()
2462 {
2463 int n = 3;
def4ed7d 2464 set_zorkmids(n); /* BAD */
d7889f52
JH
2465 int q = 4;
2466
0bec6c03
JH
2467That is C99 or C++. Some C compilers allow that, but you shouldn't.
2468
63796a85
JH
2469The gcc option C<-Wdeclaration-after-statements> scans for such problems
2470(by default on starting from Perl 5.9.4).
2471
0bec6c03
JH
2472=item *
2473
2474Introducing variables inside for()
2475
def4ed7d 2476 for(int i = ...; ...; ...) { /* BAD */
0bec6c03
JH
2477
2478That is C99 or C++. While it would indeed be awfully nice to have that
2479also in C89, to limit the scope of the loop variable, alas, we cannot.
d7889f52
JH
2480
2481=item *
2482
2483Mixing signed char pointers with unsigned char pointers
2484
2485 int foo(char *s) { ... }
2486 ...
2487 unsigned char *t = ...; /* Or U8* t = ... */
def4ed7d 2488 foo(t); /* BAD */
d7889f52
JH
2489
2490While this is legal practice, it is certainly dubious, and downright
2491fatal in at least one platform: for example VMS cc considers this a
def4ed7d
JH
2492fatal error. One cause for people often making this mistake is that a
2493"naked char" and therefore dereferencing a "naked char pointer" have
2494an undefined signedness: it depends on the compiler and the flags of
2495the compiler and the underlying platform whether the result is signed
2496or unsigned. For this very same reason using a 'char' as an array
2497index is bad.
d7889f52
JH
2498
2499=item *
2500
2501Macros that have string constants and their arguments as substrings of
2502the string constants
2503
def4ed7d 2504 #define FOO(n) printf("number = %d\n", n) /* BAD */
d7889f52
JH
2505 FOO(10);
2506
2507Pre-ANSI semantics for that was equivalent to
2508
2509 printf("10umber = %d\10");
2510
0bec6c03
JH
2511which is probably not what you were expecting. Unfortunately at least
2512one reasonably common and modern C compiler does "real backward
63796a85 2513compatibility" here, in AIX that is what still happens even though the
0bec6c03
JH
2514rest of the AIX compiler is very happily C89.
2515
2516=item *
2517
ee9468a2
RGS
2518Using printf formats for non-basic C types
2519
2520 IV i = ...;
def4ed7d 2521 printf("i = %d\n", i); /* BAD */
ee9468a2
RGS
2522
2523While this might by accident work in some platform (where IV happens
2524to be an C<int>), in general it cannot. IV might be something larger.
2525Even worse the situation is with more specific types (defined by Perl's
2526configuration step in F<config.h>):
2527
2528 Uid_t who = ...;
def4ed7d 2529 printf("who = %d\n", who); /* BAD */
ee9468a2
RGS
2530
2531The problem here is that Uid_t might be not only not C<int>-wide
2532but it might also be unsigned, in which case large uids would be
2533printed as negative values.
2534
2535There is no simple solution to this because of printf()'s limited
2536intelligence, but for many types the right format is available as
2537with either 'f' or '_f' suffix, for example:
2538
2539 IVdf /* IV in decimal */
2540 UVxf /* UV is hexadecimal */
2541
2542 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2543
2544 Uid_t_f /* Uid_t in decimal */
2545
2546 printf("who = %"Uid_t_f"\n", who);
2547
63796a85
JH
2548Or you can try casting to a "wide enough" type:
2549
2550 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2551
2552Also remember that the C<%p> format really does require a void pointer:
2553
2554 U8* p = ...;
2555 printf("p = %p\n", (void*)p);
2556
ee9468a2
RGS
2557The gcc option C<-Wformat> scans for such problems.
2558
2559=item *
2560
0bec6c03
JH
2561Blindly using variadic macros
2562
63796a85
JH
2563gcc has had them for a while with its own syntax, and C99 brought
2564them with a standardized syntax. Don't use the former, and use
2565the latter only if the HAS_C99_VARIADIC_MACROS is defined.
