Update Changes.
[p5sagit/p5-mst-13.2.git] / pod / perlhack.pod
CommitLineData
e8cd7eae 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 18
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
20
f8e3975a 21or
22
23 http://archive.develooper.com/perl5-porters@perl.org/
24
e8cd7eae 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
37releases of Perl are shepherded by a ``pumpking'', a porter
38responsible for gathering patches, deciding on a patch-by-patch
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
41Perl, and Jarkko Hietaniemi is the pumpking for the 5.8 release, and
42Hugo van der Sanden will be the pumpking for the 5.10 release.
e8cd7eae 43
44In addition, various people are pumpkings for different things. For
45instance, Andy Dougherty and Jarkko Hietaniemi share the I<Configure>
caf100c0 46pumpkin.
e8cd7eae 47
48Larry sees Perl development along the lines of the US government:
49there's the Legislature (the porters), the Executive branch (the
50pumpkings), and the Supreme Court (Larry). The legislature can
51discuss and submit patches to the executive branch all they like, but
52the executive branch is free to veto them. Rarely, the Supreme Court
53will side with the executive branch over the legislature, or the
54legislature over the executive branch. Mostly, however, the
55legislature and the executive branch are supposed to get along and
56work out their differences without impeachment or court cases.
57
58You might sometimes see reference to Rule 1 and Rule 2. Larry's power
59as Supreme Court is expressed in The Rules:
60
61=over 4
62
63=item 1
64
65Larry is always by definition right about how Perl should behave.
66This means he has final veto power on the core functionality.
67
68=item 2
69
70Larry is allowed to change his mind about any matter at a later date,
71regardless of whether he previously invoked Rule 1.
72
73=back
74
75Got that? Larry is always right, even when he was wrong. It's rare
76to see either Rule exercised, but they are often alluded to.
77
78New features and extensions to the language are contentious, because
79the criteria used by the pumpkings, Larry, and other porters to decide
80which features should be implemented and incorporated are not codified
81in a few small design goals as with some other languages. Instead,
82the heuristics are flexible and often difficult to fathom. Here is
83one person's list, roughly in decreasing order of importance, of
84heuristics that new features have to be weighed against:
85
86=over 4
87
88=item Does concept match the general goals of Perl?
89
90These haven't been written anywhere in stone, but one approximation
91is:
92
93 1. Keep it fast, simple, and useful.
94 2. Keep features/concepts as orthogonal as possible.
95 3. No arbitrary limits (platforms, data sizes, cultures).
96 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
97 5. Either assimilate new technologies, or build bridges to them.
98
99=item Where is the implementation?
100
101All the talk in the world is useless without an implementation. In
102almost every case, the person or people who argue for a new feature
103will be expected to be the ones who implement it. Porters capable
104of coding new features have their own agendas, and are not available
105to implement your (possibly good) idea.
106
107=item Backwards compatibility
108
109It's a cardinal sin to break existing Perl programs. New warnings are
110contentious--some say that a program that emits warnings is not
111broken, while others say it is. Adding keywords has the potential to
112break programs, changing the meaning of existing token sequences or
113functions might break programs.
114
115=item Could it be a module instead?
116
117Perl 5 has extension mechanisms, modules and XS, specifically to avoid
118the need to keep changing the Perl interpreter. You can write modules
119that export functions, you can give those functions prototypes so they
120can be called like built-in functions, you can even write XS code to
121mess with the runtime data structures of the Perl interpreter if you
122want to implement really complicated things. If it can be done in a
123module instead of in the core, it's highly unlikely to be added.
124
125=item Is the feature generic enough?
126
127Is this something that only the submitter wants added to the language,
128or would it be broadly useful? Sometimes, instead of adding a feature
129with a tight focus, the porters might decide to wait until someone
130implements the more generalized feature. For instance, instead of
131implementing a ``delayed evaluation'' feature, the porters are waiting
132for a macro system that would permit delayed evaluation and much more.
133
134=item Does it potentially introduce new bugs?
135
136Radical rewrites of large chunks of the Perl interpreter have the
137potential to introduce new bugs. The smaller and more localized the
138change, the better.
139
140=item Does it preclude other desirable features?
141
142A patch is likely to be rejected if it closes off future avenues of
143development. For instance, a patch that placed a true and final
144interpretation on prototypes is likely to be rejected because there
145are still options for the future of prototypes that haven't been
146addressed.
147
148=item Is the implementation robust?
149
150Good patches (tight code, complete, correct) stand more chance of
151going in. Sloppy or incorrect patches might be placed on the back
152burner until the pumpking has time to fix, or might be discarded
153altogether without further notice.
154
155=item Is the implementation generic enough to be portable?
156
157The worst patches make use of a system-specific features. It's highly
158unlikely that nonportable additions to the Perl language will be
159accepted.
160
a936dd3c 161=item Is the implementation tested?
162
163Patches which change behaviour (fixing bugs or introducing new features)
164must include regression tests to verify that everything works as expected.
165Without tests provided by the original author, how can anyone else changing
166perl in the future be sure that they haven't unwittingly broken the behaviour
167the patch implements? And without tests, how can the patch's author be
9d077eaa 168confident that his/her hard work put into the patch won't be accidentally
a936dd3c 169thrown away by someone in the future?
170
e8cd7eae 171=item Is there enough documentation?
172
173Patches without documentation are probably ill-thought out or
174incomplete. Nothing can be added without documentation, so submitting
175a patch for the appropriate manpages as well as the source code is
a936dd3c 176always a good idea.
e8cd7eae 177
178=item Is there another way to do it?
179
180Larry said ``Although the Perl Slogan is I<There's More Than One Way
181to Do It>, I hesitate to make 10 ways to do something''. This is a
182tricky heuristic to navigate, though--one man's essential addition is
183another man's pointless cruft.
184
185=item Does it create too much work?
186
187Work for the pumpking, work for Perl programmers, work for module
188authors, ... Perl is supposed to be easy.
189
f6c51b38 190=item Patches speak louder than words
191
192Working code is always preferred to pie-in-the-sky ideas. A patch to
193add a feature stands a much higher chance of making it to the language
194than does a random feature request, no matter how fervently argued the
195request might be. This ties into ``Will it be useful?'', as the fact
196that someone took the time to make the patch demonstrates a strong
197desire for the feature.
198
e8cd7eae 199=back
200
201If you're on the list, you might hear the word ``core'' bandied
202around. It refers to the standard distribution. ``Hacking on the
203core'' means you're changing the C source code to the Perl
204interpreter. ``A core module'' is one that ships with Perl.
205
a1f349fd 206=head2 Keeping in sync
207
e8cd7eae 208The source code to the Perl interpreter, in its different versions, is
209kept in a repository managed by a revision control system (which is
210currently the Perforce program, see http://perforce.com/). The
211pumpkings and a few others have access to the repository to check in
212changes. Periodically the pumpking for the development version of Perl
213will release a new version, so the rest of the porters can see what's
2be4c08b 214changed. The current state of the main trunk of repository, and patches
215that describe the individual changes that have happened since the last
216public release are available at this location:
217
218 ftp://ftp.linux.activestate.com/pub/staff/gsar/APC/
219
a1f349fd 220If you are a member of the perl5-porters mailing list, it is a good
221thing to keep in touch with the most recent changes. If not only to
222verify if what you would have posted as a bug report isn't already
223solved in the most recent available perl development branch, also
224known as perl-current, bleading edge perl, bleedperl or bleadperl.
2be4c08b 225
226Needless to say, the source code in perl-current is usually in a perpetual
227state of evolution. You should expect it to be very buggy. Do B<not> use
228it for any purpose other than testing and development.
e8cd7eae 229
3e148164 230Keeping in sync with the most recent branch can be done in several ways,
231but the most convenient and reliable way is using B<rsync>, available at
232ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
233branch by FTP.)
a1f349fd 234
235If you choose to keep in sync using rsync, there are two approaches
3e148164 236to doing so:
a1f349fd 237
238=over 4
239
240=item rsync'ing the source tree
241
3e148164 242Presuming you are in the directory where your perl source resides
a1f349fd 243and you have rsync installed and available, you can `upgrade' to
244the bleadperl using:
245
246 # rsync -avz rsync://ftp.linux.activestate.com/perl-current/ .
247
248This takes care of updating every single item in the source tree to
249the latest applied patch level, creating files that are new (to your
250distribution) and setting date/time stamps of existing files to
251reflect the bleadperl status.
252
c6d0653e 253Note that this will not delete any files that were in '.' before
254the rsync. Once you are sure that the rsync is running correctly,
255run it with the --delete and the --dry-run options like this:
256
257 # rsync -avz --delete --dry-run rsync://ftp.linux.activestate.com/perl-current/ .
258
259This will I<simulate> an rsync run that also deletes files not
260present in the bleadperl master copy. Observe the results from
261this run closely. If you are sure that the actual run would delete
262no files precious to you, you could remove the '--dry-run' option.
263
a1f349fd 264You can than check what patch was the latest that was applied by
265looking in the file B<.patch>, which will show the number of the
266latest patch.
267
268If you have more than one machine to keep in sync, and not all of
269them have access to the WAN (so you are not able to rsync all the
270source trees to the real source), there are some ways to get around
271this problem.
272
273=over 4
274
275=item Using rsync over the LAN
276
277Set up a local rsync server which makes the rsynced source tree
3e148164 278available to the LAN and sync the other machines against this
a1f349fd 279directory.
280
281From http://rsync.samba.org/README.html:
282
283 "Rsync uses rsh or ssh for communication. It does not need to be
284 setuid and requires no special privileges for installation. It
3958b146 285 does not require an inetd entry or a daemon. You must, however,
a1f349fd 286 have a working rsh or ssh system. Using ssh is recommended for
287 its security features."
