This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Fix typo in pod/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
324You might also want to look at Gisle Aas's illustrated perlguts -
325there's no guarantee that this will be absolutely up-to-date with the
326latest documentation in the Perl core, but the fundamentals will be
1577cd80 327right. ( http://gisle.aas.no/perl/illguts/ )
a422fd2d
SC
328
329=item L<perlxstut> and L<perlxs>
330
331A working knowledge of XSUB programming is incredibly useful for core
332hacking; XSUBs use techniques drawn from the PP code, the portion of the
333guts that actually executes a Perl program. It's a lot gentler to learn
334those techniques from simple examples and explanation than from the core
335itself.
336
337=item L<perlapi>
338
339The documentation for the Perl API explains what some of the internal
340functions do, as well as the many macros used in the source.
341
342=item F<Porting/pumpkin.pod>
343
344This is a collection of words of wisdom for a Perl porter; some of it is
345only useful to the pumpkin holder, but most of it applies to anyone
346wanting to go about Perl development.
347
348=item The perl5-porters FAQ
349
902821cc
RGS
350This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html .
351It contains hints on reading perl5-porters, information on how
352perl5-porters works and how Perl development in general works.
a422fd2d
SC
353
354=back
355
356=head2 Finding Your Way Around
357
358Perl maintenance can be split into a number of areas, and certain people
359(pumpkins) will have responsibility for each area. These areas sometimes
360correspond to files or directories in the source kit. Among the areas are:
361
362=over 3
363
364=item Core modules
365
366Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
367subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
368contains the core XS modules.
369
f7e1e956
MS
370=item Tests
371
372There are tests for nearly all the modules, built-ins and major bits
373of functionality. Test files all have a .t suffix. Module tests live
374in the F<lib/> and F<ext/> directories next to the module being
375tested. Others live in F<t/>. See L<Writing a test>
376
a422fd2d
SC
377=item Documentation
378
379Documentation maintenance includes looking after everything in the
380F<pod/> directory, (as well as contributing new documentation) and
381the documentation to the modules in core.
382
383=item Configure
384
385The configure process is the way we make Perl portable across the
386myriad of operating systems it supports. Responsibility for the
387configure, build and installation process, as well as the overall
388portability of the core code rests with the configure pumpkin - others
389help out with individual operating systems.
390
391The files involved are the operating system directories, (F<win32/>,
392F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
393and F<Makefile>, as well as the metaconfig files which generate
394F<Configure>. (metaconfig isn't included in the core distribution.)
395
396=item Interpreter
397
398And of course, there's the core of the Perl interpreter itself. Let's
399have a look at that in a little more detail.
400
401=back
402
403Before we leave looking at the layout, though, don't forget that
404F<MANIFEST> contains not only the file names in the Perl distribution,
405but short descriptions of what's in them, too. For an overview of the
406important files, try this:
407
408 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
409
410=head2 Elements of the interpreter
411
412The work of the interpreter has two main stages: compiling the code
413into the internal representation, or bytecode, and then executing it.
414L<perlguts/Compiled code> explains exactly how the compilation stage
415happens.
416
417Here is a short breakdown of perl's operation:
418
419=over 3
420
421=item Startup
422
423The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
424This is very high-level code, enough to fit on a single screen, and it
425resembles the code found in L<perlembed>; most of the real action takes
426place in F<perl.c>
427
9df8f87f
LB
428F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at
429make time, so you should make perl to follow this along.
430
a422fd2d 431First, F<perlmain.c> allocates some memory and constructs a Perl
9df8f87f 432interpreter, along these lines:
a422fd2d
SC
433
434 1 PERL_SYS_INIT3(&argc,&argv,&env);
435 2
436 3 if (!PL_do_undump) {
437 4 my_perl = perl_alloc();
438 5 if (!my_perl)
439 6 exit(1);
440 7 perl_construct(my_perl);
441 8 PL_perl_destruct_level = 0;
442 9 }
443
444Line 1 is a macro, and its definition is dependent on your operating
445system. Line 3 references C<PL_do_undump>, a global variable - all
446global variables in Perl start with C<PL_>. This tells you whether the
447current running program was created with the C<-u> flag to perl and then
448F<undump>, which means it's going to be false in any sane context.
449
450Line 4 calls a function in F<perl.c> to allocate memory for a Perl
451interpreter. It's quite a simple function, and the guts of it looks like
452this:
453
454 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
455
456Here you see an example of Perl's system abstraction, which we'll see
457later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
458own C<malloc> as defined in F<malloc.c> if you selected that option at
459configure time.
460
9df8f87f
LB
461Next, in line 7, we construct the interpreter using perl_construct,
462also in F<perl.c>; this sets up all the special variables that Perl
463needs, the stacks, and so on.
a422fd2d
SC
464
465Now we pass Perl the command line options, and tell it to go:
466
467 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
9df8f87f
LB
468 if (!exitstatus)
469 perl_run(my_perl);
470
471 exitstatus = perl_destruct(my_perl);
a422fd2d 472
9df8f87f 473 perl_free(my_perl);
a422fd2d
SC
474
475C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
476in F<perl.c>, which processes the command line options, sets up any
477statically linked XS modules, opens the program and calls C<yyparse> to
478parse it.
479
480=item Parsing
481
482The aim of this stage is to take the Perl source, and turn it into an op
483tree. We'll see what one of those looks like later. Strictly speaking,
484there's three things going on here.
485
486C<yyparse>, the parser, lives in F<perly.c>, although you're better off
487reading the original YACC input in F<perly.y>. (Yes, Virginia, there
488B<is> a YACC grammar for Perl!) The job of the parser is to take your
b432a672 489code and "understand" it, splitting it into sentences, deciding which
a422fd2d
SC
490operands go with which operators and so on.
491
492The parser is nobly assisted by the lexer, which chunks up your input
493into tokens, and decides what type of thing each token is: a variable
494name, an operator, a bareword, a subroutine, a core function, and so on.
495The main point of entry to the lexer is C<yylex>, and that and its
496associated routines can be found in F<toke.c>. Perl isn't much like
497other computer languages; it's highly context sensitive at times, it can
498be tricky to work out what sort of token something is, or where a token
499ends. As such, there's a lot of interplay between the tokeniser and the
500parser, which can get pretty frightening if you're not used to it.
501
502As the parser understands a Perl program, it builds up a tree of
503operations for the interpreter to perform during execution. The routines
504which construct and link together the various operations are to be found
505in F<op.c>, and will be examined later.
506
507=item Optimization
508
509Now the parsing stage is complete, and the finished tree represents
510the operations that the Perl interpreter needs to perform to execute our
511program. Next, Perl does a dry run over the tree looking for
512optimisations: constant expressions such as C<3 + 4> will be computed
513now, and the optimizer will also see if any multiple operations can be
514replaced with a single one. For instance, to fetch the variable C<$foo>,
515instead of grabbing the glob C<*foo> and looking at the scalar
516component, the optimizer fiddles the op tree to use a function which
517directly looks up the scalar in question. The main optimizer is C<peep>
518in F<op.c>, and many ops have their own optimizing functions.
519
520=item Running
521
522Now we're finally ready to go: we have compiled Perl byte code, and all
523that's left to do is run it. The actual execution is done by the
524C<runops_standard> function in F<run.c>; more specifically, it's done by
525these three innocent looking lines:
526
527 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
528 PERL_ASYNC_CHECK();
529 }
530
531You may be more comfortable with the Perl version of that:
532
533 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
534
535Well, maybe not. Anyway, each op contains a function pointer, which
536stipulates the function which will actually carry out the operation.
537This function will return the next op in the sequence - this allows for
538things like C<if> which choose the next op dynamically at run time.
539The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
540execution if required.
541
542The actual functions called are known as PP code, and they're spread
b432a672 543between four files: F<pp_hot.c> contains the "hot" code, which is most
a422fd2d
SC
544often used and highly optimized, F<pp_sys.c> contains all the
545system-specific functions, F<pp_ctl.c> contains the functions which
546implement control structures (C<if>, C<while> and the like) and F<pp.c>
547contains everything else. These are, if you like, the C code for Perl's
548built-in functions and operators.
549
dfc98234
DM
550Note that each C<pp_> function is expected to return a pointer to the next
551op. Calls to perl subs (and eval blocks) are handled within the same
552runops loop, and do not consume extra space on the C stack. For example,
553C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
554struct onto the context stack which contain the address of the op
555following the sub call or eval. They then return the first op of that sub
556or eval block, and so execution continues of that sub or block. Later, a
557C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
558retrieves the return op from it, and returns it.
559
560=item Exception handing
561
0503309d 562Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level
dfc98234 563C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
28a5cf3b 564way to capture the current PC and SP registers and later restore them; i.e.
dfc98234
DM
565a C<longjmp()> continues at the point in code where a previous C<setjmp()>
566was done, with anything further up on the C stack being lost. This is why
567code should always save values using C<SAVE_FOO> rather than in auto
568variables.
569
570The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
571C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
572C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
573C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
574
575At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
576C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
577loop or whatever, and handle possible exception returns. For a 2 return,
578final cleanup is performed, such as popping stacks and calling C<CHECK> or
579C<END> blocks. Amongst other things, this is how scope cleanup still
580occurs during an C<exit>.
581
582If a C<die> can find a C<CxEVAL> block on the context stack, then the
583stack is popped to that level and the return op in that block is assigned
584to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
585passes control back to the guard. In the case of C<perl_run> and
586C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
587loop. The is the normal way that C<die> or C<croak> is handled within an
588C<eval>.
589
590Sometimes ops are executed within an inner runops loop, such as tie, sort
591or overload code. In this case, something like
592
593 sub FETCH { eval { die } }
594
595would cause a longjmp right back to the guard in C<perl_run>, popping both
596runops loops, which is clearly incorrect. One way to avoid this is for the
597tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
598runops loop, but for efficiency reasons, perl in fact just sets a flag,
599using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
600C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
601which does a C<JMPENV_PUSH> and starts a new runops level to execute the
602code, rather than doing it on the current loop.
603
604As a further optimisation, on exit from the eval block in the C<FETCH>,
605execution of the code following the block is still carried on in the inner
606loop. When an exception is raised, C<docatch> compares the C<JMPENV>
607level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
608re-throws the exception. In this way any inner loops get popped.
609
610Here's an example.
611
612 1: eval { tie @a, 'A' };
613 2: sub A::TIEARRAY {
614 3: eval { die };
615 4: die;
616 5: }
617
618To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
619enters a runops loop. This loop executes the eval and tie ops on line 1,
620with the eval pushing a C<CxEVAL> onto the context stack.
621
622The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
623to execute the body of C<TIEARRAY>. When it executes the entertry op on
624line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
625does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
626the die op. At this point the C call stack looks like this:
627
628 Perl_pp_die
629 Perl_runops # third loop
630 S_docatch_body
631 S_docatch
632 Perl_pp_entertry
633 Perl_runops # second loop
634 S_call_body
635 Perl_call_sv
636 Perl_pp_tie
637 Perl_runops # first loop
638 S_run_body
639 perl_run
640 main
641
642and the context and data stacks, as shown by C<-Dstv>, look like:
643
644 STACK 0: MAIN
645 CX 0: BLOCK =>
646 CX 1: EVAL => AV() PV("A"\0)
647 retop=leave
648 STACK 1: MAGIC
649 CX 0: SUB =>
650 retop=(null)
651 CX 1: EVAL => *
652 retop=nextstate
653
654The die pops the first C<CxEVAL> off the context stack, sets
655C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
656the top C<docatch>. This then starts another third-level runops level,
657which executes the nextstate, pushmark and die ops on line 4. At the point
658that the second C<pp_die> is called, the C call stack looks exactly like
659that above, even though we are no longer within an inner eval; this is
660because of the optimization mentioned earlier. However, the context stack
661now looks like this, ie with the top CxEVAL popped:
662
663 STACK 0: MAIN
664 CX 0: BLOCK =>
665 CX 1: EVAL => AV() PV("A"\0)
666 retop=leave
667 STACK 1: MAGIC
668 CX 0: SUB =>
669 retop=(null)
670
671The die on line 4 pops the context stack back down to the CxEVAL, leaving
672it as:
673
674 STACK 0: MAIN
675 CX 0: BLOCK =>
676
677As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
678C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
679
680 S_docatch
681 Perl_pp_entertry
682 Perl_runops # second loop
683 S_call_body
684 Perl_call_sv
685 Perl_pp_tie
686 Perl_runops # first loop
687 S_run_body
688 perl_run
689 main
690
691In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
692differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
693and the C stack unwinds to:
694
695 perl_run
696 main
697
698Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
699and execution continues.
700
a422fd2d
SC
701=back
702
703=head2 Internal Variable Types
704
705You should by now have had a look at L<perlguts>, which tells you about
706Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
707that now.
708
709These variables are used not only to represent Perl-space variables, but
710also any constants in the code, as well as some structures completely
711internal to Perl. The symbol table, for instance, is an ordinary Perl
712hash. Your code is represented by an SV as it's read into the parser;
713any program files you call are opened via ordinary Perl filehandles, and
714so on.
715
716The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
717Perl program. Let's see, for instance, how Perl treats the constant
718C<"hello">.
719
720 % perl -MDevel::Peek -e 'Dump("hello")'
721 1 SV = PV(0xa041450) at 0xa04ecbc
722 2 REFCNT = 1
723 3 FLAGS = (POK,READONLY,pPOK)
724 4 PV = 0xa0484e0 "hello"\0
725 5 CUR = 5
726 6 LEN = 6
727
728Reading C<Devel::Peek> output takes a bit of practise, so let's go
729through it line by line.