0bec6c03
JH
2566
2567=item *
2568
2569Blindly passing va_list
2570
2571Not all platforms support passing va_list to further varargs (stdarg)
2572functions. The right thing to do is to copy the va_list using the
2573Perl_va_copy() if the NEED_VA_COPY is defined.
d7889f52 2574
ee9468a2
RGS
2575=item *
2576
e5afc1ae 2577Using gcc statement expressions
63796a85 2578
def4ed7d 2579 val = ({...;...;...}); /* BAD */
63796a85 2580
def4ed7d
JH
2581While a nice extension, it's not portable. The Perl code does
2582admittedly use them if available to gain some extra speed
2583(essentially as a funky form of inlining), but you shouldn't.
63796a85
JH
2584
2585=item *
2586
2bbc8d55 2587Binding together several statements in a macro
63796a85
JH
2588
2589Use the macros STMT_START and STMT_END.
2590
2591 STMT_START {
2592 ...
2593 } STMT_END
2594
2595=item *
2596
ee9468a2
RGS
2597Testing for operating systems or versions when should be testing for features
2598
def4ed7d 2599 #ifdef __FOONIX__ /* BAD */
ee9468a2
RGS
2600 foo = quux();
2601 #endif
2602
2603Unless you know with 100% certainty that quux() is only ever available
2604for the "Foonix" operating system B<and> that is available B<and>
2605correctly working for B<all> past, present, B<and> future versions of
2606"Foonix", the above is very wrong. This is more correct (though still
2607not perfect, because the below is a compile-time check):
2608
2609 #ifdef HAS_QUUX
2610 foo = quux();
2611 #endif
2612
def4ed7d 2613How does the HAS_QUUX become defined where it needs to be? Well, if
353c6505 2614Foonix happens to be UNIXy enough to be able to run the Configure
ee9468a2
RGS
2615script, and Configure has been taught about detecting and testing
2616quux(), the HAS_QUUX will be correctly defined. In other platforms,
2617the corresponding configuration step will hopefully do the same.
2618
2619In a pinch, if you cannot wait for Configure to be educated,
2620or if you have a good hunch of where quux() might be available,
2621you can temporarily try the following:
2622
2623 #if (defined(__FOONIX__) || defined(__BARNIX__))
2624 # define HAS_QUUX
2625 #endif
2626
2627 ...
2628
2629 #ifdef HAS_QUUX
2630 foo = quux();
2631 #endif
2632
2633But in any case, try to keep the features and operating systems separate.
2634
d7889f52
JH
2635=back
2636
ad7244db
JH
2637=head2 Problematic System Interfaces
2638
2639=over 4
2640
2641=item *
2642
353c6505 2643malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
ad7244db
JH
2644allocate at least one byte. (In general you should rarely need to
2645work at this low level, but instead use the various malloc wrappers.)
2646
2647=item *
2648
2649snprintf() - the return type is unportable. Use my_snprintf() instead.
2650
2651=back
2652
d7889f52
JH
2653=head2 Security problems
2654
2655Last but not least, here are various tips for safer coding.
2656
2657=over 4
2658
2659=item *
2660
2661Do not use gets()
2662
2663Or we will publicly ridicule you. Seriously.
2664
2665=item *
2666
d1307786 2667Do not use strcpy() or strcat() or strncpy() or strncat()
d7889f52 2668
d1307786
JH
2669Use my_strlcpy() and my_strlcat() instead: they either use the native
2670implementation, or Perl's own implementation (borrowed from the public
2671domain implementation of INN).
d7889f52
JH
2672
2673=item *
2674
2675Do not use sprintf() or vsprintf()
2676
0bec6c03 2677If you really want just plain byte strings, use my_snprintf()
64d9b66b 2678and my_vsnprintf() instead, which will try to use snprintf() and
0bec6c03
JH
2679vsnprintf() if those safer APIs are available. If you want something
2680fancier than a plain byte string, use SVs and Perl_sv_catpvf().
d7889f52
JH
2681
2682=back
2683
902b9dbf
MF
2684=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2685
2686Sometimes it helps to use external tools while debugging and
2687testing Perl. This section tries to guide you through using
2688some common testing and debugging tools with Perl. This is
2689meant as a guide to interfacing these tools with Perl, not
2690as any kind of guide to the use of the tools themselves.
2691
a958818a
JH
2692B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2693Third Degree greatly slows down the execution: seconds become minutes,
2694minutes become hours. For example as of Perl 5.8.1, the
2695ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2696e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2697than six hours, even on a snappy computer-- the said test must be
2698doing something that is quite unfriendly for memory debuggers. If you
2699don't feel like waiting, that you can simply kill away the perl
2700process.