288
289=item Using pushing over the NFS
290
291Having the other systems mounted over the NFS, you can take an
3e148164 292active pushing approach by checking the just updated tree against
293the other not-yet synced trees. An example would be
294
295 #!/usr/bin/perl -w
296
297 use strict;
298 use File::Copy;
299
300 my %MF = map {
301 m/(\S+)/;
302 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
303 } `cat MANIFEST`;
304
305 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
306
307 foreach my $host (keys %remote) {
308 unless (-d $remote{$host}) {
309 print STDERR "Cannot Xsync for host $host\n";
310 next;
311 }
312 foreach my $file (keys %MF) {
313 my $rfile = "$remote{$host}/$file";
314 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
315 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
316 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
317 printf "%4s %-34s %8d %9d %8d %9d\n",
318 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
319 unlink $rfile;
320 copy ($file, $rfile);
321 utime time, $MF{$file}[2], $rfile;
322 chmod $MF{$file}[0], $rfile;
323 }
324 }
325
326though this is not perfect. It could be improved with checking
a1f349fd 327file checksums before updating. Not all NFS systems support
328reliable utime support (when used over the NFS).
329
330=back
331
332=item rsync'ing the patches
333
334The source tree is maintained by the pumpking who applies patches to
335the files in the tree. These patches are either created by the
336pumpking himself using C<diff -c> after updating the file manually or
337by applying patches sent in by posters on the perl5-porters list.
338These patches are also saved and rsync'able, so you can apply them
339yourself to the source files.
340
341Presuming you are in a directory where your patches reside, you can
3e148164 342get them in sync with
a1f349fd 343
344 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
345
346This makes sure the latest available patch is downloaded to your
347patch directory.
348
3e148164 349It's then up to you to apply these patches, using something like
a1f349fd 350
df3477ff 351 # last=`ls -t *.gz | sed q`
a1f349fd 352 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
353 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
354 # cd ../perl-current
355 # patch -p1 -N <../perl-current-diffs/blead.patch
356
357or, since this is only a hint towards how it works, use CPAN-patchaperl
358from Andreas König to have better control over the patching process.
359
360=back
361
f7e1e956 362=head2 Why rsync the source tree
a1f349fd 363
364=over 4
365
10f58044 366=item It's easier to rsync the source tree
a1f349fd 367
368Since you don't have to apply the patches yourself, you are sure all
369files in the source tree are in the right state.
370
371=item It's more recent
372
373According to Gurusamy Sarathy:
374
375 "... The rsync mirror is automatic and syncs with the repository
376 every five minutes.
377
3e148164 378 "Updating the patch area still requires manual intervention
a1f349fd 379 (with all the goofiness that implies, which you've noted) and
380 is typically on a daily cycle. Making this process automatic
381 is on my tuit list, but don't ask me when."
382
383=item It's more reliable
384
3e148164 385Well, since the patches are updated by hand, I don't have to say any
a1f349fd 386more ... (see Sarathy's remark).
387
388=back
389
f7e1e956 390=head2 Why rsync the patches
a1f349fd 391
392=over 4
393
10f58044 394=item It's easier to rsync the patches
a1f349fd 395
396If you have more than one machine that you want to keep in track with
3e148164 397bleadperl, it's easier to rsync the patches only once and then apply
a1f349fd 398them to all the source trees on the different machines.
399
400In case you try to keep in pace on 5 different machines, for which
401only one of them has access to the WAN, rsync'ing all the source
3e148164 402trees should than be done 5 times over the NFS. Having
a1f349fd 403rsync'ed the patches only once, I can apply them to all the source
3e148164 404trees automatically. Need you say more ;-)
a1f349fd 405
406=item It's a good reference
407
408If you do not only like to have the most recent development branch,
409but also like to B<fix> bugs, or extend features, you want to dive
410into the sources. If you are a seasoned perl core diver, you don't
411need no manuals, tips, roadmaps, perlguts.pod or other aids to find
412your way around. But if you are a starter, the patches may help you
413in finding where you should start and how to change the bits that
414bug you.
415
416The file B<Changes> is updated on occasions the pumpking sees as his
417own little sync points. On those occasions, he releases a tar-ball of
418the current source tree (i.e. perl@7582.tar.gz), which will be an
419excellent point to start with when choosing to use the 'rsync the
420patches' scheme. Starting with perl@7582, which means a set of source
421files on which the latest applied patch is number 7582, you apply all
f18956b7 422succeeding patches available from then on (7583, 7584, ...).
a1f349fd 423
424You can use the patches later as a kind of search archive.
425
426=over 4
427
428=item Finding a start point
429
430If you want to fix/change the behaviour of function/feature Foo, just
431scan the patches for patches that mention Foo either in the subject,
3e148164 432the comments, or the body of the fix. A good chance the patch shows
a1f349fd 433you the files that are affected by that patch which are very likely
434to be the starting point of your journey into the guts of perl.
435
436=item Finding how to fix a bug
437
438If you've found I<where> the function/feature Foo misbehaves, but you
439don't know how to fix it (but you do know the change you want to
440make), you can, again, peruse the patches for similar changes and
441look how others apply the fix.
442
443=item Finding the source of misbehaviour
444
445When you keep in sync with bleadperl, the pumpking would love to
3958b146 446I<see> that the community efforts really work. So after each of his
a1f349fd 447sync points, you are to 'make test' to check if everything is still
448in working order. If it is, you do 'make ok', which will send an OK
449report to perlbug@perl.org. (If you do not have access to a mailer
3e148164 450from the system you just finished successfully 'make test', you can
a1f349fd 451do 'make okfile', which creates the file C<perl.ok>, which you can
452than take to your favourite mailer and mail yourself).
453
3958b146 454But of course, as always, things will not always lead to a success
a1f349fd 455path, and one or more test do not pass the 'make test'. Before
456sending in a bug report (using 'make nok' or 'make nokfile'), check
457the mailing list if someone else has reported the bug already and if
458so, confirm it by replying to that message. If not, you might want to
459trace the source of that misbehaviour B<before> sending in the bug,
460which will help all the other porters in finding the solution.
461
3e148164 462Here the saved patches come in very handy. You can check the list of
463patches to see which patch changed what file and what change caused
464the misbehaviour. If you note that in the bug report, it saves the
465one trying to solve it, looking for that point.
a1f349fd 466
467=back
468
469If searching the patches is too bothersome, you might consider using
470perl's bugtron to find more information about discussions and
471ramblings on posted bugs.
472
3e148164 473If you want to get the best of both worlds, rsync both the source
474tree for convenience, reliability and ease and rsync the patches
475for reference.
476
52315700 477=back
478
479
480=head2 Perlbug remote interface
481
482=over 4
483
484There are three (3) remote administrative interfaces for modifying bug status, category, etc. In all cases an admin must be first registered with the Perlbug database by sending an email request to richard@perl.org or bugmongers@perl.org.
485
486The main requirement is the willingness to classify, (with the emphasis on closing where possible :), outstanding bugs. Further explanation can be garnered from the web at http://bugs.perl.org/, or by asking on the admin mailing list at: bugmongers@perl.org
487
488For more info on the web see
489
490 http://bugs.perl.org/perlbug.cgi?req=spec
491
492
493B<The interfaces:>
494
495
496=item 1 http://bugs.perl.org
497
498Login via the web, (remove B<admin/> if only browsing), where interested Cc's, tests, patches and change-ids, etc. may be assigned.
499
500 http://bugs.perl.org/admin/index.html
501
502
503=item 2 bugdb@perl.org
504
505Where the subject line is used for commands:
506
507 To: bugdb@perl.org
508 Subject: -a close bugid1 bugid2 aix install
509
510 To: bugdb@perl.org
511 Subject: -h
512
513
514=item 3 commands_and_bugdids@bugs.perl.org
515
516Where the address itself is the source for the commands:
517
518 To: close_bugid1_bugid2_aix@bugs.perl.org
519
520 To: help@bugs.perl.org
521
522
523=item notes, patches, tests
524
525For patches and tests, the message body is assigned to the appropriate bug/s and forwarded to p5p for their attention.
526
527 To: test_<bugid1>_aix_close@bugs.perl.org
528 Subject: this is a test for the (now closed) aix bug
529
530 Test is the body of the mail
531
532=back
533
a1f349fd 534=head2 Submitting patches
535
f7e1e956 536Always submit patches to I<perl5-porters@perl.org>. If you're
537patching a core module and there's an author listed, send the author a
538copy (see L<Patching a core module>). This lets other porters review
539your patch, which catches a surprising number of errors in patches.
540Either use the diff program (available in source code form from
541I<ftp://ftp.gnu.org/pub/gnu/>), or use Johan Vromans' I<makepatch>
542(available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
543but context diffs are accepted. Do not send RCS-style diffs or diffs
544without context lines. More information is given in the
545I<Porting/patching.pod> file in the Perl source distribution. Please
546patch against the latest B<development> version (e.g., if you're
547fixing a bug in the 5.005 track, patch against the latest 5.005_5x
548version). Only patches that survive the heat of the development
549branch get applied to maintenance versions.
550
551Your patch should update the documentation and test suite. See
552L<Writing a test>.
e8cd7eae 553
554To report a bug in Perl, use the program I<perlbug> which comes with
555Perl (if you can't get Perl to work, send mail to the address
f18956b7 556I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
e8cd7eae 557I<perlbug> feeds into the automated bug-tracking system, access to
558which is provided through the web at I<http://bugs.perl.org/>. It
559often pays to check the archives of the perl5-porters mailing list to
560see whether the bug you're reporting has been reported before, and if
561so whether it was considered a bug. See above for the location of
562the searchable archives.