730
731Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
732memory. SVs themselves are very simple structures, but they contain a
733pointer to a more complex structure. In this case, it's a PV, a
734structure which holds a string value, at location C<0xa041450>. Line 2
735is the reference count; there are no other references to this data, so
736it's 1.
737
738Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
739read-only SV (because it's a constant) and the data is a PV internally.
740Next we've got the contents of the string, starting at location
741C<0xa0484e0>.
742
743Line 5 gives us the current length of the string - note that this does
744B<not> include the null terminator. Line 6 is not the length of the
745string, but the length of the currently allocated buffer; as the string
746grows, Perl automatically extends the available storage via a routine
747called C<SvGROW>.
748
749You can get at any of these quantities from C very easily; just add
750C<Sv> to the name of the field shown in the snippet, and you've got a
751macro which will return the value: C<SvCUR(sv)> returns the current
752length of the string, C<SvREFCOUNT(sv)> returns the reference count,
753C<SvPV(sv, len)> returns the string itself with its length, and so on.
754More macros to manipulate these properties can be found in L<perlguts>.
755
756Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
757
758 1 void
759 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
760 3 {
761 4 STRLEN tlen;
762 5 char *junk;
763
764 6 junk = SvPV_force(sv, tlen);
765 7 SvGROW(sv, tlen + len + 1);
766 8 if (ptr == junk)
767 9 ptr = SvPVX(sv);
768 10 Move(ptr,SvPVX(sv)+tlen,len,char);
769 11 SvCUR(sv) += len;
770 12 *SvEND(sv) = '\0';
771 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
772 14 SvTAINT(sv);
773 15 }
774
775This is a function which adds a string, C<ptr>, of length C<len> onto
776the end of the PV stored in C<sv>. The first thing we do in line 6 is
777make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
778macro to force a PV. As a side effect, C<tlen> gets set to the current
779value of the PV, and the PV itself is returned to C<junk>.
780
b1866b2d 781In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d
SC
782the old string, the new string and the null terminator. If C<LEN> isn't
783big enough, C<SvGROW> will reallocate space for us.
784
785Now, if C<junk> is the same as the string we're trying to add, we can
786grab the string directly from the SV; C<SvPVX> is the address of the PV
787in the SV.
788
789Line 10 does the actual catenation: the C<Move> macro moves a chunk of
790memory around: we move the string C<ptr> to the end of the PV - that's
791the start of the PV plus its current length. We're moving C<len> bytes
792of type C<char>. After doing so, we need to tell Perl we've extended the
793string, by altering C<CUR> to reflect the new length. C<SvEND> is a
794macro which gives us the end of the string, so that needs to be a
795C<"\0">.
796
797Line 13 manipulates the flags; since we've changed the PV, any IV or NV
798values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1e54db1a 799want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
a422fd2d
SC
800version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
801and turns on POK. The final C<SvTAINT> is a macro which launders tainted
802data if taint mode is turned on.
803
804AVs and HVs are more complicated, but SVs are by far the most common
805variable type being thrown around. Having seen something of how we
806manipulate these, let's go on and look at how the op tree is
807constructed.
808
809=head2 Op Trees
810
811First, what is the op tree, anyway? The op tree is the parsed
812representation of your program, as we saw in our section on parsing, and
813it's the sequence of operations that Perl goes through to execute your
814program, as we saw in L</Running>.
815
816An op is a fundamental operation that Perl can perform: all the built-in
817functions and operators are ops, and there are a series of ops which
818deal with concepts the interpreter needs internally - entering and
819leaving a block, ending a statement, fetching a variable, and so on.
820
821The op tree is connected in two ways: you can imagine that there are two
822"routes" through it, two orders in which you can traverse the tree.
823First, parse order reflects how the parser understood the code, and
824secondly, execution order tells perl what order to perform the
825operations in.
826
827The easiest way to examine the op tree is to stop Perl after it has
828finished parsing, and get it to dump out the tree. This is exactly what
7d7d5695
RGS
829the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
830and L<B::Debug|B::Debug> do.
a422fd2d
SC
831
832Let's have a look at how Perl sees C<$a = $b + $c>:
833
834 % perl -MO=Terse -e '$a=$b+$c'
835 1 LISTOP (0x8179888) leave
836 2 OP (0x81798b0) enter
837 3 COP (0x8179850) nextstate
838 4 BINOP (0x8179828) sassign
839 5 BINOP (0x8179800) add [1]
840 6 UNOP (0x81796e0) null [15]
841 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
842 8 UNOP (0x81797e0) null [15]
843 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
844 10 UNOP (0x816b4f0) null [15]
845 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
846
847Let's start in the middle, at line 4. This is a BINOP, a binary
848operator, which is at location C<0x8179828>. The specific operator in
849question is C<sassign> - scalar assignment - and you can find the code
850which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
851binary operator, it has two children: the add operator, providing the
852result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
853line 10.
854
855Line 10 is the null op: this does exactly nothing. What is that doing
856there? If you see the null op, it's a sign that something has been
857optimized away after parsing. As we mentioned in L</Optimization>,
858the optimization stage sometimes converts two operations into one, for
859example when fetching a scalar variable. When this happens, instead of
860rewriting the op tree and cleaning up the dangling pointers, it's easier
861just to replace the redundant operation with the null op. Originally,
862the tree would have looked like this:
863
864 10 SVOP (0x816b4f0) rv2sv [15]
865 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
866
867That is, fetch the C<a> entry from the main symbol table, and then look
868at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
869happens to do both these things.
870
871The right hand side, starting at line 5 is similar to what we've just
872seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
873two C<gvsv>s.
874
875Now, what's this about?
876
877 1 LISTOP (0x8179888) leave
878 2 OP (0x81798b0) enter
879 3 COP (0x8179850) nextstate
880
881C<enter> and C<leave> are scoping ops, and their job is to perform any
882housekeeping every time you enter and leave a block: lexical variables
883are tidied up, unreferenced variables are destroyed, and so on. Every
884program will have those first three lines: C<leave> is a list, and its
885children are all the statements in the block. Statements are delimited
886by C<nextstate>, so a block is a collection of C<nextstate> ops, with
887the ops to be performed for each statement being the children of
888C<nextstate>. C<enter> is a single op which functions as a marker.
889
890That's how Perl parsed the program, from top to bottom:
891
892 Program
893 |
894 Statement
895 |
896 =
897 / \
898 / \
899 $a +
900 / \
901 $b $c
902
903However, it's impossible to B<perform> the operations in this order:
904you have to find the values of C<$b> and C<$c> before you add them
905together, for instance. So, the other thread that runs through the op
906tree is the execution order: each op has a field C<op_next> which points
907to the next op to be run, so following these pointers tells us how perl
908executes the code. We can traverse the tree in this order using
909the C<exec> option to C<B::Terse>:
910
911 % perl -MO=Terse,exec -e '$a=$b+$c'
912 1 OP (0x8179928) enter
913 2 COP (0x81798c8) nextstate
914 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
915 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
916 5 BINOP (0x8179878) add [1]
917 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
918 7 BINOP (0x81798a0) sassign
919 8 LISTOP (0x8179900) leave
920
921This probably makes more sense for a human: enter a block, start a
922statement. Get the values of C<$b> and C<$c>, and add them together.
923Find C<$a>, and assign one to the other. Then leave.
924
925The way Perl builds up these op trees in the parsing process can be
926unravelled by examining F<perly.y>, the YACC grammar. Let's take the
927piece we need to construct the tree for C<$a = $b + $c>
928
929 1 term : term ASSIGNOP term
930 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
931 3 | term ADDOP term
932 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
933
934If you're not used to reading BNF grammars, this is how it works: You're
935fed certain things by the tokeniser, which generally end up in upper
936case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
937code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
b432a672 938"terminal symbols", because you can't get any simpler than them.
a422fd2d
SC
939
940The grammar, lines one and three of the snippet above, tells you how to
b432a672 941build up more complex forms. These complex forms, "non-terminal symbols"
a422fd2d
SC
942are generally placed in lower case. C<term> here is a non-terminal
943symbol, representing a single expression.
944
945The grammar gives you the following rule: you can make the thing on the
946left of the colon if you see all the things on the right in sequence.
947This is called a "reduction", and the aim of parsing is to completely
948reduce the input. There are several different ways you can perform a
949reduction, separated by vertical bars: so, C<term> followed by C<=>
950followed by C<term> makes a C<term>, and C<term> followed by C<+>
951followed by C<term> can also make a C<term>.
952
953So, if you see two terms with an C<=> or C<+>, between them, you can
954turn them into a single expression. When you do this, you execute the
955code in the block on the next line: if you see C<=>, you'll do the code
956in line 2. If you see C<+>, you'll do the code in line 4. It's this code
957which contributes to the op tree.
958
959 | term ADDOP term
960 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
961
962What this does is creates a new binary op, and feeds it a number of
963variables. The variables refer to the tokens: C<$1> is the first token in
964the input, C<$2> the second, and so on - think regular expression
965backreferences. C<$$> is the op returned from this reduction. So, we
966call C<newBINOP> to create a new binary operator. The first parameter to
967C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
968operator, so we want the type to be C<ADDOP>. We could specify this
969directly, but it's right there as the second token in the input, so we
b432a672
AL
970use C<$2>. The second parameter is the op's flags: 0 means "nothing
971special". Then the things to add: the left and right hand side of our
a422fd2d
SC
972expression, in scalar context.
973
974=head2 Stacks
975
976When perl executes something like C<addop>, how does it pass on its
977results to the next op? The answer is, through the use of stacks. Perl
978has a number of stacks to store things it's currently working on, and
979we'll look at the three most important ones here.
980
981=over 3
982
983=item Argument stack
984
985Arguments are passed to PP code and returned from PP code using the
986argument stack, C<ST>. The typical way to handle arguments is to pop
987them off the stack, deal with them how you wish, and then push the result
988back onto the stack. This is how, for instance, the cosine operator
989works:
990
991 NV value;
992 value = POPn;
993 value = Perl_cos(value);
994 XPUSHn(value);
995
996We'll see a more tricky example of this when we consider Perl's macros
997below. C<POPn> gives you the NV (floating point value) of the top SV on
998the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
999the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1000should be extended if necessary - it can't be necessary here, because we
1001know there's room for one more item on the stack, since we've just
1002removed one! The C<XPUSH*> macros at least guarantee safety.
1003
1004Alternatively, you can fiddle with the stack directly: C<SP> gives you
1005the first element in your portion of the stack, and C<TOP*> gives you
1006the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1007negation of an integer:
1008
1009 SETi(-TOPi);
1010
1011Just set the integer value of the top stack entry to its negation.
1012
1013Argument stack manipulation in the core is exactly the same as it is in
1014XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1015description of the macros used in stack manipulation.
1016
1017=item Mark stack
1018
b432a672 1019I say "your portion of the stack" above because PP code doesn't
a422fd2d
SC
1020necessarily get the whole stack to itself: if your function calls
1021another function, you'll only want to expose the arguments aimed for the
1022called function, and not (necessarily) let it get at your own data. The
b432a672 1023way we do this is to have a "virtual" bottom-of-stack, exposed to each
a422fd2d
SC
1024function. The mark stack keeps bookmarks to locations in the argument
1025stack usable by each function. For instance, when dealing with a tied
b432a672 1026variable, (internally, something with "P" magic) Perl has to call
a422fd2d
SC
1027methods for accesses to the tied variables. However, we need to separate
1028the arguments exposed to the method to the argument exposed to the
ed233832
DM
1029original function - the store or fetch or whatever it may be. Here's
1030roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
a422fd2d
SC
1031
1032 1 PUSHMARK(SP);
1033 2 EXTEND(SP,2);
1034 3 PUSHs(SvTIED_obj((SV*)av, mg));
1035 4 PUSHs(val);
1036 5 PUTBACK;
1037 6 ENTER;
1038 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1039 8 LEAVE;
13a2d996 1040
a422fd2d
SC
1041Let's examine the whole implementation, for practice:
1042
1043 1 PUSHMARK(SP);
1044
1045Push the current state of the stack pointer onto the mark stack. This is
1046so that when we've finished adding items to the argument stack, Perl
1047knows how many things we've added recently.
1048
1049 2 EXTEND(SP,2);
1050 3 PUSHs(SvTIED_obj((SV*)av, mg));
1051 4 PUSHs(val);
1052
1053We're going to add two more items onto the argument stack: when you have
1054a tied array, the C<PUSH> subroutine receives the object and the value
1055to be pushed, and that's exactly what we have here - the tied object,
1056retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1057
1058 5 PUTBACK;
1059
e89a6d4e
JD
1060Next we tell Perl to update the global stack pointer from our internal
1061variable: C<dSP> only gave us a local copy, not a reference to the global.
a422fd2d
SC
1062
1063 6 ENTER;
1064 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1065 8 LEAVE;
1066
1067C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1068variables are tidied up, everything that has been localised gets
1069its previous value returned, and so on. Think of them as the C<{> and
1070C<}> of a Perl block.