2701
2702B<NOTE 2>: To minimize the number of memory leak false alarms (see
2703L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2704environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2705and harness scripts do that automatically. But if you are running
2706some of the tests manually-- for csh-like shells:
2707
2708 setenv PERL_DESTRUCT_LEVEL 2
2709
2710and for Bourne-type shells:
2711
2712 PERL_DESTRUCT_LEVEL=2
2713 export PERL_DESTRUCT_LEVEL
2714
2715or in UNIXy environments you can also use the C<env> command:
2716
2717 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
a1b65709 2718
37c0adeb
JH
2719B<NOTE 3>: There are known memory leaks when there are compile-time
2720errors within eval or require, seeing C<S_doeval> in the call stack
2721is a good sign of these. Fixing these leaks is non-trivial,
2722unfortunately, but they must be fixed eventually.
2723
f50e5b73
MH
2724B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2725unless Perl is built with the Configure option
2726C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2727
902b9dbf
MF
2728=head2 Rational Software's Purify
2729
2730Purify is a commercial tool that is helpful in identifying
2731memory overruns, wild pointers, memory leaks and other such
2732badness. Perl must be compiled in a specific way for
2733optimal testing with Purify. Purify is available under
2734Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2735
902b9dbf
MF
2736=head2 Purify on Unix
2737
2738On Unix, Purify creates a new Perl binary. To get the most
2739benefit out of Purify, you should create the perl to Purify
2740using:
2741
2742 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2743 -Uusemymalloc -Dusemultiplicity
2744
2745where these arguments mean:
2746
2747=over 4
2748
2749=item -Accflags=-DPURIFY
2750
2751Disables Perl's arena memory allocation functions, as well as
2752forcing use of memory allocation functions derived from the
2753system malloc.
2754
2755=item -Doptimize='-g'
2756
2757Adds debugging information so that you see the exact source
2758statements where the problem occurs. Without this flag, all
2759you will see is the source filename of where the error occurred.
2760
2761=item -Uusemymalloc
2762
2763Disable Perl's malloc so that Purify can more closely monitor
2764allocations and leaks. Using Perl's malloc will make Purify
2765report most leaks in the "potential" leaks category.
2766
2767=item -Dusemultiplicity
2768
2769Enabling the multiplicity option allows perl to clean up
2770thoroughly when the interpreter shuts down, which reduces the
2771number of bogus leak reports from Purify.
2772
2773=back
2774
2775Once you've compiled a perl suitable for Purify'ing, then you
2776can just:
2777
07aa3531 2778 make pureperl
902b9dbf
MF
2779
2780which creates a binary named 'pureperl' that has been Purify'ed.
2781This binary is used in place of the standard 'perl' binary
2782when you want to debug Perl memory problems.
2783
2784As an example, to show any memory leaks produced during the
2785standard Perl testset you would create and run the Purify'ed
2786perl as:
2787
2788 make pureperl
2789 cd t
07aa3531 2790 ../pureperl -I../lib harness
902b9dbf
MF
2791
2792which would run Perl on test.pl and report any memory problems.
2793
2794Purify outputs messages in "Viewer" windows by default. If
2795you don't have a windowing environment or if you simply
2796want the Purify output to unobtrusively go to a log file
2797instead of to the interactive window, use these following
2798options to output to the log file "perl.log":
2799
2800 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2801 -log-file=perl.log -append-logfile=yes"
2802
2803If you plan to use the "Viewer" windows, then you only need this option:
2804
2805 setenv PURIFYOPTIONS "-chain-length=25"
2806
c406981e
JH
2807In Bourne-type shells:
2808
98631ff8
JL
2809 PURIFYOPTIONS="..."
2810 export PURIFYOPTIONS
c406981e
JH
2811
2812or if you have the "env" utility:
2813
98631ff8 2814 env PURIFYOPTIONS="..." ../pureperl ...
c406981e 2815
902b9dbf
MF
2816=head2 Purify on NT
2817
2818Purify on Windows NT instruments the Perl binary 'perl.exe'
2819on the fly. There are several options in the makefile you
2820should change to get the most use out of Purify:
2821
2822=over 4
2823
2824=item DEFINES
2825
2826You should add -DPURIFY to the DEFINES line so the DEFINES
2827line looks something like:
2828
07aa3531 2829 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
902b9dbf
MF
2830
2831to disable Perl's arena memory allocation functions, as
2832well as to force use of memory allocation functions derived
2833from the system malloc.