563
564The CPAN testers (I<http://testers.cpan.org/>) are a group of
565volunteers who test CPAN modules on a variety of platforms. Perl Labs
f6c51b38 566(I<http://labs.perl.org/>) automatically tests Perl source releases on
567platforms and gives feedback to the CPAN testers mailing list. Both
568efforts welcome volunteers.
e8cd7eae 569
e8cd7eae 570It's a good idea to read and lurk for a while before chipping in.
571That way you'll get to see the dynamic of the conversations, learn the
572personalities of the players, and hopefully be better prepared to make
573a useful contribution when do you speak up.
574
575If after all this you still think you want to join the perl5-porters
f6c51b38 576mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
577unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
e8cd7eae 578
a422fd2d 579To hack on the Perl guts, you'll need to read the following things:
580
581=over 3
582
583=item L<perlguts>
584
585This is of paramount importance, since it's the documentation of what
586goes where in the Perl source. Read it over a couple of times and it
587might start to make sense - don't worry if it doesn't yet, because the
588best way to study it is to read it in conjunction with poking at Perl
589source, and we'll do that later on.
590
591You might also want to look at Gisle Aas's illustrated perlguts -
592there's no guarantee that this will be absolutely up-to-date with the
593latest documentation in the Perl core, but the fundamentals will be
594right. (http://gisle.aas.no/perl/illguts/)
595
596=item L<perlxstut> and L<perlxs>
597
598A working knowledge of XSUB programming is incredibly useful for core
599hacking; XSUBs use techniques drawn from the PP code, the portion of the
600guts that actually executes a Perl program. It's a lot gentler to learn
601those techniques from simple examples and explanation than from the core
602itself.
603
604=item L<perlapi>
605
606The documentation for the Perl API explains what some of the internal
607functions do, as well as the many macros used in the source.
608
609=item F<Porting/pumpkin.pod>
610
611This is a collection of words of wisdom for a Perl porter; some of it is
612only useful to the pumpkin holder, but most of it applies to anyone
613wanting to go about Perl development.
614
615=item The perl5-porters FAQ
616
617This is posted to perl5-porters at the beginning on every month, and
618should be available from http://perlhacker.org/p5p-faq; alternatively,
619you can get the FAQ emailed to you by sending mail to
620C<perl5-porters-faq@perl.org>. It contains hints on reading
621perl5-porters, information on how perl5-porters works and how Perl
622development in general works.
623
624=back
625
626=head2 Finding Your Way Around
627
628Perl maintenance can be split into a number of areas, and certain people
629(pumpkins) will have responsibility for each area. These areas sometimes
630correspond to files or directories in the source kit. Among the areas are:
631
632=over 3
633
634=item Core modules
635
636Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
637subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
638contains the core XS modules.
639
f7e1e956 640=item Tests
641
642There are tests for nearly all the modules, built-ins and major bits
643of functionality. Test files all have a .t suffix. Module tests live
644in the F<lib/> and F<ext/> directories next to the module being
645tested. Others live in F<t/>. See L<Writing a test>
646
a422fd2d 647=item Documentation
648
649Documentation maintenance includes looking after everything in the
650F<pod/> directory, (as well as contributing new documentation) and
651the documentation to the modules in core.
652
653=item Configure
654
655The configure process is the way we make Perl portable across the
656myriad of operating systems it supports. Responsibility for the
657configure, build and installation process, as well as the overall
658portability of the core code rests with the configure pumpkin - others
659help out with individual operating systems.
660
661The files involved are the operating system directories, (F<win32/>,
662F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
663and F<Makefile>, as well as the metaconfig files which generate
664F<Configure>. (metaconfig isn't included in the core distribution.)
665
666=item Interpreter
667
668And of course, there's the core of the Perl interpreter itself. Let's
669have a look at that in a little more detail.
670
671=back
672
673Before we leave looking at the layout, though, don't forget that
674F<MANIFEST> contains not only the file names in the Perl distribution,
675but short descriptions of what's in them, too. For an overview of the
676important files, try this:
677
678 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
679
680=head2 Elements of the interpreter
681
682The work of the interpreter has two main stages: compiling the code
683into the internal representation, or bytecode, and then executing it.
684L<perlguts/Compiled code> explains exactly how the compilation stage
685happens.
686
687Here is a short breakdown of perl's operation:
688
689=over 3
690
691=item Startup
692
693The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
694This is very high-level code, enough to fit on a single screen, and it
695resembles the code found in L<perlembed>; most of the real action takes
696place in F<perl.c>
697
698First, F<perlmain.c> allocates some memory and constructs a Perl
699interpreter:
700
701 1 PERL_SYS_INIT3(&argc,&argv,&env);
702 2
703 3 if (!PL_do_undump) {
704 4 my_perl = perl_alloc();
705 5 if (!my_perl)
706 6 exit(1);
707 7 perl_construct(my_perl);
708 8 PL_perl_destruct_level = 0;
709 9 }
710
711Line 1 is a macro, and its definition is dependent on your operating
712system. Line 3 references C<PL_do_undump>, a global variable - all
713global variables in Perl start with C<PL_>. This tells you whether the
714current running program was created with the C<-u> flag to perl and then
715F<undump>, which means it's going to be false in any sane context.
716
717Line 4 calls a function in F<perl.c> to allocate memory for a Perl
718interpreter. It's quite a simple function, and the guts of it looks like
719this:
720
721 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
722
723Here you see an example of Perl's system abstraction, which we'll see
724later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
725own C<malloc> as defined in F<malloc.c> if you selected that option at
726configure time.
727
728Next, in line 7, we construct the interpreter; this sets up all the
729special variables that Perl needs, the stacks, and so on.
730
731Now we pass Perl the command line options, and tell it to go:
732
733 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
734 if (!exitstatus) {
735 exitstatus = perl_run(my_perl);
736 }
737
738
739C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
740in F<perl.c>, which processes the command line options, sets up any
741statically linked XS modules, opens the program and calls C<yyparse> to
742parse it.
743
744=item Parsing
745
746The aim of this stage is to take the Perl source, and turn it into an op
747tree. We'll see what one of those looks like later. Strictly speaking,
748there's three things going on here.
749
750C<yyparse>, the parser, lives in F<perly.c>, although you're better off
751reading the original YACC input in F<perly.y>. (Yes, Virginia, there
752B<is> a YACC grammar for Perl!) The job of the parser is to take your
753code and `understand' it, splitting it into sentences, deciding which
754operands go with which operators and so on.
755
756The parser is nobly assisted by the lexer, which chunks up your input
757into tokens, and decides what type of thing each token is: a variable
758name, an operator, a bareword, a subroutine, a core function, and so on.
759The main point of entry to the lexer is C<yylex>, and that and its
760associated routines can be found in F<toke.c>. Perl isn't much like
761other computer languages; it's highly context sensitive at times, it can
762be tricky to work out what sort of token something is, or where a token
763ends. As such, there's a lot of interplay between the tokeniser and the
764parser, which can get pretty frightening if you're not used to it.
765
766As the parser understands a Perl program, it builds up a tree of
767operations for the interpreter to perform during execution. The routines
768which construct and link together the various operations are to be found
769in F<op.c>, and will be examined later.
770
771=item Optimization
772
773Now the parsing stage is complete, and the finished tree represents
774the operations that the Perl interpreter needs to perform to execute our
775program. Next, Perl does a dry run over the tree looking for
776optimisations: constant expressions such as C<3 + 4> will be computed
777now, and the optimizer will also see if any multiple operations can be
778replaced with a single one. For instance, to fetch the variable C<$foo>,
779instead of grabbing the glob C<*foo> and looking at the scalar
780component, the optimizer fiddles the op tree to use a function which
781directly looks up the scalar in question. The main optimizer is C<peep>
782in F<op.c>, and many ops have their own optimizing functions.
783
784=item Running
785
786Now we're finally ready to go: we have compiled Perl byte code, and all
787that's left to do is run it. The actual execution is done by the
788C<runops_standard> function in F<run.c>; more specifically, it's done by
789these three innocent looking lines:
790
791 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
792 PERL_ASYNC_CHECK();
793 }
794
795You may be more comfortable with the Perl version of that:
796
797 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
798
799Well, maybe not. Anyway, each op contains a function pointer, which
800stipulates the function which will actually carry out the operation.
801This function will return the next op in the sequence - this allows for
802things like C<if> which choose the next op dynamically at run time.
803The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
804execution if required.
805
806The actual functions called are known as PP code, and they're spread
807between four files: F<pp_hot.c> contains the `hot' code, which is most
808often used and highly optimized, F<pp_sys.c> contains all the
809system-specific functions, F<pp_ctl.c> contains the functions which
810implement control structures (C<if>, C<while> and the like) and F<pp.c>
811contains everything else. These are, if you like, the C code for Perl's
812built-in functions and operators.
813
814=back
815
816=head2 Internal Variable Types
817
818You should by now have had a look at L<perlguts>, which tells you about
819Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
820that now.
821
822These variables are used not only to represent Perl-space variables, but
823also any constants in the code, as well as some structures completely
824internal to Perl. The symbol table, for instance, is an ordinary Perl
825hash. Your code is represented by an SV as it's read into the parser;
826any program files you call are opened via ordinary Perl filehandles, and
827so on.
828
829The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
830Perl program. Let's see, for instance, how Perl treats the constant
831C<"hello">.
832
833 % perl -MDevel::Peek -e 'Dump("hello")'
834 1 SV = PV(0xa041450) at 0xa04ecbc
835 2 REFCNT = 1
836 3 FLAGS = (POK,READONLY,pPOK)
837 4 PV = 0xa0484e0 "hello"\0
838 5 CUR = 5
839 6 LEN = 6
840
841Reading C<Devel::Peek> output takes a bit of practise, so let's go
842through it line by line.