1071
1072To actually do the magic method call, we have to call a subroutine in
1073Perl space: C<call_method> takes care of that, and it's described in
1074L<perlcall>. We call the C<PUSH> method in scalar context, and we're
e89a6d4e
JD
1075going to discard its return value. The call_method() function
1076removes the top element of the mark stack, so there is nothing for
1077the caller to clean up.
a422fd2d 1078
a422fd2d
SC
1079=item Save stack
1080
1081C doesn't have a concept of local scope, so perl provides one. We've
1082seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1083stack implements the C equivalent of, for example:
1084
1085 {
1086 local $foo = 42;
1087 ...
1088 }
1089
1090See L<perlguts/Localising Changes> for how to use the save stack.
1091
1092=back
1093
1094=head2 Millions of Macros
1095
1096One thing you'll notice about the Perl source is that it's full of
1097macros. Some have called the pervasive use of macros the hardest thing
1098to understand, others find it adds to clarity. Let's take an example,
1099the code which implements the addition operator:
1100
1101 1 PP(pp_add)
1102 2 {
39644a26 1103 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1104 4 {
1105 5 dPOPTOPnnrl_ul;
1106 6 SETn( left + right );
1107 7 RETURN;
1108 8 }
1109 9 }
1110
1111Every line here (apart from the braces, of course) contains a macro. The
1112first line sets up the function declaration as Perl expects for PP code;
1113line 3 sets up variable declarations for the argument stack and the
1114target, the return value of the operation. Finally, it tries to see if
1115the addition operation is overloaded; if so, the appropriate subroutine
1116is called.
1117
1118Line 5 is another variable declaration - all variable declarations start
1119with C<d> - which pops from the top of the argument stack two NVs (hence
1120C<nn>) and puts them into the variables C<right> and C<left>, hence the
1121C<rl>. These are the two operands to the addition operator. Next, we
1122call C<SETn> to set the NV of the return value to the result of adding
1123the two values. This done, we return - the C<RETURN> macro makes sure
1124that our return value is properly handled, and we pass the next operator
1125to run back to the main run loop.
1126
1127Most of these macros are explained in L<perlapi>, and some of the more
1128important ones are explained in L<perlxs> as well. Pay special attention
1129to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1130the C<[pad]THX_?> macros.
1131
52d59bef
JH
1132=head2 The .i Targets
1133
1134You can expand the macros in a F<foo.c> file by saying
1135
1136 make foo.i
1137
1138which will expand the macros using cpp. Don't be scared by the results.
1139
955fec6b
JH
1140=head1 SOURCE CODE STATIC ANALYSIS
1141
1142Various tools exist for analysing C source code B<statically>, as
1143opposed to B<dynamically>, that is, without executing the code.
1144It is possible to detect resource leaks, undefined behaviour, type
1145mismatches, portability problems, code paths that would cause illegal
1146memory accesses, and other similar problems by just parsing the C code
1147and looking at the resulting graph, what does it tell about the
1148execution and data flows. As a matter of fact, this is exactly
1149how C compilers know to give warnings about dubious code.
1150
1151=head2 lint, splint
1152
1153The good old C code quality inspector, C<lint>, is available in
1154several platforms, but please be aware that there are several
1155different implementations of it by different vendors, which means that
1156the flags are not identical across different platforms.
1157
1158There is a lint variant called C<splint> (Secure Programming Lint)
1159available from http://www.splint.org/ that should compile on any
1160Unix-like platform.
1161
1162There are C<lint> and <splint> targets in Makefile, but you may have
1163to diddle with the flags (see above).
1164
1165=head2 Coverity
1166
1167Coverity (http://www.coverity.com/) is a product similar to lint and
1168as a testbed for their product they periodically check several open
1169source projects, and they give out accounts to open source developers
1170to the defect databases.
1171
1172=head2 cpd (cut-and-paste detector)
1173
1174The cpd tool detects cut-and-paste coding. If one instance of the
1175cut-and-pasted code changes, all the other spots should probably be
1176changed, too. Therefore such code should probably be turned into a
1177subroutine or a macro.
1178
1179cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1180(http://pmd.sourceforge.net/). pmd was originally written for static
1181analysis of Java code, but later the cpd part of it was extended to
1182parse also C and C++.
1183
a52aaefa
RGS
1184Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1185pmd-X.Y.jar from it, and then run that on source code thusly:
955fec6b
JH
1186
1187 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1188
1189You may run into memory limits, in which case you should use the -Xmx option:
1190
1191 java -Xmx512M ...
1192
1193=head2 gcc warnings
1194
1195Though much can be written about the inconsistency and coverage
1196problems of gcc warnings (like C<-Wall> not meaning "all the
1197warnings", or some common portability problems not being covered by
1198C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1199collection of warnings, and so forth), gcc is still a useful tool in
1200keeping our coding nose clean.
1201
1202The C<-Wall> is by default on.
1203
a8e98a71
JH
1204The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1205always, but unfortunately they are not safe on all platforms, they can
1206for example cause fatal conflicts with the system headers (Solaris
1207being a prime example). If Configure C<-Dgccansipedantic> is used,
1208the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1209where they are known to be safe.
955fec6b
JH
1210
1211Starting from Perl 5.9.4 the following extra flags are added:
1212
1213=over 4
1214
1215=item *
1216
1217C<-Wendif-labels>
1218
1219=item *
1220
1221C<-Wextra>
1222
1223=item *
1224
1225C<-Wdeclaration-after-statement>
1226
1227=back
1228
1229The following flags would be nice to have but they would first need
0503309d 1230their own Augean stablemaster:
955fec6b
JH
1231
1232=over 4
1233
1234=item *
1235
1236C<-Wpointer-arith>
1237
1238=item *
1239
1240C<-Wshadow>
1241
1242=item *
1243
1244C<-Wstrict-prototypes>
1245
955fec6b
JH
1246=back
1247
1248The C<-Wtraditional> is another example of the annoying tendency of
1249gcc to bundle a lot of warnings under one switch -- it would be
1250impossible to deploy in practice because it would complain a lot -- but
1251it does contain some warnings that would be beneficial to have available
1252on their own, such as the warning about string constants inside macros
1253containing the macro arguments: this behaved differently pre-ANSI
1254than it does in ANSI, and some C compilers are still in transition,
1255AIX being an example.
1256
1257=head2 Warnings of other C compilers
1258
1259Other C compilers (yes, there B<are> other C compilers than gcc) often
1260have their "strict ANSI" or "strict ANSI with some portability extensions"
1261modes on, like for example the Sun Workshop has its C<-Xa> mode on
1262(though implicitly), or the DEC (these days, HP...) has its C<-std1>
1263mode on.
1264
1265=head2 DEBUGGING
1266
1267You can compile a special debugging version of Perl, which allows you
1268to use the C<-D> option of Perl to tell more about what Perl is doing.
1269But sometimes there is no alternative than to dive in with a debugger,
1270either to see the stack trace of a core dump (very useful in a bug
1271report), or trying to figure out what went wrong before the core dump
1272happened, or how did we end up having wrong or unexpected results.
1273
a422fd2d
SC
1274=head2 Poking at Perl
1275
1276To really poke around with Perl, you'll probably want to build Perl for
1277debugging, like this:
1278
1279 ./Configure -d -D optimize=-g
1280 make
1281
1282C<-g> is a flag to the C compiler to have it produce debugging
955fec6b
JH
1283information which will allow us to step through a running program,
1284and to see in which C function we are at (without the debugging
1285information we might see only the numerical addresses of the functions,
1286which is not very helpful).
1287
a422fd2d
SC
1288F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1289enables all the internal debugging code in Perl. There are a whole bunch
1290of things you can debug with this: L<perlrun> lists them all, and the
1291best way to find out about them is to play about with them. The most
1292useful options are probably
1293
1294 l Context (loop) stack processing
1295 t Trace execution
1296 o Method and overloading resolution
1297 c String/numeric conversions
1298
1299Some of the functionality of the debugging code can be achieved using XS
1300modules.
13a2d996 1301
a422fd2d
SC
1302 -Dr => use re 'debug'
1303 -Dx => use O 'Debug'
1304
1305=head2 Using a source-level debugger
1306
1307If the debugging output of C<-D> doesn't help you, it's time to step
1308through perl's execution with a source-level debugger.
1309
1310=over 3
1311
1312=item *
1313
955fec6b
JH
1314We'll use C<gdb> for our examples here; the principles will apply to
1315any debugger (many vendors call their debugger C<dbx>), but check the
1316manual of the one you're using.
a422fd2d
SC
1317
1318=back
1319
1320To fire up the debugger, type
1321
1322 gdb ./perl
1323
955fec6b
JH
1324Or if you have a core dump:
1325
1326 gdb ./perl core
1327
a422fd2d
SC
1328You'll want to do that in your Perl source tree so the debugger can read
1329the source code. You should see the copyright message, followed by the
1330prompt.
1331
1332 (gdb)
1333
1334C<help> will get you into the documentation, but here are the most
1335useful commands:
1336
1337=over 3
1338
1339=item run [args]
1340
1341Run the program with the given arguments.
1342
1343=item break function_name
1344
1345=item break source.c:xxx
1346
1347Tells the debugger that we'll want to pause execution when we reach
cea6626f 1348either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d
SC
1349line in the named source file.
1350
1351=item step
1352
1353Steps through the program a line at a time.
1354
1355=item next
1356
1357Steps through the program a line at a time, without descending into
1358functions.
1359
1360=item continue
1361
1362Run until the next breakpoint.
1363
1364=item finish
1365
1366Run until the end of the current function, then stop again.
1367
13a2d996 1368=item 'enter'
a422fd2d
SC
1369
1370Just pressing Enter will do the most recent operation again - it's a
1371blessing when stepping through miles of source code.
1372
1373=item print
1374
1375Execute the given C code and print its results. B<WARNING>: Perl makes
52d59bef
JH
1376heavy use of macros, and F<gdb> does not necessarily support macros
1377(see later L</"gdb macro support">). You'll have to substitute them
1378yourself, or to invoke cpp on the source code files
1379(see L</"The .i Targets">)
1380So, for instance, you can't say
a422fd2d
SC
1381
1382 print SvPV_nolen(sv)
1383
1384but you have to say
1385
1386 print Perl_sv_2pv_nolen(sv)
1387
ffc145e8
RK
1388=back
1389
a422fd2d
SC
1390You may find it helpful to have a "macro dictionary", which you can
1391produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
07aa3531 1392recursively apply those macros for you.
52d59bef
JH
1393
1394=head2 gdb macro support
a422fd2d 1395
52d59bef 1396Recent versions of F<gdb> have fairly good macro support, but
ea031e66
RGS
1397in order to use it you'll need to compile perl with macro definitions
1398included in the debugging information. Using F<gcc> version 3.1, this
1399means configuring with C<-Doptimize=-g3>. Other compilers might use a
1400different switch (if they support debugging macros at all).
1401
a422fd2d
SC
1402=head2 Dumping Perl Data Structures
1403
1404One way to get around this macro hell is to use the dumping functions in
1405F<dump.c>; these work a little like an internal
1406L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1407that you can't get at from Perl. Let's take an example. We'll use the
07aa3531 1408C<$a = $b + $c> we used before, but give it a bit of context:
a422fd2d
SC
1409C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1410
1411What about C<pp_add>, the function we examined earlier to implement the
1412C<+> operator:
1413
1414 (gdb) break Perl_pp_add
1415 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1416
cea6626f 1417Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d
SC
1418With the breakpoint in place, we can run our program:
1419
1420 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1421
1422Lots of junk will go past as gdb reads in the relevant source files and
1423libraries, and then:
1424
1425 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1426 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1427 (gdb) step
1428 311 dPOPTOPnnrl_ul;
1429 (gdb)
1430
1431We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1432arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1433slightly expand it:
1434
1435 #define dPOPTOPnnrl_ul NV right = POPn; \
1436 SV *leftsv = TOPs; \
1437 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1438
1439C<POPn> takes the SV from the top of the stack and obtains its NV either
1440directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1441C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1442C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
07aa3531 1443C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
a422fd2d
SC
1444
1445Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1446convert it. If we step again, we'll find ourselves there:
1447
1448 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1449 1669 if (!sv)
1450 (gdb)
1451
1452We can now use C<Perl_sv_dump> to investigate the SV:
1453
1454 SV = PV(0xa057cc0) at 0xa0675d0
1455 REFCNT = 1
1456 FLAGS = (POK,pPOK)
1457 PV = 0xa06a510 "6XXXX"\0
1458 CUR = 5
1459 LEN = 6
1460 $1 = void
1461
1462We know we're going to get C<6> from this, so let's finish the
1463subroutine:
1464
1465 (gdb) finish
1466 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1467 0x462669 in Perl_pp_add () at pp_hot.c:311
1468 311 dPOPTOPnnrl_ul;
1469
1470We can also dump out this op: the current op is always stored in
1471C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1472similar output to L<B::Debug|B::Debug>.
1473
1474 {
1475 13 TYPE = add ===> 14
1476 TARG = 1
1477 FLAGS = (SCALAR,KIDS)
1478 {
1479 TYPE = null ===> (12)
1480 (was rv2sv)
1481 FLAGS = (SCALAR,KIDS)
1482 {
1483 11 TYPE = gvsv ===> 12
1484 FLAGS = (SCALAR)
1485 GV = main::b
1486 }
1487 }
1488
10f58044 1489# finish this later #
a422fd2d
SC
1490
1491=head2 Patching
1492
1493All right, we've now had a look at how to navigate the Perl sources and
1494some things you'll need to know when fiddling with them. Let's now get
1495on and create a simple patch. Here's something Larry suggested: if a
07aa3531 1496C<U> is the first active format during a C<pack>, (for example,
a422fd2d 1497C<pack "U3C8", @stuff>) then the resulting string should be treated as
1e54db1a 1498UTF-8 encoded.
a422fd2d
SC
1499
1500How do we prepare to fix this up? First we locate the code in question -
1501the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1502files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1503altering this file, let's copy it to F<pp.c~>.