2834
2835=item USE_MULTI = define
2836
2837Enabling the multiplicity option allows perl to clean up
2838thoroughly when the interpreter shuts down, which reduces the
2839number of bogus leak reports from Purify.
2840
2841=item #PERL_MALLOC = define
2842
2843Disable Perl's malloc so that Purify can more closely monitor
2844allocations and leaks. Using Perl's malloc will make Purify
2845report most leaks in the "potential" leaks category.
2846
2847=item CFG = Debug
2848
2849Adds debugging information so that you see the exact source
2850statements where the problem occurs. Without this flag, all
2851you will see is the source filename of where the error occurred.
2852
2853=back
2854
2855As an example, to show any memory leaks produced during the
2856standard Perl testset you would create and run Purify as:
2857
2858 cd win32
2859 make
2860 cd ../t
07aa3531 2861 purify ../perl -I../lib harness
902b9dbf
MF
2862
2863which would instrument Perl in memory, run Perl on test.pl,
2864then finally report any memory problems.
2865
7a834142
JH
2866=head2 valgrind
2867
2868The excellent valgrind tool can be used to find out both memory leaks
9df8f87f
LB
2869and illegal memory accesses. As of version 3.3.0, Valgrind only
2870supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind"
2871target can be used to run the tests under valgrind. Found errors
2872and memory leaks are logged in files named F<testfile.valgrind>.
07aa3531
JC
2873
2874Valgrind also provides a cachegrind tool, invoked on perl as:
2875
038c294a 2876 VG_OPTS=--tool=cachegrind make test.valgrind
d44161bf
MHM
2877
2878As system libraries (most notably glibc) are also triggering errors,
2879valgrind allows to suppress such errors using suppression files. The
2880default suppression file that comes with valgrind already catches a lot
2881of them. Some additional suppressions are defined in F<t/perl.supp>.
7a834142
JH
2882
2883To get valgrind and for more information see
2884
2885 http://developer.kde.org/~sewardj/
2886
f134cc4e 2887=head2 Compaq's/Digital's/HP's Third Degree
09187cb1
JH
2888
2889Third Degree is a tool for memory leak detection and memory access checks.
2890It is one of the many tools in the ATOM toolkit. The toolkit is only
2891available on Tru64 (formerly known as Digital UNIX formerly known as
2892DEC OSF/1).
2893
2894When building Perl, you must first run Configure with -Doptimize=-g
2895and -Uusemymalloc flags, after that you can use the make targets
51a35ef1
JH
2896"perl.third" and "test.third". (What is required is that Perl must be
2897compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 2898
64cea5fd 2899The short story is that with "atom" you can instrument the Perl
83f0ef60 2900executable to create a new executable called F<perl.third>. When the
4ae3d70a 2901instrumented executable is run, it creates a log of dubious memory
83f0ef60 2902traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a
JH
2903third for more information. The most extensive Third Degree
2904documentation is available in the Compaq "Tru64 UNIX Programmer's
2905Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 2906
9c54ecba 2907The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
64cea5fd
JH
2908subdirectory. There is a problem with these files: Third Degree is so
2909effective that it finds problems also in the system libraries.
9c54ecba
JH
2910Therefore you should used the Porting/thirdclean script to cleanup
2911the F<*.3log> files.
64cea5fd
JH
2912
2913There are also leaks that for given certain definition of a leak,
2914aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2915
2916=head2 PERL_DESTRUCT_LEVEL
2917
a958818a
JH
2918If you want to run any of the tests yourself manually using e.g.
2919valgrind, or the pureperl or perl.third executables, please note that
2920by default perl B<does not> explicitly cleanup all the memory it has
2921allocated (such as global memory arenas) but instead lets the exit()
2922of the whole program "take care" of such allocations, also known as
2923"global destruction of objects".
64cea5fd
JH
2924
2925There is a way to tell perl to do complete cleanup: set the
2926environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2927The t/TEST wrapper does set this to 2, and this is what you
2928need to do too, if you don't want to see the "global leaks":
1f56d61a 2929For example, for "third-degreed" Perl:
64cea5fd 2930
1f56d61a 2931 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
09187cb1 2932
414f2397
RGS
2933(Note: the mod_perl apache module uses also this environment variable
2934for its own purposes and extended its semantics. Refer to the mod_perl
287a822c
RGS
2935documentation for more information. Also, spawned threads do the
2936equivalent of setting this variable to the value 1.)