843
844Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
845memory. SVs themselves are very simple structures, but they contain a
846pointer to a more complex structure. In this case, it's a PV, a
847structure which holds a string value, at location C<0xa041450>. Line 2
848is the reference count; there are no other references to this data, so
849it's 1.
850
851Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
852read-only SV (because it's a constant) and the data is a PV internally.
853Next we've got the contents of the string, starting at location
854C<0xa0484e0>.
855
856Line 5 gives us the current length of the string - note that this does
857B<not> include the null terminator. Line 6 is not the length of the
858string, but the length of the currently allocated buffer; as the string
859grows, Perl automatically extends the available storage via a routine
860called C<SvGROW>.
861
862You can get at any of these quantities from C very easily; just add
863C<Sv> to the name of the field shown in the snippet, and you've got a
864macro which will return the value: C<SvCUR(sv)> returns the current
865length of the string, C<SvREFCOUNT(sv)> returns the reference count,
866C<SvPV(sv, len)> returns the string itself with its length, and so on.
867More macros to manipulate these properties can be found in L<perlguts>.
868
869Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
870
871 1 void
872 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
873 3 {
874 4 STRLEN tlen;
875 5 char *junk;
876
877 6 junk = SvPV_force(sv, tlen);
878 7 SvGROW(sv, tlen + len + 1);
879 8 if (ptr == junk)
880 9 ptr = SvPVX(sv);
881 10 Move(ptr,SvPVX(sv)+tlen,len,char);
882 11 SvCUR(sv) += len;
883 12 *SvEND(sv) = '\0';
884 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
885 14 SvTAINT(sv);
886 15 }
887
888This is a function which adds a string, C<ptr>, of length C<len> onto
889the end of the PV stored in C<sv>. The first thing we do in line 6 is
890make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
891macro to force a PV. As a side effect, C<tlen> gets set to the current
892value of the PV, and the PV itself is returned to C<junk>.
893
b1866b2d 894In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d 895the old string, the new string and the null terminator. If C<LEN> isn't
896big enough, C<SvGROW> will reallocate space for us.
897
898Now, if C<junk> is the same as the string we're trying to add, we can
899grab the string directly from the SV; C<SvPVX> is the address of the PV
900in the SV.
901
902Line 10 does the actual catenation: the C<Move> macro moves a chunk of
903memory around: we move the string C<ptr> to the end of the PV - that's
904the start of the PV plus its current length. We're moving C<len> bytes
905of type C<char>. After doing so, we need to tell Perl we've extended the
906string, by altering C<CUR> to reflect the new length. C<SvEND> is a
907macro which gives us the end of the string, so that needs to be a
908C<"\0">.
909
910Line 13 manipulates the flags; since we've changed the PV, any IV or NV
911values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
912want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF8-aware
913version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
914and turns on POK. The final C<SvTAINT> is a macro which launders tainted
915data if taint mode is turned on.
916
917AVs and HVs are more complicated, but SVs are by far the most common
918variable type being thrown around. Having seen something of how we
919manipulate these, let's go on and look at how the op tree is
920constructed.
921
922=head2 Op Trees
923
924First, what is the op tree, anyway? The op tree is the parsed
925representation of your program, as we saw in our section on parsing, and
926it's the sequence of operations that Perl goes through to execute your
927program, as we saw in L</Running>.
928
929An op is a fundamental operation that Perl can perform: all the built-in
930functions and operators are ops, and there are a series of ops which
931deal with concepts the interpreter needs internally - entering and
932leaving a block, ending a statement, fetching a variable, and so on.
933
934The op tree is connected in two ways: you can imagine that there are two
935"routes" through it, two orders in which you can traverse the tree.
936First, parse order reflects how the parser understood the code, and
937secondly, execution order tells perl what order to perform the
938operations in.
939
940The easiest way to examine the op tree is to stop Perl after it has
941finished parsing, and get it to dump out the tree. This is exactly what
942the compiler backends L<B::Terse|B::Terse> and L<B::Debug|B::Debug> do.
943
944Let's have a look at how Perl sees C<$a = $b + $c>:
945
946 % perl -MO=Terse -e '$a=$b+$c'
947 1 LISTOP (0x8179888) leave
948 2 OP (0x81798b0) enter
949 3 COP (0x8179850) nextstate
950 4 BINOP (0x8179828) sassign
951 5 BINOP (0x8179800) add [1]
952 6 UNOP (0x81796e0) null [15]
953 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
954 8 UNOP (0x81797e0) null [15]
955 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
956 10 UNOP (0x816b4f0) null [15]
957 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
958
959Let's start in the middle, at line 4. This is a BINOP, a binary
960operator, which is at location C<0x8179828>. The specific operator in
961question is C<sassign> - scalar assignment - and you can find the code
962which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
963binary operator, it has two children: the add operator, providing the
964result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
965line 10.
966
967Line 10 is the null op: this does exactly nothing. What is that doing
968there? If you see the null op, it's a sign that something has been
969optimized away after parsing. As we mentioned in L</Optimization>,
970the optimization stage sometimes converts two operations into one, for
971example when fetching a scalar variable. When this happens, instead of
972rewriting the op tree and cleaning up the dangling pointers, it's easier
973just to replace the redundant operation with the null op. Originally,
974the tree would have looked like this:
975
976 10 SVOP (0x816b4f0) rv2sv [15]
977 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
978
979That is, fetch the C<a> entry from the main symbol table, and then look
980at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
981happens to do both these things.
982
983The right hand side, starting at line 5 is similar to what we've just
984seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
985two C<gvsv>s.
986
987Now, what's this about?
988
989 1 LISTOP (0x8179888) leave
990 2 OP (0x81798b0) enter
991 3 COP (0x8179850) nextstate
992
993C<enter> and C<leave> are scoping ops, and their job is to perform any
994housekeeping every time you enter and leave a block: lexical variables
995are tidied up, unreferenced variables are destroyed, and so on. Every
996program will have those first three lines: C<leave> is a list, and its
997children are all the statements in the block. Statements are delimited
998by C<nextstate>, so a block is a collection of C<nextstate> ops, with
999the ops to be performed for each statement being the children of
1000C<nextstate>. C<enter> is a single op which functions as a marker.
1001
1002That's how Perl parsed the program, from top to bottom:
1003
1004 Program
1005 |
1006 Statement
1007 |
1008 =
1009 / \
1010 / \
1011 $a +
1012 / \
1013 $b $c
1014
1015However, it's impossible to B<perform> the operations in this order:
1016you have to find the values of C<$b> and C<$c> before you add them
1017together, for instance. So, the other thread that runs through the op
1018tree is the execution order: each op has a field C<op_next> which points
1019to the next op to be run, so following these pointers tells us how perl
1020executes the code. We can traverse the tree in this order using
1021the C<exec> option to C<B::Terse>:
1022
1023 % perl -MO=Terse,exec -e '$a=$b+$c'
1024 1 OP (0x8179928) enter
1025 2 COP (0x81798c8) nextstate
1026 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
1027 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
1028 5 BINOP (0x8179878) add [1]
1029 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
1030 7 BINOP (0x81798a0) sassign
1031 8 LISTOP (0x8179900) leave
1032
1033This probably makes more sense for a human: enter a block, start a
1034statement. Get the values of C<$b> and C<$c>, and add them together.
1035Find C<$a>, and assign one to the other. Then leave.
1036
1037The way Perl builds up these op trees in the parsing process can be
1038unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1039piece we need to construct the tree for C<$a = $b + $c>
1040
1041 1 term : term ASSIGNOP term
1042 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1043 3 | term ADDOP term
1044 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1045
1046If you're not used to reading BNF grammars, this is how it works: You're
1047fed certain things by the tokeniser, which generally end up in upper
1048case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1049code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
1050`terminal symbols', because you can't get any simpler than them.
1051
1052The grammar, lines one and three of the snippet above, tells you how to
1053build up more complex forms. These complex forms, `non-terminal symbols'
1054are generally placed in lower case. C<term> here is a non-terminal
1055symbol, representing a single expression.
1056
1057The grammar gives you the following rule: you can make the thing on the
1058left of the colon if you see all the things on the right in sequence.
1059This is called a "reduction", and the aim of parsing is to completely
1060reduce the input. There are several different ways you can perform a
1061reduction, separated by vertical bars: so, C<term> followed by C<=>
1062followed by C<term> makes a C<term>, and C<term> followed by C<+>
1063followed by C<term> can also make a C<term>.
1064
1065So, if you see two terms with an C<=> or C<+>, between them, you can
1066turn them into a single expression. When you do this, you execute the
1067code in the block on the next line: if you see C<=>, you'll do the code
1068in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1069which contributes to the op tree.
1070
1071 | term ADDOP term
1072 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1073
1074What this does is creates a new binary op, and feeds it a number of
1075variables. The variables refer to the tokens: C<$1> is the first token in
1076the input, C<$2> the second, and so on - think regular expression
1077backreferences. C<$$> is the op returned from this reduction. So, we
1078call C<newBINOP> to create a new binary operator. The first parameter to
1079C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1080operator, so we want the type to be C<ADDOP>. We could specify this
1081directly, but it's right there as the second token in the input, so we
1082use C<$2>. The second parameter is the op's flags: 0 means `nothing
1083special'. Then the things to add: the left and right hand side of our
1084expression, in scalar context.
1085
1086=head2 Stacks
1087
1088When perl executes something like C<addop>, how does it pass on its
1089results to the next op? The answer is, through the use of stacks. Perl
1090has a number of stacks to store things it's currently working on, and
1091we'll look at the three most important ones here.