1504
a6ec74c1
JH
1505[Well, it was in F<pp.c> when this tutorial was written. It has now been
1506split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1507
a422fd2d
SC
1508Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1509loop over the pattern, taking each format character in turn into
1510C<datum_type>. Then for each possible format character, we swallow up
1511the other arguments in the pattern (a field width, an asterisk, and so
1512on) and convert the next chunk input into the specified format, adding
1513it onto the output SV C<cat>.
1514
1515How do we know if the C<U> is the first format in the C<pat>? Well, if
1516we have a pointer to the start of C<pat> then, if we see a C<U> we can
1517test whether we're still at the start of the string. So, here's where
1518C<pat> is set up:
1519
1520 STRLEN fromlen;
1521 register char *pat = SvPVx(*++MARK, fromlen);
1522 register char *patend = pat + fromlen;
1523 register I32 len;
1524 I32 datumtype;
1525 SV *fromstr;
1526
1527We'll have another string pointer in there:
1528
1529 STRLEN fromlen;
1530 register char *pat = SvPVx(*++MARK, fromlen);
1531 register char *patend = pat + fromlen;
1532 + char *patcopy;
1533 register I32 len;
1534 I32 datumtype;
1535 SV *fromstr;
1536
1537And just before we start the loop, we'll set C<patcopy> to be the start
1538of C<pat>:
1539
1540 items = SP - MARK;
1541 MARK++;
1542 sv_setpvn(cat, "", 0);
1543 + patcopy = pat;
1544 while (pat < patend) {
1545
1546Now if we see a C<U> which was at the start of the string, we turn on
1e54db1a 1547the C<UTF8> flag for the output SV, C<cat>:
a422fd2d
SC
1548
1549 + if (datumtype == 'U' && pat==patcopy+1)
1550 + SvUTF8_on(cat);
1551 if (datumtype == '#') {
1552 while (pat < patend && *pat != '\n')
1553 pat++;
1554
1555Remember that it has to be C<patcopy+1> because the first character of
1556the string is the C<U> which has been swallowed into C<datumtype!>
1557
1558Oops, we forgot one thing: what if there are spaces at the start of the
1559pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1560character, even though it's not the first thing in the pattern. In this
1561case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1562
1563 if (isSPACE(datumtype))
1564 continue;
1565
1566needs to become
1567
1568 if (isSPACE(datumtype)) {
1569 patcopy++;
1570 continue;
1571 }
1572
1573OK. That's the C part done. Now we must do two additional things before
1574this patch is ready to go: we've changed the behaviour of Perl, and so
1575we must document that change. We must also provide some more regression
1576tests to make sure our patch works and doesn't create a bug somewhere
1577else along the line.
1578
b23b8711
MS
1579The regression tests for each operator live in F<t/op/>, and so we
1580make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1581tests to the end. First, we'll test that the C<U> does indeed create
07aa3531 1582Unicode strings.
b23b8711
MS
1583
1584t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1585use the one from t/test.pl.
b23b8711 1586
35c336e6
MS
1587 require './test.pl';
1588 plan( tests => 159 );
b23b8711
MS
1589
1590so instead of this:
a422fd2d
SC
1591
1592 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1593 print "ok $test\n"; $test++;
1594
35c336e6
MS
1595we can write the more sensible (see L<Test::More> for a full
1596explanation of is() and other testing functions).
b23b8711 1597
07aa3531 1598 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
38a44b82 1599 "U* produces Unicode" );
b23b8711 1600
a422fd2d
SC
1601Now we'll test that we got that space-at-the-beginning business right:
1602
35c336e6 1603 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1604 " with spaces at the beginning" );
a422fd2d
SC
1605
1606And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1607the first active format:
1608
35c336e6 1609 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
38a44b82 1610 "U* not first isn't Unicode" );
a422fd2d 1611
35c336e6
MS
1612Mustn't forget to change the number of tests which appears at the top,
1613or else the automated tester will get confused. This will either look
1614like this:
a422fd2d 1615
35c336e6
MS
1616 print "1..156\n";
1617
1618or this:
1619
1620 plan( tests => 156 );
a422fd2d
SC
1621
1622We now compile up Perl, and run it through the test suite. Our new
1623tests pass, hooray!
1624
1625Finally, the documentation. The job is never done until the paperwork is
1626over, so let's describe the change we've just made. The relevant place
1627is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1628this text in the description of C<pack>:
1629
1630 =item *
1631
1632 If the pattern begins with a C<U>, the resulting string will be treated
1e54db1a
JH
1633 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1634 with an initial C<U0>, and the bytes that follow will be interpreted as
1635 Unicode characters. If you don't want this to happen, you can begin your
1636 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
a422fd2d
SC
1637 string, and then follow this with a C<U*> somewhere in your pattern.
1638
1639All done. Now let's create the patch. F<Porting/patching.pod> tells us
1640that if we're making major changes, we should copy the entire directory
1641to somewhere safe before we begin fiddling, and then do
13a2d996 1642
a422fd2d
SC
1643 diff -ruN old new > patch
1644
1645However, we know which files we've changed, and we can simply do this:
1646
1647 diff -u pp.c~ pp.c > patch
1648 diff -u t/op/pack.t~ t/op/pack.t >> patch
1649 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1650
1651We end up with a patch looking a little like this:
1652
1653 --- pp.c~ Fri Jun 02 04:34:10 2000
1654 +++ pp.c Fri Jun 16 11:37:25 2000
1655 @@ -4375,6 +4375,7 @@
1656 register I32 items;
1657 STRLEN fromlen;
1658 register char *pat = SvPVx(*++MARK, fromlen);
1659 + char *patcopy;
1660 register char *patend = pat + fromlen;
1661 register I32 len;
1662 I32 datumtype;
1663 @@ -4405,6 +4406,7 @@
1664 ...
1665
1666And finally, we submit it, with our rationale, to perl5-porters. Job
1667done!
1668
f7e1e956
MS
1669=head2 Patching a core module
1670
1671This works just like patching anything else, with an extra
1672consideration. Many core modules also live on CPAN. If this is so,
1673patch the CPAN version instead of the core and send the patch off to
1674the module maintainer (with a copy to p5p). This will help the module
1675maintainer keep the CPAN version in sync with the core version without
1676constantly scanning p5p.
1677
db300100
RGS
1678The list of maintainers of core modules is usefully documented in
1679F<Porting/Maintainers.pl>.
1680
acbe17fc
JP
1681=head2 Adding a new function to the core
1682
1683If, as part of a patch to fix a bug, or just because you have an
1684especially good idea, you decide to add a new function to the core,
1685discuss your ideas on p5p well before you start work. It may be that
1686someone else has already attempted to do what you are considering and
1687can give lots of good advice or even provide you with bits of code
1688that they already started (but never finished).
1689
1690You have to follow all of the advice given above for patching. It is
1691extremely important to test any addition thoroughly and add new tests
1692to explore all boundary conditions that your new function is expected
1693to handle. If your new function is used only by one module (e.g. toke),
1694then it should probably be named S_your_function (for static); on the
210b36aa 1695other hand, if you expect it to accessible from other functions in
acbe17fc
JP
1696Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1697for more details.
1698
1699The location of any new code is also an important consideration. Don't
1700just create a new top level .c file and put your code there; you would
1701have to make changes to Configure (so the Makefile is created properly),
1702as well as possibly lots of include files. This is strictly pumpking
1703business.
1704
1705It is better to add your function to one of the existing top level
1706source code files, but your choice is complicated by the nature of
1707the Perl distribution. Only the files that are marked as compiled
1708static are located in the perl executable. Everything else is located
1709in the shared library (or DLL if you are running under WIN32). So,
1710for example, if a function was only used by functions located in
1711toke.c, then your code can go in toke.c. If, however, you want to call
1712the function from universal.c, then you should put your code in another
1713location, for example util.c.
1714
1715In addition to writing your c-code, you will need to create an
1716appropriate entry in embed.pl describing your function, then run
1717'make regen_headers' to create the entries in the numerous header
1718files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1719for information on the various options that you can set in embed.pl.
1720You will forget to do this a few (or many) times and you will get
1721warnings during the compilation phase. Make sure that you mention
1722this when you post your patch to P5P; the pumpking needs to know this.
1723
1724When you write your new code, please be conscious of existing code
884bad00 1725conventions used in the perl source files. See L<perlstyle> for
acbe17fc
JP
1726details. Although most of the guidelines discussed seem to focus on
1727Perl code, rather than c, they all apply (except when they don't ;).
1728See also I<Porting/patching.pod> file in the Perl source distribution
1729for lots of details about both formatting and submitting patches of
1730your changes.
1731
1732Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1733Test on as many platforms as you can find. Test as many perl
1734Configure options as you can (e.g. MULTIPLICITY). If you have
1735profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 1736below for how to use them to further test your code. Remember that
acbe17fc
JP
1737most of the people on P5P are doing this on their own time and
1738don't have the time to debug your code.
f7e1e956
MS
1739
1740=head2 Writing a test
1741
1742Every module and built-in function has an associated test file (or
1743should...). If you add or change functionality, you have to write a
1744test. If you fix a bug, you have to write a test so that bug never
1745comes back. If you alter the docs, it would be nice to test what the
1746new documentation says.
1747
1748In short, if you submit a patch you probably also have to patch the
1749tests.
1750
1751For modules, the test file is right next to the module itself.
1752F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1753so there are some snags (and it would be wonderful for you to brush
1754them out), but it basically works that way. Everything else lives in
1755F<t/>.
1756
1757=over 3
1758
1759=item F<t/base/>
1760
1761Testing of the absolute basic functionality of Perl. Things like
1762C<if>, basic file reads and writes, simple regexes, etc. These are
1763run first in the test suite and if any of them fail, something is
1764I<really> broken.
1765
1766=item F<t/cmd/>
1767
1768These test the basic control structures, C<if/else>, C<while>,
35c336e6 1769subroutines, etc.
f7e1e956
MS
1770
1771=item F<t/comp/>
1772
1773Tests basic issues of how Perl parses and compiles itself.
1774
1775=item F<t/io/>
1776
1777Tests for built-in IO functions, including command line arguments.
1778
1779=item F<t/lib/>
1780
1781The old home for the module tests, you shouldn't put anything new in
1782here. There are still some bits and pieces hanging around in here
1783that need to be moved. Perhaps you could move them? Thanks!
1784
3c295041
RGS
1785=item F<t/mro/>
1786
0503309d 1787Tests for perl's method resolution order implementations
3c295041
RGS
1788(see L<mro>).
1789
f7e1e956
MS
1790=item F<t/op/>
1791
1792Tests for perl's built in functions that don't fit into any of the
1793other directories.
1794
1795=item F<t/pod/>
1796
1797Tests for POD directives. There are still some tests for the Pod
1798modules hanging around in here that need to be moved out into F<lib/>.
1799
1800=item F<t/run/>
1801
1802Testing features of how perl actually runs, including exit codes and
1803handling of PERL* environment variables.
1804
244d9cb7
RGS
1805=item F<t/uni/>
1806
1807Tests for the core support of Unicode.
1808
1809=item F<t/win32/>
1810
1811Windows-specific tests.
1812
1813=item F<t/x2p>
1814
1815A test suite for the s2p converter.
1816
f7e1e956
MS
1817=back
1818
1819The core uses the same testing style as the rest of Perl, a simple
1820"ok/not ok" run through Test::Harness, but there are a few special
1821considerations.
1822
35c336e6
MS
1823There are three ways to write a test in the core. Test::More,
1824t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1825decision of which to use depends on what part of the test suite you're
1826working on. This is a measure to prevent a high-level failure (such
1827as Config.pm breaking) from causing basic functionality tests to fail.
1828
07aa3531 1829=over 4
35c336e6
MS
1830
1831=item t/base t/comp
1832
1833Since we don't know if require works, or even subroutines, use ad hoc
1834tests for these two. Step carefully to avoid using the feature being
1835tested.
1836
1837=item t/cmd t/run t/io t/op
1838
1839Now that basic require() and subroutines are tested, you can use the
1840t/test.pl library which emulates the important features of Test::More
1841while using a minimum of core features.
1842
1843You can also conditionally use certain libraries like Config, but be
1844sure to skip the test gracefully if it's not there.
1845
1846=item t/lib ext lib
1847
1848Now that the core of Perl is tested, Test::More can be used. You can
1849also use the full suite of core modules in the tests.
1850
1851=back
f7e1e956
MS
1852
1853When you say "make test" Perl uses the F<t/TEST> program to run the
07aa3531
JC
1854test suite (except under Win32 where it uses F<t/harness> instead.)
1855All tests are run from the F<t/> directory, B<not> the directory
1856which contains the test. This causes some problems with the tests
7205a85d 1857in F<lib/>, so here's some opportunity for some patching.
f7e1e956
MS
1858
1859You must be triply conscious of cross-platform concerns. This usually
1860boils down to using File::Spec and avoiding things like C<fork()> and
1861C<system()> unless absolutely necessary.