5a6c59ef
DM
2937
2938If, at the end of a run you get the message I<N scalars leaked>, you can
fd0854ff
DM
2939recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2940of all those leaked SVs to be dumped along with details as to where each
2941SV was originally allocated. This information is also displayed by
2942Devel::Peek. Note that the extra details recorded with each SV increases
2943memory usage, so it shouldn't be used in production environments. It also
2944converts C<new_SV()> from a macro into a real function, so you can use
2945your favourite debugger to discover where those pesky SVs were allocated.
414f2397 2946
d7a2c63c
MHM
2947If you see that you're leaking memory at runtime, but neither valgrind
2948nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
2949leaking SVs that are still reachable and will be properly cleaned up
2950during destruction of the interpreter. In such cases, using the C<-Dm>
2951switch can point you to the source of the leak. If the executable was
2952built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations
2953in addition to memory allocations. Each SV allocation has a distinct
2954serial number that will be written on creation and destruction of the SV.
2955So if you're executing the leaking code in a loop, you need to look for
2956SVs that are created, but never destroyed between each cycle. If such an
2957SV is found, set a conditional breakpoint within C<new_SV()> and make it
2958break only when C<PL_sv_serial> is equal to the serial number of the
2959leaking SV. Then you will catch the interpreter in exactly the state
2960where the leaking SV is allocated, which is sufficient in many cases to
2961find the source of the leak.
2962
2963As C<-Dm> is using the PerlIO layer for output, it will by itself
2964allocate quite a bunch of SVs, which are hidden to avoid recursion.
2965You can bypass the PerlIO layer if you use the SV logging provided
2966by C<-DPERL_MEM_LOG> instead.
2967
46c6c7e2
JH
2968=head2 PERL_MEM_LOG
2969
10a879f5
JC
2970If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go
2971through logging functions, which is handy for breakpoint setting.
2972
2973Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging
2e5b5004 2974functions read $ENV{PERL_MEM_LOG} to determine whether to log the
10a879f5
JC
2975event, and if so how:
2976
2e5b5004
RGS
2977 $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops
2978 $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops
2979 $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log
2980 $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2)
10a879f5
JC
2981
2982Memory logging is somewhat similar to C<-Dm> but is independent of
2983C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(),
2e5b5004 2984and Safefree() are logged with the caller's source code file and line
10a879f5
JC
2985number (and C function name, if supported by the C compiler). In
2986contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging
2987is similar.
2988
2989Since the logging doesn't use PerlIO, all SV allocations are logged
2990and no extra SV allocations are introduced by enabling the logging.
2991If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for
2992each SV allocation is also logged.
d7a2c63c 2993
51a35ef1
JH
2994=head2 Profiling
2995
2996Depending on your platform there are various of profiling Perl.
2997
2998There are two commonly used techniques of profiling executables:
10f58044 2999I<statistical time-sampling> and I<basic-block counting>.
51a35ef1
JH
3000
3001The first method takes periodically samples of the CPU program
3002counter, and since the program counter can be correlated with the code
3003generated for functions, we get a statistical view of in which
3004functions the program is spending its time. The caveats are that very
3005small/fast functions have lower probability of showing up in the
3006profile, and that periodically interrupting the program (this is
3007usually done rather frequently, in the scale of milliseconds) imposes
3008an additional overhead that may skew the results. The first problem
3009can be alleviated by running the code for longer (in general this is a
3010good idea for profiling), the second problem is usually kept in guard
3011by the profiling tools themselves.
3012
10f58044 3013The second method divides up the generated code into I<basic blocks>.
51a35ef1
JH
3014Basic blocks are sections of code that are entered only in the
3015beginning and exited only at the end. For example, a conditional jump
3016starts a basic block. Basic block profiling usually works by
10f58044 3017I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1
JH
3018book-keeping code to the generated code. During the execution of the
3019code the basic block counters are then updated appropriately. The
3020caveat is that the added extra code can skew the results: again, the
3021profiling tools usually try to factor their own effects out of the
3022results.