1092
1093=over 3
1094
1095=item Argument stack
1096
1097Arguments are passed to PP code and returned from PP code using the
1098argument stack, C<ST>. The typical way to handle arguments is to pop
1099them off the stack, deal with them how you wish, and then push the result
1100back onto the stack. This is how, for instance, the cosine operator
1101works:
1102
1103 NV value;
1104 value = POPn;
1105 value = Perl_cos(value);
1106 XPUSHn(value);
1107
1108We'll see a more tricky example of this when we consider Perl's macros
1109below. C<POPn> gives you the NV (floating point value) of the top SV on
1110the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1111the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1112should be extended if necessary - it can't be necessary here, because we
1113know there's room for one more item on the stack, since we've just
1114removed one! The C<XPUSH*> macros at least guarantee safety.
1115
1116Alternatively, you can fiddle with the stack directly: C<SP> gives you
1117the first element in your portion of the stack, and C<TOP*> gives you
1118the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1119negation of an integer:
1120
1121 SETi(-TOPi);
1122
1123Just set the integer value of the top stack entry to its negation.
1124
1125Argument stack manipulation in the core is exactly the same as it is in
1126XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1127description of the macros used in stack manipulation.
1128
1129=item Mark stack
1130
1131I say `your portion of the stack' above because PP code doesn't
1132necessarily get the whole stack to itself: if your function calls
1133another function, you'll only want to expose the arguments aimed for the
1134called function, and not (necessarily) let it get at your own data. The
1135way we do this is to have a `virtual' bottom-of-stack, exposed to each
1136function. The mark stack keeps bookmarks to locations in the argument
1137stack usable by each function. For instance, when dealing with a tied
1138variable, (internally, something with `P' magic) Perl has to call
1139methods for accesses to the tied variables. However, we need to separate
1140the arguments exposed to the method to the argument exposed to the
1141original function - the store or fetch or whatever it may be. Here's how
1142the tied C<push> is implemented; see C<av_push> in F<av.c>:
1143
1144 1 PUSHMARK(SP);
1145 2 EXTEND(SP,2);
1146 3 PUSHs(SvTIED_obj((SV*)av, mg));
1147 4 PUSHs(val);
1148 5 PUTBACK;
1149 6 ENTER;
1150 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1151 8 LEAVE;
1152 9 POPSTACK;
13a2d996 1153
a422fd2d 1154The lines which concern the mark stack are the first, fifth and last
1155lines: they save away, restore and remove the current position of the
1156argument stack.
1157
1158Let's examine the whole implementation, for practice:
1159
1160 1 PUSHMARK(SP);
1161
1162Push the current state of the stack pointer onto the mark stack. This is
1163so that when we've finished adding items to the argument stack, Perl
1164knows how many things we've added recently.
1165
1166 2 EXTEND(SP,2);
1167 3 PUSHs(SvTIED_obj((SV*)av, mg));
1168 4 PUSHs(val);
1169
1170We're going to add two more items onto the argument stack: when you have
1171a tied array, the C<PUSH> subroutine receives the object and the value
1172to be pushed, and that's exactly what we have here - the tied object,
1173retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1174
1175 5 PUTBACK;
1176
1177Next we tell Perl to make the change to the global stack pointer: C<dSP>
1178only gave us a local copy, not a reference to the global.
1179
1180 6 ENTER;
1181 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1182 8 LEAVE;
1183
1184C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1185variables are tidied up, everything that has been localised gets
1186its previous value returned, and so on. Think of them as the C<{> and
1187C<}> of a Perl block.
1188
1189To actually do the magic method call, we have to call a subroutine in
1190Perl space: C<call_method> takes care of that, and it's described in
1191L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1192going to discard its return value.
1193
1194 9 POPSTACK;
1195
1196Finally, we remove the value we placed on the mark stack, since we
1197don't need it any more.
1198
1199=item Save stack
1200
1201C doesn't have a concept of local scope, so perl provides one. We've
1202seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1203stack implements the C equivalent of, for example:
1204
1205 {
1206 local $foo = 42;
1207 ...
1208 }
1209
1210See L<perlguts/Localising Changes> for how to use the save stack.
1211
1212=back
1213
1214=head2 Millions of Macros
1215
1216One thing you'll notice about the Perl source is that it's full of
1217macros. Some have called the pervasive use of macros the hardest thing
1218to understand, others find it adds to clarity. Let's take an example,
1219the code which implements the addition operator:
1220
1221 1 PP(pp_add)
1222 2 {
39644a26 1223 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d 1224 4 {
1225 5 dPOPTOPnnrl_ul;
1226 6 SETn( left + right );
1227 7 RETURN;
1228 8 }
1229 9 }
1230
1231Every line here (apart from the braces, of course) contains a macro. The
1232first line sets up the function declaration as Perl expects for PP code;
1233line 3 sets up variable declarations for the argument stack and the
1234target, the return value of the operation. Finally, it tries to see if
1235the addition operation is overloaded; if so, the appropriate subroutine
1236is called.
1237
1238Line 5 is another variable declaration - all variable declarations start
1239with C<d> - which pops from the top of the argument stack two NVs (hence
1240C<nn>) and puts them into the variables C<right> and C<left>, hence the
1241C<rl>. These are the two operands to the addition operator. Next, we
1242call C<SETn> to set the NV of the return value to the result of adding
1243the two values. This done, we return - the C<RETURN> macro makes sure
1244that our return value is properly handled, and we pass the next operator
1245to run back to the main run loop.
1246
1247Most of these macros are explained in L<perlapi>, and some of the more
1248important ones are explained in L<perlxs> as well. Pay special attention
1249to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1250the C<[pad]THX_?> macros.
1251
1252
1253=head2 Poking at Perl
1254
1255To really poke around with Perl, you'll probably want to build Perl for
1256debugging, like this:
1257
1258 ./Configure -d -D optimize=-g
1259 make
1260
1261C<-g> is a flag to the C compiler to have it produce debugging
1262information which will allow us to step through a running program.
1263F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1264enables all the internal debugging code in Perl. There are a whole bunch
1265of things you can debug with this: L<perlrun> lists them all, and the
1266best way to find out about them is to play about with them. The most
1267useful options are probably
1268
1269 l Context (loop) stack processing
1270 t Trace execution
1271 o Method and overloading resolution
1272 c String/numeric conversions
1273
1274Some of the functionality of the debugging code can be achieved using XS
1275modules.
13a2d996 1276
a422fd2d 1277 -Dr => use re 'debug'
1278 -Dx => use O 'Debug'
1279
1280=head2 Using a source-level debugger
1281
1282If the debugging output of C<-D> doesn't help you, it's time to step
1283through perl's execution with a source-level debugger.
1284
1285=over 3
1286
1287=item *
1288
1289We'll use C<gdb> for our examples here; the principles will apply to any
1290debugger, but check the manual of the one you're using.
1291
1292=back
1293
1294To fire up the debugger, type
1295
1296 gdb ./perl
1297
1298You'll want to do that in your Perl source tree so the debugger can read
1299the source code. You should see the copyright message, followed by the
1300prompt.
1301
1302 (gdb)
1303
1304C<help> will get you into the documentation, but here are the most
1305useful commands:
1306
1307=over 3
1308
1309=item run [args]
1310
1311Run the program with the given arguments.
1312
1313=item break function_name
1314
1315=item break source.c:xxx
1316
1317Tells the debugger that we'll want to pause execution when we reach
cea6626f 1318either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d 1319line in the named source file.
1320
1321=item step
1322
1323Steps through the program a line at a time.
1324
1325=item next
1326
1327Steps through the program a line at a time, without descending into
1328functions.
1329
1330=item continue
1331
1332Run until the next breakpoint.
1333
1334=item finish
1335
1336Run until the end of the current function, then stop again.
1337
13a2d996 1338=item 'enter'
a422fd2d 1339
1340Just pressing Enter will do the most recent operation again - it's a
1341blessing when stepping through miles of source code.
1342
1343=item print
1344
1345Execute the given C code and print its results. B<WARNING>: Perl makes
1346heavy use of macros, and F<gdb> is not aware of macros. You'll have to
1347substitute them yourself. So, for instance, you can't say
1348
1349 print SvPV_nolen(sv)
1350
1351but you have to say
1352
1353 print Perl_sv_2pv_nolen(sv)
1354
1355You may find it helpful to have a "macro dictionary", which you can
1356produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1357recursively apply the macros for you.
1358
1359=back
1360
1361=head2 Dumping Perl Data Structures
1362
1363One way to get around this macro hell is to use the dumping functions in
1364F<dump.c>; these work a little like an internal
1365L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1366that you can't get at from Perl. Let's take an example. We'll use the
1367C<$a = $b + $c> we used before, but give it a bit of context:
1368C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1369
1370What about C<pp_add>, the function we examined earlier to implement the
1371C<+> operator:
1372
1373 (gdb) break Perl_pp_add
1374 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1375
cea6626f 1376Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d 1377With the breakpoint in place, we can run our program:
1378
1379 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1380
1381Lots of junk will go past as gdb reads in the relevant source files and
1382libraries, and then:
1383
1384 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1385 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d 1386 (gdb) step
1387 311 dPOPTOPnnrl_ul;
1388 (gdb)
1389
1390We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1391arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1392slightly expand it:
1393
1394 #define dPOPTOPnnrl_ul NV right = POPn; \
1395 SV *leftsv = TOPs; \
1396 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1397
1398C<POPn> takes the SV from the top of the stack and obtains its NV either
1399directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1400C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1401C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1402C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1403
1404Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1405convert it. If we step again, we'll find ourselves there:
1406
1407 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1408 1669 if (!sv)
1409 (gdb)
1410
1411We can now use C<Perl_sv_dump> to investigate the SV:
1412
1413 SV = PV(0xa057cc0) at 0xa0675d0
1414 REFCNT = 1
1415 FLAGS = (POK,pPOK)
1416 PV = 0xa06a510 "6XXXX"\0
1417 CUR = 5
1418 LEN = 6
1419 $1 = void
1420
1421We know we're going to get C<6> from this, so let's finish the
1422subroutine:
1423
1424 (gdb) finish
1425 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1426 0x462669 in Perl_pp_add () at pp_hot.c:311
1427 311 dPOPTOPnnrl_ul;
1428
1429We can also dump out this op: the current op is always stored in
1430C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1431similar output to L<B::Debug|B::Debug>.