1862
e018f8be
JH
1863=head2 Special Make Test Targets
1864
1865There are various special make targets that can be used to test Perl
1866slightly differently than the standard "test" target. Not all them
1867are expected to give a 100% success rate. Many of them have several
7205a85d
YO
1868aliases, and many of them are not available on certain operating
1869systems.
e018f8be
JH
1870
1871=over 4
1872
1873=item coretest
1874
7d7d5695 1875Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
e018f8be 1876
7205a85d
YO
1877(Not available on Win32)
1878
e018f8be
JH
1879=item test.deparse
1880
b26492ee
RGS
1881Run all the tests through B::Deparse. Not all tests will succeed.
1882
7205a85d
YO
1883(Not available on Win32)
1884
b26492ee
RGS
1885=item test.taintwarn
1886
1887Run all tests with the B<-t> command-line switch. Not all tests
1888are expected to succeed (until they're specifically fixed, of course).
e018f8be 1889
7205a85d
YO
1890(Not available on Win32)
1891
e018f8be
JH
1892=item minitest
1893
1894Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
8cebccf4 1895F<t/op>, F<t/uni> and F<t/mro> tests.
e018f8be 1896
7a834142
JH
1897=item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1898
1899(Only in Linux) Run all the tests using the memory leak + naughty
1900memory access tool "valgrind". The log files will be named
1901F<testname.valgrind>.
1902
e018f8be
JH
1903=item test.third check.third utest.third ucheck.third
1904
1905(Only in Tru64) Run all the tests using the memory leak + naughty
1906memory access tool "Third Degree". The log files will be named
60a57c1c 1907F<perl.3log.testname>.
e018f8be
JH
1908
1909=item test.torture torturetest
1910
1911Run all the usual tests and some extra tests. As of Perl 5.8.0 the
244d9cb7 1912only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
e018f8be
JH
1913
1914You can also run the torture test with F<t/harness> by giving
1915C<-torture> argument to F<t/harness>.
1916
1917=item utest ucheck test.utf8 check.utf8
1918
1919Run all the tests with -Mutf8. Not all tests will succeed.
1920
7205a85d
YO
1921(Not available on Win32)
1922
cc0710ff
RGS
1923=item minitest.utf16 test.utf16
1924
1925Runs the tests with UTF-16 encoded scripts, encoded with different
1926versions of this encoding.
1927
1928C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
1929C<-utf16> arguments to F<t/TEST>.
1930
7205a85d
YO
1931(Not available on Win32)
1932
244d9cb7
RGS
1933=item test_harness
1934
1935Run the test suite with the F<t/harness> controlling program, instead of
1936F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1937L<Test::Harness> module, thus using this test target supposes that perl
1938mostly works. The main advantage for our purposes is that it prints a
00bf5cd9
RGS
1939detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1940doesn't redirect stderr to stdout.
244d9cb7 1941
7205a85d
YO
1942Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
1943there is no special "test_harness" target.
1944
1945Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
1946environment variables to control the behaviour of F<t/harness>. This means
1947you can say
1948
1949 nmake test TEST_FILES="op/*.t"
1950 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
1951
1952=item test-notty test_notty
1953
1954Sets PERL_SKIP_TTY_TEST to true before running normal test.
1955
244d9cb7
RGS
1956=back
1957
1958=head2 Running tests by hand
1959
1960You can run part of the test suite by hand by using one the following
1961commands from the F<t/> directory :
1962
1963 ./perl -I../lib TEST list-of-.t-files
1964
1965or
1966
1967 ./perl -I../lib harness list-of-.t-files
1968
1969(if you don't specify test scripts, the whole test suite will be run.)
1970
7205a85d
YO
1971=head3 Using t/harness for testing
1972
1973If you use C<harness> for testing you have several command line options
1974available to you. The arguments are as follows, and are in the order
1975that they must appear if used together.
1976
1977 harness -v -torture -re=pattern LIST OF FILES TO TEST
1978 harness -v -torture -re LIST OF PATTERNS TO MATCH
1979
1980If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
07aa3531 1981the manifest. The file list may include shell wildcards which will be
7205a85d
YO
1982expanded out.
1983
1984=over 4
1985
1986=item -v
1987
07aa3531 1988Run the tests under verbose mode so you can see what tests were run,
8550bf48 1989and debug output.
7205a85d
YO
1990
1991=item -torture
1992
1993Run the torture tests as well as the normal set.
1994
1995=item -re=PATTERN
1996
1997Filter the file list so that all the test files run match PATTERN.
1998Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
1999in that it allows the file list to be provided as well.
2000
2001=item -re LIST OF PATTERNS
2002
07aa3531 2003Filter the file list so that all the test files run match
7205a85d
YO
2004/(LIST|OF|PATTERNS)/. Note that with this form the patterns
2005are joined by '|' and you cannot supply a list of files, instead
2006the test files are obtained from the MANIFEST.
2007
2008=back
2009
244d9cb7
RGS
2010You can run an individual test by a command similar to
2011
2012 ./perl -I../lib patho/to/foo.t
2013
2014except that the harnesses set up some environment variables that may
2015affect the execution of the test :
2016
07aa3531 2017=over 4
244d9cb7
RGS
2018
2019=item PERL_CORE=1
2020
2021indicates that we're running this test part of the perl core test suite.
2022This is useful for modules that have a dual life on CPAN.
2023
2024=item PERL_DESTRUCT_LEVEL=2
2025
2026is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2027
2028=item PERL
2029
2030(used only by F<t/TEST>) if set, overrides the path to the perl executable
2031that should be used to run the tests (the default being F<./perl>).
2032
2033=item PERL_SKIP_TTY_TEST
2034
2035if set, tells to skip the tests that need a terminal. It's actually set
2036automatically by the Makefile, but can also be forced artificially by
2037running 'make test_notty'.
2038
e018f8be 2039=back
f7e1e956 2040
7cd58830
RGS
2041=head3 Other environment variables that may influence tests
2042
2043=over 4
2044
2045=item PERL_TEST_Net_Ping
2046
2047Setting this variable runs all the Net::Ping modules tests,
2048otherwise some tests that interact with the outside world are skipped.
2049See L<perl58delta>.
2050
2051=item PERL_TEST_NOVREXX
2052
2053Setting this variable skips the vrexx.t tests for OS2::REXX.
2054
2055=item PERL_TEST_NUMCONVERTS
2056
2057This sets a variable in op/numconvert.t.
2058
2059=back
2060
2061See also the documentation for the Test and Test::Harness modules,
2062for more environment variables that affect testing.
2063
d7889f52
JH
2064=head2 Common problems when patching Perl source code
2065
2066Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2067some cases we have to take pre-ANSI requirements into consideration.
2068You don't care about some particular platform having broken Perl?
2069I hear there is still a strong demand for J2EE programmers.
2070
2071=head2 Perl environment problems
2072
2073=over 4
2074
2075=item *
2076
2077Not compiling with threading
2078
2079Compiling with threading (-Duseithreads) completely rewrites
2080the function prototypes of Perl. You better try your changes
0bec6c03 2081with that. Related to this is the difference between "Perl_-less"
d7889f52
JH
2082and "Perl_-ly" APIs, for example:
2083
2084 Perl_sv_setiv(aTHX_ ...);
2085 sv_setiv(...);
2086
ee9468a2
RGS
2087The first one explicitly passes in the context, which is needed for e.g.
2088threaded builds. The second one does that implicitly; do not get them
def4ed7d
JH
2089mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2090(or a dVAR) as the first thing in the function.
d7889f52
JH
2091
2092See L<perlguts/"How multiple interpreters and concurrency are supported">
2093for further discussion about context.
2094
2095=item *
2096
2097Not compiling with -DDEBUGGING
2098
2099The DEBUGGING define exposes more code to the compiler,
0bec6c03 2100therefore more ways for things to go wrong. You should try it.
d7889f52
JH
2101
2102=item *
2103
ee9468a2
RGS
2104Introducing (non-read-only) globals
2105
2106Do not introduce any modifiable globals, truly global or file static.
bc028b6b
JH
2107They are bad form and complicate multithreading and other forms of
2108concurrency. The right way is to introduce them as new interpreter
2109variables, see F<intrpvar.h> (at the very end for binary compatibility).
ee9468a2
RGS
2110
2111Introducing read-only (const) globals is okay, as long as you verify
2112with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2113BSD-style output) that the data you added really is read-only.
2114(If it is, it shouldn't show up in the output of that command.)
2115
2116If you want to have static strings, make them constant:
2117
2118 static const char etc[] = "...";
2119
bc028b6b 2120If you want to have arrays of constant strings, note carefully
ee9468a2
RGS
2121the right combination of C<const>s:
2122
2123 static const char * const yippee[] =
2124 {"hi", "ho", "silver"};
2125
bc028b6b
JH
2126There is a way to completely hide any modifiable globals (they are all
2127moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2128It is not normally used, but can be used for testing, read more
def4ed7d 2129about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
bc028b6b 2130
ee9468a2
RGS
2131=item *
2132
d7889f52
JH
2133Not exporting your new function
2134
2135Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2136function that is part of the public API (the shared Perl library)
2137to be explicitly marked as exported. See the discussion about
2138F<embed.pl> in L<perlguts>.
2139
2140=item *
2141
2142Exporting your new function
2143
2144The new shiny result of either genuine new functionality or your
2145arduous refactoring is now ready and correctly exported. So what
def4ed7d 2146could possibly go wrong?
d7889f52
JH
2147
2148Maybe simply that your function did not need to be exported in the
2149first place. Perl has a long and not so glorious history of exporting
2150functions that it should not have.
2151
2152If the function is used only inside one source code file, make it
2153static. See the discussion about F<embed.pl> in L<perlguts>.
2154
2155If the function is used across several files, but intended only for
2156Perl's internal use (and this should be the common case), do not
2157export it to the public API. See the discussion about F<embed.pl>
2158in L<perlguts>.
2159
2160=back
2161
5b38b9cd 2162=head2 Portability problems
d7889f52
JH
2163
2164The following are common causes of compilation and/or execution
2165failures, not common to Perl as such. The C FAQ is good bedtime
0bec6c03
JH
2166reading. Please test your changes with as many C compilers and
2167platforms as possible -- we will, anyway, and it's nice to save
2168oneself from public embarrassment.
2169
9aaf14db
RGS
2170If using gcc, you can add the C<-std=c89> option which will hopefully
2171catch most of these unportabilities. (However it might also catch
2172incompatibilities in your system's header files.)
d1307786 2173
a8e98a71
JH
2174Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2175C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2176
def4ed7d
JH
2177If using the C<gcc -Wall> note that not all the possible warnings
2178(like C<-Wunitialized>) are given unless you also compile with C<-O>.
2179
2180Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2181code files (the ones at the top level of the source code distribution,
2182but not e.g. the extensions under ext/) are automatically compiled
2183with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2184and a selection of C<-W> flags (see cflags.SH).
27565cb6 2185
0bec6c03 2186Also study L<perlport> carefully to avoid any bad assumptions
def4ed7d 2187about the operating system, filesystems, and so forth.
0bec6c03 2188
606fd33d 2189You may once in a while try a "make microperl" to see whether we
63796a85 2190can still compile Perl with just the bare minimum of interfaces.
606fd33d 2191(See README.micro.)
ee9468a2 2192
0bec6c03 2193Do not assume an operating system indicates a certain compiler.
d7889f52
JH
2194
2195=over 4
2196
2197=item *
2198
2199Casting pointers to integers or casting integers to pointers
2200
2201 void castaway(U8* p)
2202 {
2203 IV i = p;
2204
2205or
2206
2207 void castaway(U8* p)
2208 {
2209 IV i = (IV)p;
2210
ee9468a2 2211Both are bad, and broken, and unportable. Use the PTR2IV()
d7889f52
JH
2212macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2213INT2PTR(), and NUM2PTR().)
2214
2215=item *
2216
0bec6c03
JH
2217Casting between data function pointers and data pointers
2218
d7889f52
JH
2219Technically speaking casting between function pointers and data
2220pointers is unportable and undefined, but practically speaking
2221it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
0bec6c03 2222macros. Sometimes you can also play games with unions.
d7889f52
JH
2223
2224=item *
2225
2226Assuming sizeof(int) == sizeof(long)
2227
2228There are platforms where longs are 64 bits, and platforms where ints
2229are 64 bits, and while we are out to shock you, even platforms where
2230shorts are 64 bits. This is all legal according to the C standard.
2231(In other words, "long long" is not a portable way to specify 64 bits,
2232and "long long" is not even guaranteed to be any wider than "long".)
63796a85
JH
2233
2234Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2235Avoid things like I32 because they are B<not> guaranteed to be
2236I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2237guaranteed to be B<int> or B<long>. If you really explicitly need
223864-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
d7889f52
JH
2239
2240=item *
2241
2242Assuming one can dereference any type of pointer for any type of data
2243
2244 char *p = ...;
def4ed7d 2245 long pony = *p; /* BAD */
d7889f52
JH
2246
2247Many platforms, quite rightly so, will give you a core dump instead
2248of a pony if the p happens not be correctly aligned.
2249
2250=item *
2251
2252Lvalue casts
2253
def4ed7d 2254 (int)*p = ...; /* BAD */
d7889f52
JH
2255
2256Simply not portable. Get your lvalue to be of the right type,
27565cb6
JH
2257or maybe use temporary variables, or dirty tricks with unions.