3023
83f0ef60
JH
3024=head2 Gprof Profiling
3025
51a35ef1
JH
3026gprof is a profiling tool available in many UNIX platforms,
3027it uses F<statistical time-sampling>.
83f0ef60
JH
3028
3029You can build a profiled version of perl called "perl.gprof" by
51a35ef1
JH
3030invoking the make target "perl.gprof" (What is required is that Perl
3031must be compiled using the C<-pg> flag, you may need to re-Configure).
3032Running the profiled version of Perl will create an output file called
3033F<gmon.out> is created which contains the profiling data collected
3034during the execution.
83f0ef60
JH
3035
3036The gprof tool can then display the collected data in various ways.
3037Usually gprof understands the following options:
3038
3039=over 4
3040
3041=item -a
3042
3043Suppress statically defined functions from the profile.
3044
3045=item -b
3046
3047Suppress the verbose descriptions in the profile.
3048
3049=item -e routine
3050
3051Exclude the given routine and its descendants from the profile.
3052
3053=item -f routine
3054
3055Display only the given routine and its descendants in the profile.
3056
3057=item -s
3058
3059Generate a summary file called F<gmon.sum> which then may be given
3060to subsequent gprof runs to accumulate data over several runs.
3061
3062=item -z
3063
3064Display routines that have zero usage.
3065
3066=back
3067
3068For more detailed explanation of the available commands and output
3069formats, see your own local documentation of gprof.
3070
038c294a 3071quick hint:
07aa3531 3072
289d61c2
JL
3073 $ sh Configure -des -Dusedevel -Doptimize='-pg' && make perl.gprof
3074 $ ./perl.gprof someprog # creates gmon.out in current directory
3075 $ gprof ./perl.gprof > out
07aa3531
JC
3076 $ view out
3077
51a35ef1
JH
3078=head2 GCC gcov Profiling
3079
10f58044 3080Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1
JH
3081for the GNU CC.
3082
3083You can build a profiled version of perl called F<perl.gcov> by
3084invoking the make target "perl.gcov" (what is required that Perl must
3085be compiled using gcc with the flags C<-fprofile-arcs
3086-ftest-coverage>, you may need to re-Configure).
3087
3088Running the profiled version of Perl will cause profile output to be
3089generated. For each source file an accompanying ".da" file will be
3090created.
3091
3092To display the results you use the "gcov" utility (which should
3093be installed if you have gcc 3.0 or newer installed). F<gcov> is
3094run on source code files, like this
3095
3096 gcov sv.c
3097
3098which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3099contain the source code annotated with relative frequencies of
3100execution indicated by "#" markers.
3101
3102Useful options of F<gcov> include C<-b> which will summarise the
3103basic block, branch, and function call coverage, and C<-c> which
3104instead of relative frequencies will use the actual counts. For
3105more information on the use of F<gcov> and basic block profiling
3106with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3107
3108 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3109
3110and its section titled "8. gcov: a Test Coverage Program"
3111
3112 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3113
07aa3531
JC
3114quick hint:
3115
3116 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3117 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3118 $ rm -f regexec.c.gcov regexec.gcda
3119 $ ./perl.gcov
3120 $ gcov regexec.c
3121 $ view regexec.c.gcov
3122
4ae3d70a
JH
3123=head2 Pixie Profiling
3124
51a35ef1
JH
3125Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3126UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 3127I<basic-block counting>.
4ae3d70a 3128
83f0ef60 3129You can build a profiled version of perl called F<perl.pixie> by
51a35ef1
JH
3130invoking the make target "perl.pixie" (what is required is that Perl
3131must be compiled using the C<-g> flag, you may need to re-Configure).
3132
3133In Tru64 a file called F<perl.Addrs> will also be silently created,
3134this file contains the addresses of the basic blocks. Running the
3135profiled version of Perl will create a new file called "perl.Counts"
3136which contains the counts for the basic block for that particular
3137program execution.
4ae3d70a 3138
51a35ef1 3139To display the results you use the F<prof> utility. The exact
4ae3d70a
JH
3140incantation depends on your operating system, "prof perl.Counts" in
3141IRIX, and "prof -pixie -all -L. perl" in Tru64.
3142
6c41479b
JH
3143In IRIX the following prof options are available:
3144
3145=over 4
3146
3147=item -h
3148
3149Reports the most heavily used lines in descending order of use.
6e36760b 3150Useful for finding the hotspot lines.