1432
1433 {
1434 13 TYPE = add ===> 14
1435 TARG = 1
1436 FLAGS = (SCALAR,KIDS)
1437 {
1438 TYPE = null ===> (12)
1439 (was rv2sv)
1440 FLAGS = (SCALAR,KIDS)
1441 {
1442 11 TYPE = gvsv ===> 12
1443 FLAGS = (SCALAR)
1444 GV = main::b
1445 }
1446 }
1447
10f58044 1448# finish this later #
a422fd2d 1449
1450=head2 Patching
1451
1452All right, we've now had a look at how to navigate the Perl sources and
1453some things you'll need to know when fiddling with them. Let's now get
1454on and create a simple patch. Here's something Larry suggested: if a
1455C<U> is the first active format during a C<pack>, (for example,
1456C<pack "U3C8", @stuff>) then the resulting string should be treated as
1457UTF8 encoded.
1458
1459How do we prepare to fix this up? First we locate the code in question -
1460the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1461files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1462altering this file, let's copy it to F<pp.c~>.
1463
a6ec74c1 1464[Well, it was in F<pp.c> when this tutorial was written. It has now been
1465split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1466
a422fd2d 1467Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1468loop over the pattern, taking each format character in turn into
1469C<datum_type>. Then for each possible format character, we swallow up
1470the other arguments in the pattern (a field width, an asterisk, and so
1471on) and convert the next chunk input into the specified format, adding
1472it onto the output SV C<cat>.
1473
1474How do we know if the C<U> is the first format in the C<pat>? Well, if
1475we have a pointer to the start of C<pat> then, if we see a C<U> we can
1476test whether we're still at the start of the string. So, here's where
1477C<pat> is set up:
1478
1479 STRLEN fromlen;
1480 register char *pat = SvPVx(*++MARK, fromlen);
1481 register char *patend = pat + fromlen;
1482 register I32 len;
1483 I32 datumtype;
1484 SV *fromstr;
1485
1486We'll have another string pointer in there:
1487
1488 STRLEN fromlen;
1489 register char *pat = SvPVx(*++MARK, fromlen);
1490 register char *patend = pat + fromlen;
1491 + char *patcopy;
1492 register I32 len;
1493 I32 datumtype;
1494 SV *fromstr;
1495
1496And just before we start the loop, we'll set C<patcopy> to be the start
1497of C<pat>:
1498
1499 items = SP - MARK;
1500 MARK++;
1501 sv_setpvn(cat, "", 0);
1502 + patcopy = pat;
1503 while (pat < patend) {
1504
1505Now if we see a C<U> which was at the start of the string, we turn on
1506the UTF8 flag for the output SV, C<cat>:
1507
1508 + if (datumtype == 'U' && pat==patcopy+1)
1509 + SvUTF8_on(cat);
1510 if (datumtype == '#') {
1511 while (pat < patend && *pat != '\n')
1512 pat++;
1513
1514Remember that it has to be C<patcopy+1> because the first character of
1515the string is the C<U> which has been swallowed into C<datumtype!>
1516
1517Oops, we forgot one thing: what if there are spaces at the start of the
1518pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1519character, even though it's not the first thing in the pattern. In this
1520case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1521
1522 if (isSPACE(datumtype))
1523 continue;
1524
1525needs to become
1526
1527 if (isSPACE(datumtype)) {
1528 patcopy++;
1529 continue;
1530 }
1531
1532OK. That's the C part done. Now we must do two additional things before
1533this patch is ready to go: we've changed the behaviour of Perl, and so
1534we must document that change. We must also provide some more regression
1535tests to make sure our patch works and doesn't create a bug somewhere
1536else along the line.
1537
b23b8711 1538The regression tests for each operator live in F<t/op/>, and so we
1539make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1540tests to the end. First, we'll test that the C<U> does indeed create
1541Unicode strings.
1542
1543t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1544use the one from t/test.pl.
b23b8711 1545
35c336e6 1546 require './test.pl';
1547 plan( tests => 159 );
b23b8711 1548
1549so instead of this:
a422fd2d 1550
1551 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1552 print "ok $test\n"; $test++;
1553
35c336e6 1554we can write the more sensible (see L<Test::More> for a full
1555explanation of is() and other testing functions).
b23b8711 1556
35c336e6 1557 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
812f5127 1558 "U* produces unicode" );
b23b8711 1559
a422fd2d 1560Now we'll test that we got that space-at-the-beginning business right:
1561
35c336e6 1562 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1563 " with spaces at the beginning" );
a422fd2d 1564
1565And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1566the first active format:
1567
35c336e6 1568 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
812f5127 1569 "U* not first isn't unicode" );
a422fd2d 1570
35c336e6 1571Mustn't forget to change the number of tests which appears at the top,
1572or else the automated tester will get confused. This will either look
1573like this:
a422fd2d 1574
35c336e6 1575 print "1..156\n";
1576
1577or this:
1578
1579 plan( tests => 156 );
a422fd2d 1580
1581We now compile up Perl, and run it through the test suite. Our new
1582tests pass, hooray!
1583
1584Finally, the documentation. The job is never done until the paperwork is
1585over, so let's describe the change we've just made. The relevant place
1586is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1587this text in the description of C<pack>:
1588
1589 =item *
1590
1591 If the pattern begins with a C<U>, the resulting string will be treated
1592 as Unicode-encoded. You can force UTF8 encoding on in a string with an
1593 initial C<U0>, and the bytes that follow will be interpreted as Unicode
1594 characters. If you don't want this to happen, you can begin your pattern
1595 with C<C0> (or anything else) to force Perl not to UTF8 encode your
1596 string, and then follow this with a C<U*> somewhere in your pattern.
1597
1598All done. Now let's create the patch. F<Porting/patching.pod> tells us
1599that if we're making major changes, we should copy the entire directory
1600to somewhere safe before we begin fiddling, and then do
13a2d996 1601
a422fd2d 1602 diff -ruN old new > patch
1603
1604However, we know which files we've changed, and we can simply do this:
1605
1606 diff -u pp.c~ pp.c > patch
1607 diff -u t/op/pack.t~ t/op/pack.t >> patch
1608 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1609
1610We end up with a patch looking a little like this:
1611
1612 --- pp.c~ Fri Jun 02 04:34:10 2000
1613 +++ pp.c Fri Jun 16 11:37:25 2000
1614 @@ -4375,6 +4375,7 @@
1615 register I32 items;
1616 STRLEN fromlen;
1617 register char *pat = SvPVx(*++MARK, fromlen);
1618 + char *patcopy;
1619 register char *patend = pat + fromlen;
1620 register I32 len;
1621 I32 datumtype;
1622 @@ -4405,6 +4406,7 @@
1623 ...
1624
1625And finally, we submit it, with our rationale, to perl5-porters. Job
1626done!
1627
f7e1e956 1628=head2 Patching a core module
1629
1630This works just like patching anything else, with an extra
1631consideration. Many core modules also live on CPAN. If this is so,
1632patch the CPAN version instead of the core and send the patch off to
1633the module maintainer (with a copy to p5p). This will help the module
1634maintainer keep the CPAN version in sync with the core version without
1635constantly scanning p5p.
1636
acbe17fc 1637=head2 Adding a new function to the core
1638
1639If, as part of a patch to fix a bug, or just because you have an
1640especially good idea, you decide to add a new function to the core,
1641discuss your ideas on p5p well before you start work. It may be that
1642someone else has already attempted to do what you are considering and
1643can give lots of good advice or even provide you with bits of code
1644that they already started (but never finished).
1645
1646You have to follow all of the advice given above for patching. It is
1647extremely important to test any addition thoroughly and add new tests
1648to explore all boundary conditions that your new function is expected
1649to handle. If your new function is used only by one module (e.g. toke),
1650then it should probably be named S_your_function (for static); on the
210b36aa 1651other hand, if you expect it to accessible from other functions in
acbe17fc 1652Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1653for more details.
1654
1655The location of any new code is also an important consideration. Don't
1656just create a new top level .c file and put your code there; you would
1657have to make changes to Configure (so the Makefile is created properly),
1658as well as possibly lots of include files. This is strictly pumpking
1659business.
1660
1661It is better to add your function to one of the existing top level
1662source code files, but your choice is complicated by the nature of
1663the Perl distribution. Only the files that are marked as compiled
1664static are located in the perl executable. Everything else is located
1665in the shared library (or DLL if you are running under WIN32). So,
1666for example, if a function was only used by functions located in
1667toke.c, then your code can go in toke.c. If, however, you want to call
1668the function from universal.c, then you should put your code in another
1669location, for example util.c.
1670
1671In addition to writing your c-code, you will need to create an
1672appropriate entry in embed.pl describing your function, then run
1673'make regen_headers' to create the entries in the numerous header
1674files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1675for information on the various options that you can set in embed.pl.
1676You will forget to do this a few (or many) times and you will get
1677warnings during the compilation phase. Make sure that you mention
1678this when you post your patch to P5P; the pumpking needs to know this.
1679
1680When you write your new code, please be conscious of existing code
1681conventions used in the perl source files. See <perlstyle> for
1682details. Although most of the guidelines discussed seem to focus on
1683Perl code, rather than c, they all apply (except when they don't ;).
1684See also I<Porting/patching.pod> file in the Perl source distribution
1685for lots of details about both formatting and submitting patches of
1686your changes.