2258
2259=item *
2260
606fd33d
JH
2261Assume B<anything> about structs (especially the ones you
2262don't control, like the ones coming from the system headers)
27565cb6
JH
2263
2264=over 8
2265
2266=item *
2267
2268That a certain field exists in a struct
2269
2270=item *
2271
902821cc 2272That no other fields exist besides the ones you know of
27565cb6
JH
2273
2274=item *
2275
606fd33d 2276That a field is of certain signedness, sizeof, or type
27565cb6
JH
2277
2278=item *
2279
2280That the fields are in a certain order
2281
606fd33d
JH
2282=over 8
2283
27565cb6
JH
2284=item *
2285
606fd33d
JH
2286While C guarantees the ordering specified in the struct definition,
2287between different platforms the definitions might differ
2288
2289=back
27565cb6
JH
2290
2291=item *
2292
606fd33d
JH
2293That the sizeof(struct) or the alignments are the same everywhere
2294
2295=over 8
27565cb6
JH
2296
2297=item *
2298
606fd33d
JH
2299There might be padding bytes between the fields to align the fields -
2300the bytes can be anything
2301
2302=item *
2303
2304Structs are required to be aligned to the maximum alignment required
2305by the fields - which for native types is for usually equivalent to
2306sizeof() of the field
2307
2308=back
27565cb6
JH
2309
2310=back
d7889f52
JH
2311
2312=item *
2313
2bbc8d55
SP
2314Assuming the character set is ASCIIish
2315
2316Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
2317This is transparent for the most part, but because the character sets
2318differ, you shouldn't use numeric (decimal, octal, nor hex) constants
2319to refer to characters. You can safely say 'A', but not 0x41.
2320You can safely say '\n', but not \012.
2321If a character doesn't have a trivial input form, you can
2322create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that
2323it resolves to different values depending on the character set being used.
2324(There are three different EBCDIC character sets defined in C<utfebcdic.h>,
2325so it might be best to insert the #define three times in that file.)
2326
2327Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case
2328alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'.
2329But '0' - '9' is an unbroken range in both systems. Don't assume anything
2330about other ranges.
2331
2332Many of the comments in the existing code ignore the possibility of EBCDIC,
2333and may be wrong therefore, even if the code works.
2334This is actually a tribute to the successful transparent insertion of being
fe749c9a 2335able to handle EBCDIC without having to change pre-existing code.
2bbc8d55
SP
2336
2337UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode
2338code points as sequences of bytes. Macros
2339with the same names (but different definitions)
2340in C<utf8.h> and C<utfebcdic.h>
fe749c9a
KW
2341are used to allow the calling code to think that there is only one such
2342encoding.
2343This is almost always referred to as C<utf8>, but it means the EBCDIC version
2344as well. Again, comments in the code may well be wrong even if the code itself
2345is right.
2bbc8d55
SP
2346For example, the concept of C<invariant characters> differs between ASCII and
2347EBCDIC.
2348On ASCII platforms, only characters that do not have the high-order
2349bit set (i.e. whose ordinals are strict ASCII, 0 - 127)
2350are invariant, and the documentation and comments in the code
2351may assume that,
2352often referring to something like, say, C<hibit>.
2353The situation differs and is not so simple on EBCDIC machines, but as long as
2354the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it
2355works, even if the comments are wrong.
2356
2357=item *
2358
2359Assuming the character set is just ASCII
2360
2361ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
2362characters have different meanings depending on the locale. Absent a locale,
2363currently these extra characters are generally considered to be unassigned,
2364and this has presented some problems.
2365This is scheduled to be changed in 5.12 so that these characters will
2366be considered to be Latin-1 (ISO-8859-1).
2367
2368=item *
2369
0bec6c03
JH
2370Mixing #define and #ifdef
2371
2372 #define BURGLE(x) ... \
def4ed7d 2373 #ifdef BURGLE_OLD_STYLE /* BAD */
0bec6c03
JH
2374 ... do it the old way ... \
2375 #else
2376 ... do it the new way ... \
2377 #endif
2378
ee9468a2
RGS
2379You cannot portably "stack" cpp directives. For example in the above
2380you need two separate BURGLE() #defines, one for each #ifdef branch.
2381
2382=item *
2383
2bbc8d55 2384Adding non-comment stuff after #endif or #else
ee9468a2
RGS
2385
2386 #ifdef SNOSH
2387 ...
def4ed7d 2388 #else !SNOSH /* BAD */
ee9468a2 2389 ...
def4ed7d 2390 #endif SNOSH /* BAD */
ee9468a2 2391
def4ed7d
JH
2392The #endif and #else cannot portably have anything non-comment after
2393them. If you want to document what is going (which is a good idea
2394especially if the branches are long), use (C) comments:
ee9468a2
RGS
2395
2396 #ifdef SNOSH
2397 ...
2398 #else /* !SNOSH */
2399 ...
2400 #endif /* SNOSH */
2401
2402The gcc option C<-Wendif-labels> warns about the bad variant
2403(by default on starting from Perl 5.9.4).
0bec6c03
JH
2404
2405=item *
2406
27565cb6
JH
2407Having a comma after the last element of an enum list
2408
2409 enum color {
2410 CERULEAN,
2411 CHARTREUSE,
def4ed7d 2412 CINNABAR, /* BAD */
27565cb6
JH
2413 };
2414
2415is not portable. Leave out the last comma.
2416
2417Also note that whether enums are implicitly morphable to ints
2418varies between compilers, you might need to (int).
2419
2420=item *
2421
d7889f52
JH
2422Using //-comments
2423
def4ed7d 2424 // This function bamfoodles the zorklator. /* BAD */
d7889f52
JH
2425
2426That is C99 or C++. Perl is C89. Using the //-comments is silently
0bec6c03
JH
2427allowed by many C compilers but cranking up the ANSI C89 strictness
2428(which we like to do) causes the compilation to fail.
d7889f52
JH
2429
2430=item *
2431
2432Mixing declarations and code
2433
2434 void zorklator()
2435 {
2436 int n = 3;
def4ed7d 2437 set_zorkmids(n); /* BAD */
d7889f52
JH
2438 int q = 4;
2439
0bec6c03
JH
2440That is C99 or C++. Some C compilers allow that, but you shouldn't.
2441
63796a85
JH
2442The gcc option C<-Wdeclaration-after-statements> scans for such problems
2443(by default on starting from Perl 5.9.4).
2444
0bec6c03
JH
2445=item *
2446
2447Introducing variables inside for()
2448
def4ed7d 2449 for(int i = ...; ...; ...) { /* BAD */
0bec6c03
JH
2450
2451That is C99 or C++. While it would indeed be awfully nice to have that
2452also in C89, to limit the scope of the loop variable, alas, we cannot.
d7889f52
JH
2453
2454=item *
2455
2456Mixing signed char pointers with unsigned char pointers
2457
2458 int foo(char *s) { ... }
2459 ...
2460 unsigned char *t = ...; /* Or U8* t = ... */
def4ed7d 2461 foo(t); /* BAD */
d7889f52
JH
2462
2463While this is legal practice, it is certainly dubious, and downright
2464fatal in at least one platform: for example VMS cc considers this a
def4ed7d
JH
2465fatal error. One cause for people often making this mistake is that a
2466"naked char" and therefore dereferencing a "naked char pointer" have
2467an undefined signedness: it depends on the compiler and the flags of
2468the compiler and the underlying platform whether the result is signed
2469or unsigned. For this very same reason using a 'char' as an array
2470index is bad.
d7889f52
JH
2471
2472=item *
2473
2474Macros that have string constants and their arguments as substrings of
2475the string constants
2476
def4ed7d 2477 #define FOO(n) printf("number = %d\n", n) /* BAD */
d7889f52
JH
2478 FOO(10);
2479
2480Pre-ANSI semantics for that was equivalent to
2481
2482 printf("10umber = %d\10");
2483
0bec6c03
JH
2484which is probably not what you were expecting. Unfortunately at least
2485one reasonably common and modern C compiler does "real backward
63796a85 2486compatibility" here, in AIX that is what still happens even though the
0bec6c03
JH
2487rest of the AIX compiler is very happily C89.
2488
2489=item *
2490
ee9468a2
RGS
2491Using printf formats for non-basic C types
2492
2493 IV i = ...;
def4ed7d 2494 printf("i = %d\n", i); /* BAD */
ee9468a2
RGS
2495
2496While this might by accident work in some platform (where IV happens
2497to be an C<int>), in general it cannot. IV might be something larger.
2498Even worse the situation is with more specific types (defined by Perl's
2499configuration step in F<config.h>):
2500
2501 Uid_t who = ...;
def4ed7d 2502 printf("who = %d\n", who); /* BAD */
ee9468a2
RGS
2503
2504The problem here is that Uid_t might be not only not C<int>-wide
2505but it might also be unsigned, in which case large uids would be
2506printed as negative values.
2507
2508There is no simple solution to this because of printf()'s limited
2509intelligence, but for many types the right format is available as
2510with either 'f' or '_f' suffix, for example:
2511
2512 IVdf /* IV in decimal */
2513 UVxf /* UV is hexadecimal */
2514
2515 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2516
2517 Uid_t_f /* Uid_t in decimal */
2518
2519 printf("who = %"Uid_t_f"\n", who);
2520
63796a85
JH
2521Or you can try casting to a "wide enough" type:
2522
2523 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2524
2525Also remember that the C<%p> format really does require a void pointer:
2526
2527 U8* p = ...;
2528 printf("p = %p\n", (void*)p);
2529
ee9468a2
RGS
2530The gcc option C<-Wformat> scans for such problems.
2531
2532=item *
2533
0bec6c03
JH
2534Blindly using variadic macros
2535
63796a85
JH
2536gcc has had them for a while with its own syntax, and C99 brought
2537them with a standardized syntax. Don't use the former, and use
2538the latter only if the HAS_C99_VARIADIC_MACROS is defined.
0bec6c03
JH
2539
2540=item *
2541
2542Blindly passing va_list
2543
2544Not all platforms support passing va_list to further varargs (stdarg)
2545functions. The right thing to do is to copy the va_list using the
2546Perl_va_copy() if the NEED_VA_COPY is defined.
d7889f52 2547
ee9468a2
RGS
2548=item *
2549
e5afc1ae 2550Using gcc statement expressions
63796a85 2551
def4ed7d 2552 val = ({...;...;...}); /* BAD */
63796a85 2553
def4ed7d
JH
2554While a nice extension, it's not portable. The Perl code does
2555admittedly use them if available to gain some extra speed
2556(essentially as a funky form of inlining), but you shouldn't.
63796a85
JH
2557
2558=item *
2559
2bbc8d55 2560Binding together several statements in a macro
63796a85
JH
2561
2562Use the macros STMT_START and STMT_END.
2563
2564 STMT_START {
2565 ...
2566 } STMT_END
2567
2568=item *
2569
ee9468a2
RGS
2570Testing for operating systems or versions when should be testing for features
2571
def4ed7d 2572 #ifdef __FOONIX__ /* BAD */
ee9468a2
RGS
2573 foo = quux();
2574 #endif
2575
2576Unless you know with 100% certainty that quux() is only ever available
2577for the "Foonix" operating system B<and> that is available B<and>
2578correctly working for B<all> past, present, B<and> future versions of
2579"Foonix", the above is very wrong. This is more correct (though still
2580not perfect, because the below is a compile-time check):
2581
2582 #ifdef HAS_QUUX
2583 foo = quux();
2584 #endif
2585
def4ed7d 2586How does the HAS_QUUX become defined where it needs to be? Well, if
353c6505 2587Foonix happens to be UNIXy enough to be able to run the Configure
ee9468a2
RGS
2588script, and Configure has been taught about detecting and testing
2589quux(), the HAS_QUUX will be correctly defined. In other platforms,
2590the corresponding configuration step will hopefully do the same.
2591
2592In a pinch, if you cannot wait for Configure to be educated,
2593or if you have a good hunch of where quux() might be available,
2594you can temporarily try the following:
2595
2596 #if (defined(__FOONIX__) || defined(__BARNIX__))
2597 # define HAS_QUUX
2598 #endif
2599
2600 ...
2601
2602 #ifdef HAS_QUUX
2603 foo = quux();
2604 #endif
2605
2606But in any case, try to keep the features and operating systems separate.
2607
d7889f52
JH
2608=back
2609
ad7244db
JH
2610=head2 Problematic System Interfaces
2611
2612=over 4
2613
2614=item *
2615
353c6505 2616malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
ad7244db
JH
2617allocate at least one byte. (In general you should rarely need to
2618work at this low level, but instead use the various malloc wrappers.)
2619
2620=item *
2621
2622snprintf() - the return type is unportable. Use my_snprintf() instead.
2623
2624=back
2625
d7889f52
JH
2626=head2 Security problems
2627
2628Last but not least, here are various tips for safer coding.
2629
2630=over 4
2631
2632=item *
2633
2634Do not use gets()
2635
2636Or we will publicly ridicule you. Seriously.