6c41479b
JH
3151
3152=item -l
3153
3154Groups lines by procedure, with procedures sorted in descending order of use.
3155Within a procedure, lines are listed in source order.
6e36760b 3156Useful for finding the hotspots of procedures.
6c41479b
JH
3157
3158=back
3159
3160In Tru64 the following options are available:
3161
3162=over 4
3163
3958b146 3164=item -p[rocedures]
6c41479b 3165
3958b146 3166Procedures sorted in descending order by the number of cycles executed
6e36760b 3167in each procedure. Useful for finding the hotspot procedures.
6c41479b
JH
3168(This is the default option.)
3169
24000d2f 3170=item -h[eavy]
6c41479b 3171
6e36760b
JH
3172Lines sorted in descending order by the number of cycles executed in
3173each line. Useful for finding the hotspot lines.
6c41479b 3174
24000d2f 3175=item -i[nvocations]
6c41479b 3176
6e36760b
JH
3177The called procedures are sorted in descending order by number of calls
3178made to the procedures. Useful for finding the most used procedures.
6c41479b 3179
24000d2f 3180=item -l[ines]
6c41479b
JH
3181
3182Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 3183Useful for finding the hotspots of procedures.
6c41479b
JH
3184
3185=item -testcoverage
3186
3187The compiler emitted code for these lines, but the code was unexecuted.
3188
24000d2f 3189=item -z[ero]
6c41479b
JH
3190
3191Unexecuted procedures.
3192
aa500c9e 3193=back
6c41479b
JH
3194
3195For further information, see your system's manual pages for pixie and prof.
4ae3d70a 3196
b8ddf6b3
SB
3197=head2 Miscellaneous tricks
3198
3199=over 4
3200
3201=item *
3202
cc177e1a 3203Those debugging perl with the DDD frontend over gdb may find the
b8ddf6b3
SB
3204following useful:
3205
3206You can extend the data conversion shortcuts menu, so for example you
3207can display an SV's IV value with one click, without doing any typing.
3208To do that simply edit ~/.ddd/init file and add after:
3209
3210 ! Display shortcuts.
3211 Ddd*gdbDisplayShortcuts: \
3212 /t () // Convert to Bin\n\
3213 /d () // Convert to Dec\n\
3214 /x () // Convert to Hex\n\
3215 /o () // Convert to Oct(\n\
3216
3217the following two lines:
3218
3219 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3220 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3221
3222so now you can do ivx and pvx lookups or you can plug there the
3223sv_peek "conversion":
3224
3225 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3226
3227(The my_perl is for threaded builds.)
3228Just remember that every line, but the last one, should end with \n\
3229
3230Alternatively edit the init file interactively via:
32313rd mouse button -> New Display -> Edit Menu
3232
3233Note: you can define up to 20 conversion shortcuts in the gdb
3234section.
3235
9965345d
JH
3236=item *
3237
7e337ee0
JH
3238If you see in a debugger a memory area mysteriously full of 0xABABABAB
3239or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3240see L<perlclib>.
9965345d 3241
f1fac472
NC
3242=item *
3243
3244Under ithreads the optree is read only. If you want to enforce this, to check
3245for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3246enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3247that allocates op memory via C<mmap>, and sets it read-only at run time.
3248Any write access to an op results in a C<SIGBUS> and abort.
3249
3250This code is intended for development only, and may not be portable even to
3251all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3252all ops read only. Specifically it
3253
3254=over
3255
3256=item 1
3257
3258Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3259later via C<require> or C<eval> will be re-write
3260
3261=item 2
3262
3263Turns an entire slab of ops read-write if the refcount of any op in the slab
3264needs to be decreased.
3265
3266=item 3
3267
3268Turns an entire slab of ops read-write if any op from the slab is freed.
3269
b8ddf6b3
SB
3270=back
3271
f1fac472
NC
3272It's not possible to turn the slabs to read-only after an action requiring
3273read-write access, as either can happen during op tree building time, so
3274there may still be legitimate write access.
3275
3276However, as an 80% solution it is still effective, as currently it catches
3277a write access during the generation of F<Config.pm>, which means that we
3278can't yet build F<perl> with this enabled.