1687
1688Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1689Test on as many platforms as you can find. Test as many perl
1690Configure options as you can (e.g. MULTIPLICITY). If you have
1691profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 1692below for how to use them to further test your code. Remember that
acbe17fc 1693most of the people on P5P are doing this on their own time and
1694don't have the time to debug your code.
f7e1e956 1695
1696=head2 Writing a test
1697
1698Every module and built-in function has an associated test file (or
1699should...). If you add or change functionality, you have to write a
1700test. If you fix a bug, you have to write a test so that bug never
1701comes back. If you alter the docs, it would be nice to test what the
1702new documentation says.
1703
1704In short, if you submit a patch you probably also have to patch the
1705tests.
1706
1707For modules, the test file is right next to the module itself.
1708F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1709so there are some snags (and it would be wonderful for you to brush
1710them out), but it basically works that way. Everything else lives in
1711F<t/>.
1712
1713=over 3
1714
1715=item F<t/base/>
1716
1717Testing of the absolute basic functionality of Perl. Things like
1718C<if>, basic file reads and writes, simple regexes, etc. These are
1719run first in the test suite and if any of them fail, something is
1720I<really> broken.
1721
1722=item F<t/cmd/>
1723
1724These test the basic control structures, C<if/else>, C<while>,
35c336e6 1725subroutines, etc.
f7e1e956 1726
1727=item F<t/comp/>
1728
1729Tests basic issues of how Perl parses and compiles itself.
1730
1731=item F<t/io/>
1732
1733Tests for built-in IO functions, including command line arguments.
1734
1735=item F<t/lib/>
1736
1737The old home for the module tests, you shouldn't put anything new in
1738here. There are still some bits and pieces hanging around in here
1739that need to be moved. Perhaps you could move them? Thanks!
1740
1741=item F<t/op/>
1742
1743Tests for perl's built in functions that don't fit into any of the
1744other directories.
1745
1746=item F<t/pod/>
1747
1748Tests for POD directives. There are still some tests for the Pod
1749modules hanging around in here that need to be moved out into F<lib/>.
1750
1751=item F<t/run/>
1752
1753Testing features of how perl actually runs, including exit codes and
1754handling of PERL* environment variables.
1755
1756=back
1757
1758The core uses the same testing style as the rest of Perl, a simple
1759"ok/not ok" run through Test::Harness, but there are a few special
1760considerations.
1761
35c336e6 1762There are three ways to write a test in the core. Test::More,
1763t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1764decision of which to use depends on what part of the test suite you're
1765working on. This is a measure to prevent a high-level failure (such
1766as Config.pm breaking) from causing basic functionality tests to fail.
1767
1768=over 4
1769
1770=item t/base t/comp
1771
1772Since we don't know if require works, or even subroutines, use ad hoc
1773tests for these two. Step carefully to avoid using the feature being
1774tested.
1775
1776=item t/cmd t/run t/io t/op
1777
1778Now that basic require() and subroutines are tested, you can use the
1779t/test.pl library which emulates the important features of Test::More
1780while using a minimum of core features.
1781
1782You can also conditionally use certain libraries like Config, but be
1783sure to skip the test gracefully if it's not there.
1784
1785=item t/lib ext lib
1786
1787Now that the core of Perl is tested, Test::More can be used. You can
1788also use the full suite of core modules in the tests.
1789
1790=back
f7e1e956 1791
1792When you say "make test" Perl uses the F<t/TEST> program to run the
1793test suite. All tests are run from the F<t/> directory, B<not> the
1794directory which contains the test. This causes some problems with the
1795tests in F<lib/>, so here's some opportunity for some patching.
1796
1797You must be triply conscious of cross-platform concerns. This usually
1798boils down to using File::Spec and avoiding things like C<fork()> and
1799C<system()> unless absolutely necessary.
1800
1801
902b9dbf 1802=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
1803
1804Sometimes it helps to use external tools while debugging and
1805testing Perl. This section tries to guide you through using
1806some common testing and debugging tools with Perl. This is
1807meant as a guide to interfacing these tools with Perl, not
1808as any kind of guide to the use of the tools themselves.
1809
1810=head2 Rational Software's Purify
1811
1812Purify is a commercial tool that is helpful in identifying
1813memory overruns, wild pointers, memory leaks and other such
1814badness. Perl must be compiled in a specific way for
1815optimal testing with Purify. Purify is available under
1816Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
1817
1818The only currently known leaks happen when there are
1819compile-time errors within eval or require. (Fixing these
1820is non-trivial, unfortunately, but they must be fixed
1821eventually.)
1822
1823=head2 Purify on Unix
1824
1825On Unix, Purify creates a new Perl binary. To get the most
1826benefit out of Purify, you should create the perl to Purify
1827using:
1828
1829 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
1830 -Uusemymalloc -Dusemultiplicity
1831
1832where these arguments mean:
1833
1834=over 4
1835
1836=item -Accflags=-DPURIFY
1837
1838Disables Perl's arena memory allocation functions, as well as
1839forcing use of memory allocation functions derived from the
1840system malloc.
1841
1842=item -Doptimize='-g'
1843
1844Adds debugging information so that you see the exact source
1845statements where the problem occurs. Without this flag, all
1846you will see is the source filename of where the error occurred.
1847
1848=item -Uusemymalloc
1849
1850Disable Perl's malloc so that Purify can more closely monitor
1851allocations and leaks. Using Perl's malloc will make Purify
1852report most leaks in the "potential" leaks category.
1853
1854=item -Dusemultiplicity
1855
1856Enabling the multiplicity option allows perl to clean up
1857thoroughly when the interpreter shuts down, which reduces the
1858number of bogus leak reports from Purify.
1859
1860=back
1861
1862Once you've compiled a perl suitable for Purify'ing, then you
1863can just:
1864
1865 make pureperl
1866
1867which creates a binary named 'pureperl' that has been Purify'ed.
1868This binary is used in place of the standard 'perl' binary
1869when you want to debug Perl memory problems.
1870
1871As an example, to show any memory leaks produced during the
1872standard Perl testset you would create and run the Purify'ed
1873perl as:
1874
1875 make pureperl
1876 cd t
1877 ../pureperl -I../lib harness
1878
1879which would run Perl on test.pl and report any memory problems.
1880
1881Purify outputs messages in "Viewer" windows by default. If
1882you don't have a windowing environment or if you simply
1883want the Purify output to unobtrusively go to a log file
1884instead of to the interactive window, use these following
1885options to output to the log file "perl.log":
1886
1887 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
1888 -log-file=perl.log -append-logfile=yes"
1889
1890If you plan to use the "Viewer" windows, then you only need this option:
1891
1892 setenv PURIFYOPTIONS "-chain-length=25"
1893
c406981e 1894In Bourne-type shells:
1895
1896 PURIFY_OPTIONS="..."
1897 export PURIFY_OPTIONS
1898
1899or if you have the "env" utility:
1900
1901 env PURIFY_OPTIONS="..." ../pureperl ...
1902
902b9dbf 1903=head2 Purify on NT
1904
1905Purify on Windows NT instruments the Perl binary 'perl.exe'
1906on the fly. There are several options in the makefile you
1907should change to get the most use out of Purify:
1908
1909=over 4
1910
1911=item DEFINES
1912
1913You should add -DPURIFY to the DEFINES line so the DEFINES
1914line looks something like:
1915
1916 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
1917
1918to disable Perl's arena memory allocation functions, as
1919well as to force use of memory allocation functions derived
1920from the system malloc.
1921
1922=item USE_MULTI = define
1923
1924Enabling the multiplicity option allows perl to clean up
1925thoroughly when the interpreter shuts down, which reduces the
1926number of bogus leak reports from Purify.
1927
1928=item #PERL_MALLOC = define
1929
1930Disable Perl's malloc so that Purify can more closely monitor
1931allocations and leaks. Using Perl's malloc will make Purify
1932report most leaks in the "potential" leaks category.
1933
1934=item CFG = Debug
1935
1936Adds debugging information so that you see the exact source
1937statements where the problem occurs. Without this flag, all
1938you will see is the source filename of where the error occurred.
1939
1940=back
1941
1942As an example, to show any memory leaks produced during the
1943standard Perl testset you would create and run Purify as:
1944
1945 cd win32
1946 make
1947 cd ../t
1948 purify ../perl -I../lib harness
1949
1950which would instrument Perl in memory, run Perl on test.pl,
1951then finally report any memory problems.
1952
09187cb1 1953=head2 Compaq's/Digital's Third Degree
1954
1955Third Degree is a tool for memory leak detection and memory access checks.
1956It is one of the many tools in the ATOM toolkit. The toolkit is only
1957available on Tru64 (formerly known as Digital UNIX formerly known as
1958DEC OSF/1).
1959
1960When building Perl, you must first run Configure with -Doptimize=-g
1961and -Uusemymalloc flags, after that you can use the make targets
51a35ef1 1962"perl.third" and "test.third". (What is required is that Perl must be
1963compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 1964
64cea5fd 1965The short story is that with "atom" you can instrument the Perl
83f0ef60 1966executable to create a new executable called F<perl.third>. When the
4ae3d70a 1967instrumented executable is run, it creates a log of dubious memory
83f0ef60 1968traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a 1969third for more information. The most extensive Third Degree
1970documentation is available in the Compaq "Tru64 UNIX Programmer's
1971Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 1972
83f0ef60 1973The "test.third" leaves a lot of files named F<perl.3log.*> in the t/
64cea5fd 1974subdirectory. There is a problem with these files: Third Degree is so
1975effective that it finds problems also in the system libraries.