2637
2638=item *
2639
d1307786 2640Do not use strcpy() or strcat() or strncpy() or strncat()
d7889f52 2641
d1307786
JH
2642Use my_strlcpy() and my_strlcat() instead: they either use the native
2643implementation, or Perl's own implementation (borrowed from the public
2644domain implementation of INN).
d7889f52
JH
2645
2646=item *
2647
2648Do not use sprintf() or vsprintf()
2649
0bec6c03 2650If you really want just plain byte strings, use my_snprintf()
64d9b66b 2651and my_vsnprintf() instead, which will try to use snprintf() and
0bec6c03
JH
2652vsnprintf() if those safer APIs are available. If you want something
2653fancier than a plain byte string, use SVs and Perl_sv_catpvf().
d7889f52
JH
2654
2655=back
2656
902b9dbf
MF
2657=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2658
2659Sometimes it helps to use external tools while debugging and
2660testing Perl. This section tries to guide you through using
2661some common testing and debugging tools with Perl. This is
2662meant as a guide to interfacing these tools with Perl, not
2663as any kind of guide to the use of the tools themselves.
2664
a958818a
JH
2665B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2666Third Degree greatly slows down the execution: seconds become minutes,
2667minutes become hours. For example as of Perl 5.8.1, the
2668ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2669e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2670than six hours, even on a snappy computer-- the said test must be
2671doing something that is quite unfriendly for memory debuggers. If you
2672don't feel like waiting, that you can simply kill away the perl
2673process.
2674
2675B<NOTE 2>: To minimize the number of memory leak false alarms (see
2676L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2677environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2678and harness scripts do that automatically. But if you are running
2679some of the tests manually-- for csh-like shells:
2680
2681 setenv PERL_DESTRUCT_LEVEL 2
2682
2683and for Bourne-type shells:
2684
2685 PERL_DESTRUCT_LEVEL=2
2686 export PERL_DESTRUCT_LEVEL
2687
2688or in UNIXy environments you can also use the C<env> command:
2689
2690 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
a1b65709 2691
37c0adeb
JH
2692B<NOTE 3>: There are known memory leaks when there are compile-time
2693errors within eval or require, seeing C<S_doeval> in the call stack
2694is a good sign of these. Fixing these leaks is non-trivial,
2695unfortunately, but they must be fixed eventually.
2696
f50e5b73
MH
2697B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2698unless Perl is built with the Configure option
2699C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2700
902b9dbf
MF
2701=head2 Rational Software's Purify
2702
2703Purify is a commercial tool that is helpful in identifying
2704memory overruns, wild pointers, memory leaks and other such
2705badness. Perl must be compiled in a specific way for
2706optimal testing with Purify. Purify is available under
2707Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2708
902b9dbf
MF
2709=head2 Purify on Unix
2710
2711On Unix, Purify creates a new Perl binary. To get the most
2712benefit out of Purify, you should create the perl to Purify
2713using:
2714
2715 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2716 -Uusemymalloc -Dusemultiplicity
2717
2718where these arguments mean:
2719
2720=over 4
2721
2722=item -Accflags=-DPURIFY
2723
2724Disables Perl's arena memory allocation functions, as well as
2725forcing use of memory allocation functions derived from the
2726system malloc.
2727
2728=item -Doptimize='-g'
2729
2730Adds debugging information so that you see the exact source
2731statements where the problem occurs. Without this flag, all
2732you will see is the source filename of where the error occurred.
2733
2734=item -Uusemymalloc
2735
2736Disable Perl's malloc so that Purify can more closely monitor
2737allocations and leaks. Using Perl's malloc will make Purify
2738report most leaks in the "potential" leaks category.
2739
2740=item -Dusemultiplicity
2741
2742Enabling the multiplicity option allows perl to clean up
2743thoroughly when the interpreter shuts down, which reduces the
2744number of bogus leak reports from Purify.
2745
2746=back
2747
2748Once you've compiled a perl suitable for Purify'ing, then you
2749can just:
2750
07aa3531 2751 make pureperl
902b9dbf
MF
2752
2753which creates a binary named 'pureperl' that has been Purify'ed.
2754This binary is used in place of the standard 'perl' binary
2755when you want to debug Perl memory problems.
2756
2757As an example, to show any memory leaks produced during the
2758standard Perl testset you would create and run the Purify'ed
2759perl as:
2760
2761 make pureperl
2762 cd t
07aa3531 2763 ../pureperl -I../lib harness
902b9dbf
MF
2764
2765which would run Perl on test.pl and report any memory problems.
2766
2767Purify outputs messages in "Viewer" windows by default. If
2768you don't have a windowing environment or if you simply
2769want the Purify output to unobtrusively go to a log file
2770instead of to the interactive window, use these following
2771options to output to the log file "perl.log":
2772
2773 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2774 -log-file=perl.log -append-logfile=yes"
2775
2776If you plan to use the "Viewer" windows, then you only need this option:
2777
2778 setenv PURIFYOPTIONS "-chain-length=25"
2779
c406981e
JH
2780In Bourne-type shells:
2781
98631ff8
JL
2782 PURIFYOPTIONS="..."
2783 export PURIFYOPTIONS
c406981e
JH
2784
2785or if you have the "env" utility:
2786
98631ff8 2787 env PURIFYOPTIONS="..." ../pureperl ...
c406981e 2788
902b9dbf
MF
2789=head2 Purify on NT
2790
2791Purify on Windows NT instruments the Perl binary 'perl.exe'
2792on the fly. There are several options in the makefile you
2793should change to get the most use out of Purify:
2794
2795=over 4
2796
2797=item DEFINES
2798
2799You should add -DPURIFY to the DEFINES line so the DEFINES
2800line looks something like:
2801
07aa3531 2802 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
902b9dbf
MF
2803
2804to disable Perl's arena memory allocation functions, as
2805well as to force use of memory allocation functions derived
2806from the system malloc.
2807
2808=item USE_MULTI = define
2809
2810Enabling the multiplicity option allows perl to clean up
2811thoroughly when the interpreter shuts down, which reduces the
2812number of bogus leak reports from Purify.
2813
2814=item #PERL_MALLOC = define
2815
2816Disable Perl's malloc so that Purify can more closely monitor
2817allocations and leaks. Using Perl's malloc will make Purify
2818report most leaks in the "potential" leaks category.
2819
2820=item CFG = Debug
2821
2822Adds debugging information so that you see the exact source
2823statements where the problem occurs. Without this flag, all
2824you will see is the source filename of where the error occurred.
2825
2826=back
2827
2828As an example, to show any memory leaks produced during the
2829standard Perl testset you would create and run Purify as:
2830
2831 cd win32
2832 make
2833 cd ../t
07aa3531 2834 purify ../perl -I../lib harness
902b9dbf
MF
2835
2836which would instrument Perl in memory, run Perl on test.pl,
2837then finally report any memory problems.
2838
7a834142
JH
2839=head2 valgrind
2840
2841The excellent valgrind tool can be used to find out both memory leaks
9df8f87f
LB
2842and illegal memory accesses. As of version 3.3.0, Valgrind only
2843supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind"
2844target can be used to run the tests under valgrind. Found errors
2845and memory leaks are logged in files named F<testfile.valgrind>.
07aa3531
JC
2846
2847Valgrind also provides a cachegrind tool, invoked on perl as:
2848
038c294a 2849 VG_OPTS=--tool=cachegrind make test.valgrind
d44161bf
MHM
2850
2851As system libraries (most notably glibc) are also triggering errors,
2852valgrind allows to suppress such errors using suppression files. The
2853default suppression file that comes with valgrind already catches a lot
2854of them. Some additional suppressions are defined in F<t/perl.supp>.
7a834142
JH
2855
2856To get valgrind and for more information see
2857
2858 http://developer.kde.org/~sewardj/
2859
f134cc4e 2860=head2 Compaq's/Digital's/HP's Third Degree
09187cb1
JH
2861
2862Third Degree is a tool for memory leak detection and memory access checks.
2863It is one of the many tools in the ATOM toolkit. The toolkit is only
2864available on Tru64 (formerly known as Digital UNIX formerly known as
2865DEC OSF/1).
2866
2867When building Perl, you must first run Configure with -Doptimize=-g
2868and -Uusemymalloc flags, after that you can use the make targets
51a35ef1
JH
2869"perl.third" and "test.third". (What is required is that Perl must be
2870compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 2871
64cea5fd 2872The short story is that with "atom" you can instrument the Perl
83f0ef60 2873executable to create a new executable called F<perl.third>. When the
4ae3d70a 2874instrumented executable is run, it creates a log of dubious memory
83f0ef60 2875traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a
JH
2876third for more information. The most extensive Third Degree
2877documentation is available in the Compaq "Tru64 UNIX Programmer's
2878Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 2879
9c54ecba 2880The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
64cea5fd
JH
2881subdirectory. There is a problem with these files: Third Degree is so
2882effective that it finds problems also in the system libraries.
9c54ecba
JH
2883Therefore you should used the Porting/thirdclean script to cleanup
2884the F<*.3log> files.
64cea5fd
JH
2885
2886There are also leaks that for given certain definition of a leak,
2887aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2888
2889=head2 PERL_DESTRUCT_LEVEL
2890
a958818a
JH
2891If you want to run any of the tests yourself manually using e.g.
2892valgrind, or the pureperl or perl.third executables, please note that
2893by default perl B<does not> explicitly cleanup all the memory it has
2894allocated (such as global memory arenas) but instead lets the exit()
2895of the whole program "take care" of such allocations, also known as
2896"global destruction of objects".
64cea5fd
JH
2897
2898There is a way to tell perl to do complete cleanup: set the
2899environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2900The t/TEST wrapper does set this to 2, and this is what you
2901need to do too, if you don't want to see the "global leaks":
1f56d61a 2902For example, for "third-degreed" Perl:
64cea5fd 2903
1f56d61a 2904 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
09187cb1 2905
414f2397
RGS
2906(Note: the mod_perl apache module uses also this environment variable
2907for its own purposes and extended its semantics. Refer to the mod_perl
287a822c
RGS
2908documentation for more information. Also, spawned threads do the
2909equivalent of setting this variable to the value 1.)
5a6c59ef
DM
2910
2911If, at the end of a run you get the message I<N scalars leaked>, you can
fd0854ff
DM
2912recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2913of all those leaked SVs to be dumped along with details as to where each
2914SV was originally allocated. This information is also displayed by
2915Devel::Peek. Note that the extra details recorded with each SV increases
2916memory usage, so it shouldn't be used in production environments. It also
2917converts C<new_SV()> from a macro into a real function, so you can use
2918your favourite debugger to discover where those pesky SVs were allocated.
414f2397 2919
d7a2c63c
MHM
2920If you see that you're leaking memory at runtime, but neither valgrind
2921nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
2922leaking SVs that are still reachable and will be properly cleaned up
2923during destruction of the interpreter. In such cases, using the C<-Dm>
2924switch can point you to the source of the leak. If the executable was
2925built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations
2926in addition to memory allocations. Each SV allocation has a distinct
2927serial number that will be written on creation and destruction of the SV.
2928So if you're executing the leaking code in a loop, you need to look for
2929SVs that are created, but never destroyed between each cycle. If such an
2930SV is found, set a conditional breakpoint within C<new_SV()> and make it
2931break only when C<PL_sv_serial> is equal to the serial number of the
2932leaking SV. Then you will catch the interpreter in exactly the state
2933where the leaking SV is allocated, which is sufficient in many cases to
2934find the source of the leak.
2935
2936As C<-Dm> is using the PerlIO layer for output, it will by itself
2937allocate quite a bunch of SVs, which are hidden to avoid recursion.
2938You can bypass the PerlIO layer if you use the SV logging provided
2939by C<-DPERL_MEM_LOG> instead.
2940
46c6c7e2
JH
2941=head2 PERL_MEM_LOG
2942
9f653bb5 2943If compiled with C<-DPERL_MEM_LOG>, all Newx() and Renew() allocations
46c6c7e2
JH
2944and Safefree() in the Perl core go through logging functions, which is
2945handy for breakpoint setting. If also compiled with C<-DPERL_MEM_LOG_STDERR>,
e352bcff
JH
2946the allocations and frees are logged to STDERR (or more precisely, to the
2947file descriptor 2) in these logging functions, with the calling source code
2948file and line number (and C function name, if supported by the C compiler).
2949
2950This logging is somewhat similar to C<-Dm> but independent of C<-DDEBUGGING>,
2951and at a higher level (the C<-Dm> is directly at the point of C<malloc()>,
a015a77e 2952while the C<PERL_MEM_LOG> is at the level of C<New()>).
46c6c7e2 2953
d7a2c63c
MHM
2954In addition to memory allocations, SV allocations will be logged, just as
2955with C<-Dm>. However, since the logging doesn't use PerlIO, all SV allocations
2956are logged and no extra SV allocations are introduced by enabling the logging.
2957If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for each SV
2958allocation is also logged.
2959
2960You can control the logging from your environment if you compile with
2961C<-DPERL_MEM_LOG_ENV>. Then you need to explicitly set C<PERL_MEM_LOG> and/or
2962C<PERL_SV_LOG> to a non-zero value to enable logging of memory and/or SV
2963allocations.
2964
51a35ef1
JH
2965=head2 Profiling
2966
2967Depending on your platform there are various of profiling Perl.
2968
2969There are two commonly used techniques of profiling executables:
10f58044 2970I<statistical time-sampling> and I<basic-block counting>.
51a35ef1
JH
2971
2972The first method takes periodically samples of the CPU program
2973counter, and since the program counter can be correlated with the code
2974generated for functions, we get a statistical view of in which
2975functions the program is spending its time. The caveats are that very
2976small/fast functions have lower probability of showing up in the
2977profile, and that periodically interrupting the program (this is
2978usually done rather frequently, in the scale of milliseconds) imposes
2979an additional overhead that may skew the results. The first problem
2980can be alleviated by running the code for longer (in general this is a
2981good idea for profiling), the second problem is usually kept in guard
2982by the profiling tools themselves.