3279
3280=back
3281
3282
955fec6b 3283=head1 CONCLUSION
a422fd2d 3284
955fec6b
JH
3285We've had a brief look around the Perl source, how to maintain quality
3286of the source code, an overview of the stages F<perl> goes through
3287when it's running your code, how to use debuggers to poke at the Perl
3288guts, and finally how to analyse the execution of Perl. We took a very
3289simple problem and demonstrated how to solve it fully - with
3290documentation, regression tests, and finally a patch for submission to
3291p5p. Finally, we talked about how to use external tools to debug and
3292test Perl.
a422fd2d
SC
3293
3294I'd now suggest you read over those references again, and then, as soon
3295as possible, get your hands dirty. The best way to learn is by doing,
07aa3531 3296so:
a422fd2d
SC
3297
3298=over 3
3299
3300=item *
3301
3302Subscribe to perl5-porters, follow the patches and try and understand
3303them; don't be afraid to ask if there's a portion you're not clear on -
3304who knows, you may unearth a bug in the patch...
3305
3306=item *
3307
3308Keep up to date with the bleeding edge Perl distributions and get
3309familiar with the changes. Try and get an idea of what areas people are
3310working on and the changes they're making.
3311
3312=item *
3313
3e148164 3314Do read the README associated with your operating system, e.g. README.aix
a1f349fd
MB
3315on the IBM AIX OS. Don't hesitate to supply patches to that README if
3316you find anything missing or changed over a new OS release.
3317
3318=item *
3319
a422fd2d
SC
3320Find an area of Perl that seems interesting to you, and see if you can
3321work out how it works. Scan through the source, and step over it in the
3322debugger. Play, poke, investigate, fiddle! You'll probably get to
3323understand not just your chosen area but a much wider range of F<perl>'s
3324activity as well, and probably sooner than you'd think.
3325
3326=back
3327
3328=over 3
3329
3330=item I<The Road goes ever on and on, down from the door where it began.>
3331
3332=back
3333
64d9b66b 3334If you can do these things, you've started on the long road to Perl porting.
a422fd2d
SC
3335Thanks for wanting to help make Perl better - and happy hacking!
3336
4ac71550
TC
3337=head2 Metaphoric Quotations
3338
3339If you recognized the quote about the Road above, you're in luck.
3340
3341Most software projects begin each file with a literal description of each
3342file's purpose. Perl instead begins each with a literary allusion to that
3343file's purpose.
3344
3345Like chapters in many books, all top-level Perl source files (along with a
3346few others here and there) begin with an epigramic inscription that alludes,
3347indirectly and metaphorically, to the material you're about to read.
3348
3349Quotations are taken from writings of J.R.R Tolkien pertaining to his
3350Legendarium, almost always from I<The Lord of the Rings>. Chapters and
3351page numbers are given using the following editions:
3352
3353=over 4
3354
3355=item *
3356
3357I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary
3358edition of 2007 was used, published in the UK by Harper Collins Publishers
3359and in the US by the Houghton Mifflin Company.
3360
3361=item *
3362
3363I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover,
336450th-anniversary edition of 2004 was used, published in the UK by Harper
3365Collins Publishers and in the US by the Houghton Mifflin Company.
3366
3367=item *
3368
3369I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his
3370son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes
3371in Christopher's mammoth I<History of Middle Earth>. Page numbers derive
3372from the hardcover edition, first published in 1983 by George Allen &
3373Unwin; no page numbers changed for the special 3-volume omnibus edition of
33742002 or the various trade-paper editions, all again now by Harper Collins
3375or Houghton Mifflin.
3376
3377=back
3378
3379Other JRRT books fair game for quotes would thus include I<The Adventures of
3380Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of
3381the Children of Hurin>, all but the first posthumously assembled by CJRT.
3382But I<The Lord of the Rings> itself is perfectly fine and probably best to
3383quote from, provided you can find a suitable quote there.
3384
3385So if you were to supply a new, complete, top-level source file to add to
3386Perl, you should conform to this peculiar practice by yourself selecting an
3387appropriate quotation from Tolkien, retaining the original spelling and
3388punctuation and using the same format the rest of the quotes are in.
3389Indirect and oblique is just fine; remember, it's a metaphor, so being meta
3390is, after all, what it's for.
3391
e8cd7eae
GS
3392=head1 AUTHOR
3393
3394This document was written by Nathan Torkington, and is maintained by
3395the perl5-porters mailing list.
4ac71550 3396
b16c2e4a
RGS
3397=head1 SEE ALSO
3398
3399L<perlrepository>