83f0ef60 1976Therefore there are certain types of errors that you should ignore in
1977your debugging. Errors with stack traces matching
64cea5fd 1978
1979 __actual_atof|__catgets|_doprnt|__exc_|__exec|_findio|__localtime|setlocale|__sia_|__strxfrm
1980
1981(all in libc.so) are known to be non-serious. You can also
1982ignore the combinations
1983
1984 Perl_gv_fetchfile() calling strcpy()
1985 S_doopen_pmc() calling strcmp()
1986
1987causing "rih" (reading invalid heap) errors.
1988
1989There are also leaks that for given certain definition of a leak,
1990aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
1991
1992=head2 PERL_DESTRUCT_LEVEL
1993
1994If you want to run any of the tests yourself manually using the
1995pureperl or perl.third executables, please note that by default
1996perl B<does not> explicitly cleanup all the memory it has allocated
1997(such as global memory arenas) but instead lets the exit() of
1998the whole program "take care" of such allocations, also known
1999as "global destruction of objects".
2000
2001There is a way to tell perl to do complete cleanup: set the
2002environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2003The t/TEST wrapper does set this to 2, and this is what you
2004need to do too, if you don't want to see the "global leaks":
2005
2006 PERL_DESTRUCT_LEVEL=2 ./perl.third t/foo/bar.t
09187cb1 2007
51a35ef1 2008=head2 Profiling
2009
2010Depending on your platform there are various of profiling Perl.
2011
2012There are two commonly used techniques of profiling executables:
10f58044 2013I<statistical time-sampling> and I<basic-block counting>.
51a35ef1 2014
2015The first method takes periodically samples of the CPU program
2016counter, and since the program counter can be correlated with the code
2017generated for functions, we get a statistical view of in which
2018functions the program is spending its time. The caveats are that very
2019small/fast functions have lower probability of showing up in the
2020profile, and that periodically interrupting the program (this is
2021usually done rather frequently, in the scale of milliseconds) imposes
2022an additional overhead that may skew the results. The first problem
2023can be alleviated by running the code for longer (in general this is a
2024good idea for profiling), the second problem is usually kept in guard
2025by the profiling tools themselves.
2026
10f58044 2027The second method divides up the generated code into I<basic blocks>.
51a35ef1 2028Basic blocks are sections of code that are entered only in the
2029beginning and exited only at the end. For example, a conditional jump
2030starts a basic block. Basic block profiling usually works by
10f58044 2031I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1 2032book-keeping code to the generated code. During the execution of the
2033code the basic block counters are then updated appropriately. The
2034caveat is that the added extra code can skew the results: again, the
2035profiling tools usually try to factor their own effects out of the
2036results.
2037
83f0ef60 2038=head2 Gprof Profiling
2039
51a35ef1 2040gprof is a profiling tool available in many UNIX platforms,
2041it uses F<statistical time-sampling>.
83f0ef60 2042
2043You can build a profiled version of perl called "perl.gprof" by
51a35ef1 2044invoking the make target "perl.gprof" (What is required is that Perl
2045must be compiled using the C<-pg> flag, you may need to re-Configure).
2046Running the profiled version of Perl will create an output file called
2047F<gmon.out> is created which contains the profiling data collected
2048during the execution.
83f0ef60 2049
2050The gprof tool can then display the collected data in various ways.
2051Usually gprof understands the following options:
2052
2053=over 4
2054
2055=item -a
2056
2057Suppress statically defined functions from the profile.
2058
2059=item -b
2060
2061Suppress the verbose descriptions in the profile.
2062
2063=item -e routine
2064
2065Exclude the given routine and its descendants from the profile.
2066
2067=item -f routine
2068
2069Display only the given routine and its descendants in the profile.
2070
2071=item -s
2072
2073Generate a summary file called F<gmon.sum> which then may be given
2074to subsequent gprof runs to accumulate data over several runs.
2075
2076=item -z
2077
2078Display routines that have zero usage.
2079
2080=back
2081
2082For more detailed explanation of the available commands and output
2083formats, see your own local documentation of gprof.
2084
51a35ef1 2085=head2 GCC gcov Profiling
2086
10f58044 2087Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1 2088for the GNU CC.
2089
2090You can build a profiled version of perl called F<perl.gcov> by
2091invoking the make target "perl.gcov" (what is required that Perl must
2092be compiled using gcc with the flags C<-fprofile-arcs
2093-ftest-coverage>, you may need to re-Configure).
2094
2095Running the profiled version of Perl will cause profile output to be
2096generated. For each source file an accompanying ".da" file will be
2097created.
2098
2099To display the results you use the "gcov" utility (which should
2100be installed if you have gcc 3.0 or newer installed). F<gcov> is
2101run on source code files, like this
2102
2103 gcov sv.c
2104
2105which will cause F<sv.c.gcov> to be created. The F<.gcov> files
2106contain the source code annotated with relative frequencies of
2107execution indicated by "#" markers.
2108
2109Useful options of F<gcov> include C<-b> which will summarise the
2110basic block, branch, and function call coverage, and C<-c> which
2111instead of relative frequencies will use the actual counts. For
2112more information on the use of F<gcov> and basic block profiling
2113with gcc, see the latest GNU CC manual, as of GCC 3.0 see
2114
2115 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
2116
2117and its section titled "8. gcov: a Test Coverage Program"
2118
2119 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
2120
4ae3d70a 2121=head2 Pixie Profiling
2122
51a35ef1 2123Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
2124UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 2125I<basic-block counting>.
4ae3d70a 2126
83f0ef60 2127You can build a profiled version of perl called F<perl.pixie> by
51a35ef1 2128invoking the make target "perl.pixie" (what is required is that Perl
2129must be compiled using the C<-g> flag, you may need to re-Configure).
2130
2131In Tru64 a file called F<perl.Addrs> will also be silently created,
2132this file contains the addresses of the basic blocks. Running the
2133profiled version of Perl will create a new file called "perl.Counts"
2134which contains the counts for the basic block for that particular
2135program execution.
4ae3d70a 2136
51a35ef1 2137To display the results you use the F<prof> utility. The exact
4ae3d70a 2138incantation depends on your operating system, "prof perl.Counts" in
2139IRIX, and "prof -pixie -all -L. perl" in Tru64.
2140
6c41479b 2141In IRIX the following prof options are available:
2142
2143=over 4
2144
2145=item -h
2146
2147Reports the most heavily used lines in descending order of use.
6e36760b 2148Useful for finding the hotspot lines.
6c41479b 2149
2150=item -l
2151
2152Groups lines by procedure, with procedures sorted in descending order of use.
2153Within a procedure, lines are listed in source order.
6e36760b 2154Useful for finding the hotspots of procedures.
6c41479b 2155
2156=back
2157
2158In Tru64 the following options are available:
2159
2160=over 4
2161
3958b146 2162=item -p[rocedures]
6c41479b 2163
3958b146 2164Procedures sorted in descending order by the number of cycles executed
6e36760b 2165in each procedure. Useful for finding the hotspot procedures.
6c41479b 2166(This is the default option.)
2167
24000d2f 2168=item -h[eavy]
6c41479b 2169
6e36760b 2170Lines sorted in descending order by the number of cycles executed in
2171each line. Useful for finding the hotspot lines.
6c41479b 2172
24000d2f 2173=item -i[nvocations]
6c41479b 2174
6e36760b 2175The called procedures are sorted in descending order by number of calls
2176made to the procedures. Useful for finding the most used procedures.
6c41479b 2177
24000d2f 2178=item -l[ines]
6c41479b 2179
2180Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 2181Useful for finding the hotspots of procedures.
6c41479b 2182
2183=item -testcoverage
2184
2185The compiler emitted code for these lines, but the code was unexecuted.
2186
24000d2f 2187=item -z[ero]
6c41479b 2188
2189Unexecuted procedures.
2190
aa500c9e 2191=back
6c41479b 2192
2193For further information, see your system's manual pages for pixie and prof.
4ae3d70a 2194
a422fd2d 2195=head2 CONCLUSION
2196
2197We've had a brief look around the Perl source, an overview of the stages
2198F<perl> goes through when it's running your code, and how to use a
902b9dbf 2199debugger to poke at the Perl guts. We took a very simple problem and
2200demonstrated how to solve it fully - with documentation, regression
2201tests, and finally a patch for submission to p5p. Finally, we talked
2202about how to use external tools to debug and test Perl.
a422fd2d 2203
2204I'd now suggest you read over those references again, and then, as soon
2205as possible, get your hands dirty. The best way to learn is by doing,
2206so:
2207
2208=over 3
2209
2210=item *
2211
2212Subscribe to perl5-porters, follow the patches and try and understand
2213them; don't be afraid to ask if there's a portion you're not clear on -
2214who knows, you may unearth a bug in the patch...
2215
2216=item *
2217
2218Keep up to date with the bleeding edge Perl distributions and get
2219familiar with the changes. Try and get an idea of what areas people are
2220working on and the changes they're making.
2221
2222=item *
2223
3e148164 2224Do read the README associated with your operating system, e.g. README.aix
a1f349fd 2225on the IBM AIX OS. Don't hesitate to supply patches to that README if
2226you find anything missing or changed over a new OS release.
2227
2228=item *
2229
a422fd2d 2230Find an area of Perl that seems interesting to you, and see if you can
2231work out how it works. Scan through the source, and step over it in the
2232debugger. Play, poke, investigate, fiddle! You'll probably get to
2233understand not just your chosen area but a much wider range of F<perl>'s
2234activity as well, and probably sooner than you'd think.
2235
2236=back
2237
2238=over 3
2239
2240=item I<The Road goes ever on and on, down from the door where it began.>
2241
2242=back
2243
2244If you can do these things, you've started on the long road to Perl porting.
2245Thanks for wanting to help make Perl better - and happy hacking!
2246
e8cd7eae 2247=head1 AUTHOR
2248
2249This document was written by Nathan Torkington, and is maintained by
2250the perl5-porters mailing list.
2251