2983
10f58044 2984The second method divides up the generated code into I<basic blocks>.
51a35ef1
JH
2985Basic blocks are sections of code that are entered only in the
2986beginning and exited only at the end. For example, a conditional jump
2987starts a basic block. Basic block profiling usually works by
10f58044 2988I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1
JH
2989book-keeping code to the generated code. During the execution of the
2990code the basic block counters are then updated appropriately. The
2991caveat is that the added extra code can skew the results: again, the
2992profiling tools usually try to factor their own effects out of the
2993results.
2994
83f0ef60
JH
2995=head2 Gprof Profiling
2996
51a35ef1
JH
2997gprof is a profiling tool available in many UNIX platforms,
2998it uses F<statistical time-sampling>.
83f0ef60
JH
2999
3000You can build a profiled version of perl called "perl.gprof" by
51a35ef1
JH
3001invoking the make target "perl.gprof" (What is required is that Perl
3002must be compiled using the C<-pg> flag, you may need to re-Configure).
3003Running the profiled version of Perl will create an output file called
3004F<gmon.out> is created which contains the profiling data collected
3005during the execution.
83f0ef60
JH
3006
3007The gprof tool can then display the collected data in various ways.
3008Usually gprof understands the following options:
3009
3010=over 4
3011
3012=item -a
3013
3014Suppress statically defined functions from the profile.
3015
3016=item -b
3017
3018Suppress the verbose descriptions in the profile.
3019
3020=item -e routine
3021
3022Exclude the given routine and its descendants from the profile.
3023
3024=item -f routine
3025
3026Display only the given routine and its descendants in the profile.
3027
3028=item -s
3029
3030Generate a summary file called F<gmon.sum> which then may be given
3031to subsequent gprof runs to accumulate data over several runs.
3032
3033=item -z
3034
3035Display routines that have zero usage.
3036
3037=back
3038
3039For more detailed explanation of the available commands and output
3040formats, see your own local documentation of gprof.
3041
038c294a 3042quick hint:
07aa3531
JC
3043
3044 $ sh Configure -des -Dusedevel -Doptimize='-g' -Accflags='-pg' -Aldflags='-pg' && make
3045 $ ./perl someprog # creates gmon.out in current directory
3046 $ gprof perl > out
3047 $ view out
3048
51a35ef1
JH
3049=head2 GCC gcov Profiling
3050
10f58044 3051Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1
JH
3052for the GNU CC.
3053
3054You can build a profiled version of perl called F<perl.gcov> by
3055invoking the make target "perl.gcov" (what is required that Perl must
3056be compiled using gcc with the flags C<-fprofile-arcs
3057-ftest-coverage>, you may need to re-Configure).
3058
3059Running the profiled version of Perl will cause profile output to be
3060generated. For each source file an accompanying ".da" file will be
3061created.
3062
3063To display the results you use the "gcov" utility (which should
3064be installed if you have gcc 3.0 or newer installed). F<gcov> is
3065run on source code files, like this
3066
3067 gcov sv.c
3068
3069which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3070contain the source code annotated with relative frequencies of
3071execution indicated by "#" markers.
3072
3073Useful options of F<gcov> include C<-b> which will summarise the
3074basic block, branch, and function call coverage, and C<-c> which
3075instead of relative frequencies will use the actual counts. For
3076more information on the use of F<gcov> and basic block profiling
3077with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3078
3079 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3080
3081and its section titled "8. gcov: a Test Coverage Program"
3082
3083 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3084
07aa3531
JC
3085quick hint:
3086
3087 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3088 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3089 $ rm -f regexec.c.gcov regexec.gcda
3090 $ ./perl.gcov
3091 $ gcov regexec.c
3092 $ view regexec.c.gcov
3093
4ae3d70a
JH
3094=head2 Pixie Profiling
3095
51a35ef1
JH
3096Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3097UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 3098I<basic-block counting>.
4ae3d70a 3099
83f0ef60 3100You can build a profiled version of perl called F<perl.pixie> by
51a35ef1
JH
3101invoking the make target "perl.pixie" (what is required is that Perl
3102must be compiled using the C<-g> flag, you may need to re-Configure).
3103
3104In Tru64 a file called F<perl.Addrs> will also be silently created,
3105this file contains the addresses of the basic blocks. Running the
3106profiled version of Perl will create a new file called "perl.Counts"
3107which contains the counts for the basic block for that particular
3108program execution.
4ae3d70a 3109
51a35ef1 3110To display the results you use the F<prof> utility. The exact
4ae3d70a
JH
3111incantation depends on your operating system, "prof perl.Counts" in
3112IRIX, and "prof -pixie -all -L. perl" in Tru64.
3113
6c41479b
JH
3114In IRIX the following prof options are available:
3115
3116=over 4
3117
3118=item -h
3119
3120Reports the most heavily used lines in descending order of use.
6e36760b 3121Useful for finding the hotspot lines.
6c41479b
JH
3122
3123=item -l
3124
3125Groups lines by procedure, with procedures sorted in descending order of use.
3126Within a procedure, lines are listed in source order.
6e36760b 3127Useful for finding the hotspots of procedures.
6c41479b
JH
3128
3129=back
3130
3131In Tru64 the following options are available:
3132
3133=over 4
3134
3958b146 3135=item -p[rocedures]
6c41479b 3136
3958b146 3137Procedures sorted in descending order by the number of cycles executed
6e36760b 3138in each procedure. Useful for finding the hotspot procedures.
6c41479b
JH
3139(This is the default option.)
3140
24000d2f 3141=item -h[eavy]
6c41479b 3142
6e36760b
JH
3143Lines sorted in descending order by the number of cycles executed in
3144each line. Useful for finding the hotspot lines.
6c41479b 3145
24000d2f 3146=item -i[nvocations]
6c41479b 3147
6e36760b
JH
3148The called procedures are sorted in descending order by number of calls
3149made to the procedures. Useful for finding the most used procedures.
6c41479b 3150
24000d2f 3151=item -l[ines]
6c41479b
JH
3152
3153Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 3154Useful for finding the hotspots of procedures.
6c41479b
JH
3155
3156=item -testcoverage
3157
3158The compiler emitted code for these lines, but the code was unexecuted.
3159
24000d2f 3160=item -z[ero]
6c41479b
JH
3161
3162Unexecuted procedures.
3163
aa500c9e 3164=back
6c41479b
JH
3165
3166For further information, see your system's manual pages for pixie and prof.
4ae3d70a 3167
b8ddf6b3
SB
3168=head2 Miscellaneous tricks
3169
3170=over 4
3171
3172=item *
3173
cc177e1a 3174Those debugging perl with the DDD frontend over gdb may find the
b8ddf6b3
SB
3175following useful:
3176
3177You can extend the data conversion shortcuts menu, so for example you
3178can display an SV's IV value with one click, without doing any typing.
3179To do that simply edit ~/.ddd/init file and add after:
3180
3181 ! Display shortcuts.
3182 Ddd*gdbDisplayShortcuts: \
3183 /t () // Convert to Bin\n\
3184 /d () // Convert to Dec\n\
3185 /x () // Convert to Hex\n\
3186 /o () // Convert to Oct(\n\
3187
3188the following two lines:
3189
3190 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3191 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3192
3193so now you can do ivx and pvx lookups or you can plug there the
3194sv_peek "conversion":
3195
3196 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3197
3198(The my_perl is for threaded builds.)
3199Just remember that every line, but the last one, should end with \n\
3200
3201Alternatively edit the init file interactively via:
32023rd mouse button -> New Display -> Edit Menu
3203
3204Note: you can define up to 20 conversion shortcuts in the gdb
3205section.
3206
9965345d
JH
3207=item *
3208
7e337ee0
JH
3209If you see in a debugger a memory area mysteriously full of 0xABABABAB
3210or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3211see L<perlclib>.
9965345d 3212
f1fac472
NC
3213=item *
3214
3215Under ithreads the optree is read only. If you want to enforce this, to check
3216for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3217enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3218that allocates op memory via C<mmap>, and sets it read-only at run time.
3219Any write access to an op results in a C<SIGBUS> and abort.
3220
3221This code is intended for development only, and may not be portable even to
3222all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3223all ops read only. Specifically it
3224
3225=over
3226
3227=item 1
3228
3229Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3230later via C<require> or C<eval> will be re-write
3231
3232=item 2
3233
3234Turns an entire slab of ops read-write if the refcount of any op in the slab
3235needs to be decreased.
3236
3237=item 3
3238
3239Turns an entire slab of ops read-write if any op from the slab is freed.
3240
b8ddf6b3
SB
3241=back
3242
f1fac472
NC
3243It's not possible to turn the slabs to read-only after an action requiring
3244read-write access, as either can happen during op tree building time, so
3245there may still be legitimate write access.
3246
3247However, as an 80% solution it is still effective, as currently it catches
3248a write access during the generation of F<Config.pm>, which means that we
3249can't yet build F<perl> with this enabled.
3250
3251=back
3252
3253
955fec6b 3254=head1 CONCLUSION
a422fd2d 3255
955fec6b
JH
3256We've had a brief look around the Perl source, how to maintain quality
3257of the source code, an overview of the stages F<perl> goes through
3258when it's running your code, how to use debuggers to poke at the Perl
3259guts, and finally how to analyse the execution of Perl. We took a very
3260simple problem and demonstrated how to solve it fully - with
3261documentation, regression tests, and finally a patch for submission to
3262p5p. Finally, we talked about how to use external tools to debug and
3263test Perl.
a422fd2d
SC
3264
3265I'd now suggest you read over those references again, and then, as soon
3266as possible, get your hands dirty. The best way to learn is by doing,
07aa3531 3267so:
a422fd2d
SC
3268
3269=over 3
3270
3271=item *
3272
3273Subscribe to perl5-porters, follow the patches and try and understand
3274them; don't be afraid to ask if there's a portion you're not clear on -
3275who knows, you may unearth a bug in the patch...
3276
3277=item *
3278
3279Keep up to date with the bleeding edge Perl distributions and get
3280familiar with the changes. Try and get an idea of what areas people are
3281working on and the changes they're making.
3282
3283=item *
3284
3e148164 3285Do read the README associated with your operating system, e.g. README.aix
a1f349fd
MB
3286on the IBM AIX OS. Don't hesitate to supply patches to that README if
3287you find anything missing or changed over a new OS release.
3288
3289=item *
3290
a422fd2d
SC
3291Find an area of Perl that seems interesting to you, and see if you can
3292work out how it works. Scan through the source, and step over it in the
3293debugger. Play, poke, investigate, fiddle! You'll probably get to
3294understand not just your chosen area but a much wider range of F<perl>'s
3295activity as well, and probably sooner than you'd think.
3296
3297=back
3298
3299=over 3
3300
3301=item I<The Road goes ever on and on, down from the door where it began.>
3302
3303=back
3304
64d9b66b 3305If you can do these things, you've started on the long road to Perl porting.
a422fd2d
SC
3306Thanks for wanting to help make Perl better - and happy hacking!
3307
4ac71550
TC
3308=head2 Metaphoric Quotations
3309
3310If you recognized the quote about the Road above, you're in luck.
3311
3312Most software projects begin each file with a literal description of each
3313file's purpose. Perl instead begins each with a literary allusion to that
3314file's purpose.
3315
3316Like chapters in many books, all top-level Perl source files (along with a
3317few others here and there) begin with an epigramic inscription that alludes,
3318indirectly and metaphorically, to the material you're about to read.
3319
3320Quotations are taken from writings of J.R.R Tolkien pertaining to his
3321Legendarium, almost always from I<The Lord of the Rings>. Chapters and
3322page numbers are given using the following editions:
3323
3324=over 4
3325
3326=item *
3327
3328I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary
3329edition of 2007 was used, published in the UK by Harper Collins Publishers
3330and in the US by the Houghton Mifflin Company.
3331
3332=item *
3333
3334I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover,
333550th-anniversary edition of 2004 was used, published in the UK by Harper
3336Collins Publishers and in the US by the Houghton Mifflin Company.
3337
3338=item *
3339
3340I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his
3341son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes
3342in Christopher's mammoth I<History of Middle Earth>. Page numbers derive
3343from the hardcover edition, first published in 1983 by George Allen &
3344Unwin; no page numbers changed for the special 3-volume omnibus edition of
33452002 or the various trade-paper editions, all again now by Harper Collins
3346or Houghton Mifflin.
3347
3348=back
3349
3350Other JRRT books fair game for quotes would thus include I<The Adventures of
3351Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of
3352the Children of Hurin>, all but the first posthumously assembled by CJRT.
3353But I<The Lord of the Rings> itself is perfectly fine and probably best to
3354quote from, provided you can find a suitable quote there.
3355
3356So if you were to supply a new, complete, top-level source file to add to
3357Perl, you should conform to this peculiar practice by yourself selecting an
3358appropriate quotation from Tolkien, retaining the original spelling and
3359punctuation and using the same format the rest of the quotes are in.
3360Indirect and oblique is just fine; remember, it's a metaphor, so being meta
3361is, after all, what it's for.
3362
e8cd7eae
GS
3363=head1 AUTHOR
3364
3365This document was written by Nathan Torkington, and is maintained by
3366the perl5-porters mailing list.
4ac71550 3367
b16c2e4a
RGS
3368=head1 SEE ALSO
3369
3370L<perlrepository>