This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Build get_layers() also under -Uuseperlio.
[perl5.git] / pod / perlhack.pod
CommitLineData
e8cd7eae
GS
1=head1 NAME
2
3perlhack - How to hack at the Perl internals
4
5=head1 DESCRIPTION
6
7This document attempts to explain how Perl development takes place,
8and ends with some suggestions for people wanting to become bona fide
9porters.
10
11The perl5-porters mailing list is where the Perl standard distribution
12is maintained and developed. The list can get anywhere from 10 to 150
13messages a day, depending on the heatedness of the debate. Most days
14there are two or three patches, extensions, features, or bugs being
15discussed at a time.
16
f8e3975a 17A searchable archive of the list is at either:
e8cd7eae
GS
18
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
20
f8e3975a
IP
21or
22
23 http://archive.develooper.com/perl5-porters@perl.org/
24
e8cd7eae
GS
25List subscribers (the porters themselves) come in several flavours.
26Some are quiet curious lurkers, who rarely pitch in and instead watch
27the ongoing development to ensure they're forewarned of new changes or
28features in Perl. Some are representatives of vendors, who are there
29to make sure that Perl continues to compile and work on their
30platforms. Some patch any reported bug that they know how to fix,
31some are actively patching their pet area (threads, Win32, the regexp
32engine), while others seem to do nothing but complain. In other
33words, it's your usual mix of technical people.
34
35Over this group of porters presides Larry Wall. He has the final word
f6c51b38
GS
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
JL
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
GS
43
44In addition, various people are pumpkings for different things. For
45instance, Andy Dougherty and Jarkko Hietaniemi share the I<Configure>
caf100c0 46pumpkin.
e8cd7eae
GS
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
NC
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
NC
169thrown away by someone in the future?
170
e8cd7eae
GS
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
GS
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
GS
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
GS
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
MB
206=head2 Keeping in sync
207
e8cd7eae 208The source code to the Perl interpreter, in its different versions, is
f224927c
JH
209kept in a repository managed by a revision control system ( which is
210currently the Perforce program, see http://perforce.com/ ). The
e8cd7eae
GS
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
GS
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
0cfb3454 218 http://public.activestate.com/gsar/APC/
2be4c08b
GS
219 ftp://ftp.linux.activestate.com/pub/staff/gsar/APC/
220
0cfb3454
GS
221If you're looking for a particular change, or a change that affected
222a particular set of files, you may find the B<Perl Repository Browser>
223useful:
224
225 http://public.activestate.com/cgi-bin/perlbrowse
226
227You may also want to subscribe to the perl5-changes mailing list to
228receive a copy of each patch that gets submitted to the maintenance
229and development "branches" of the perl repository. See
230http://lists.perl.org/ for subscription information.
231
a1f349fd
MB
232If you are a member of the perl5-porters mailing list, it is a good
233thing to keep in touch with the most recent changes. If not only to
234verify if what you would have posted as a bug report isn't already
235solved in the most recent available perl development branch, also
236known as perl-current, bleading edge perl, bleedperl or bleadperl.
2be4c08b
GS
237
238Needless to say, the source code in perl-current is usually in a perpetual
239state of evolution. You should expect it to be very buggy. Do B<not> use
240it for any purpose other than testing and development.
e8cd7eae 241
3e148164
JH
242Keeping in sync with the most recent branch can be done in several ways,
243but the most convenient and reliable way is using B<rsync>, available at
244ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
245branch by FTP.)
a1f349fd
MB
246
247If you choose to keep in sync using rsync, there are two approaches
3e148164 248to doing so:
a1f349fd
MB
249
250=over 4
251
252=item rsync'ing the source tree
253
3e148164 254Presuming you are in the directory where your perl source resides
a1f349fd
MB
255and you have rsync installed and available, you can `upgrade' to
256the bleadperl using:
257
258 # rsync -avz rsync://ftp.linux.activestate.com/perl-current/ .
259
260This takes care of updating every single item in the source tree to
261the latest applied patch level, creating files that are new (to your
262distribution) and setting date/time stamps of existing files to
263reflect the bleadperl status.
264
c6d0653e
LC
265Note that this will not delete any files that were in '.' before
266the rsync. Once you are sure that the rsync is running correctly,
267run it with the --delete and the --dry-run options like this:
268
269 # rsync -avz --delete --dry-run rsync://ftp.linux.activestate.com/perl-current/ .
270
271This will I<simulate> an rsync run that also deletes files not
272present in the bleadperl master copy. Observe the results from
273this run closely. If you are sure that the actual run would delete
274no files precious to you, you could remove the '--dry-run' option.
275
a1f349fd
MB
276You can than check what patch was the latest that was applied by
277looking in the file B<.patch>, which will show the number of the
278latest patch.
279
280If you have more than one machine to keep in sync, and not all of
281them have access to the WAN (so you are not able to rsync all the
282source trees to the real source), there are some ways to get around
283this problem.
284
285=over 4
286
287=item Using rsync over the LAN
288
289Set up a local rsync server which makes the rsynced source tree
3e148164 290available to the LAN and sync the other machines against this
a1f349fd
MB
291directory.
292
1577cd80 293From http://rsync.samba.org/README.html :
a1f349fd
MB
294
295 "Rsync uses rsh or ssh for communication. It does not need to be
296 setuid and requires no special privileges for installation. It
3958b146 297 does not require an inetd entry or a daemon. You must, however,
a1f349fd
MB
298 have a working rsh or ssh system. Using ssh is recommended for
299 its security features."
300
301=item Using pushing over the NFS
302
303Having the other systems mounted over the NFS, you can take an
3e148164
JH
304active pushing approach by checking the just updated tree against
305the other not-yet synced trees. An example would be
306
307 #!/usr/bin/perl -w
308
309 use strict;
310 use File::Copy;
311
312 my %MF = map {
313 m/(\S+)/;
314 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
315 } `cat MANIFEST`;
316
317 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
318
319 foreach my $host (keys %remote) {
320 unless (-d $remote{$host}) {
321 print STDERR "Cannot Xsync for host $host\n";
322 next;
323 }
324 foreach my $file (keys %MF) {
325 my $rfile = "$remote{$host}/$file";
326 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
327 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
328 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
329 printf "%4s %-34s %8d %9d %8d %9d\n",
330 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
331 unlink $rfile;
332 copy ($file, $rfile);
333 utime time, $MF{$file}[2], $rfile;
334 chmod $MF{$file}[0], $rfile;
335 }
336 }
337
338though this is not perfect. It could be improved with checking
a1f349fd
MB
339file checksums before updating. Not all NFS systems support
340reliable utime support (when used over the NFS).
341
342=back
343
344=item rsync'ing the patches
345
346The source tree is maintained by the pumpking who applies patches to
347the files in the tree. These patches are either created by the
348pumpking himself using C<diff -c> after updating the file manually or
349by applying patches sent in by posters on the perl5-porters list.
350These patches are also saved and rsync'able, so you can apply them
351yourself to the source files.
352
353Presuming you are in a directory where your patches reside, you can
3e148164 354get them in sync with
a1f349fd
MB
355
356 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
357
358This makes sure the latest available patch is downloaded to your
359patch directory.
360
3e148164 361It's then up to you to apply these patches, using something like
a1f349fd 362
df3477ff 363 # last=`ls -t *.gz | sed q`
a1f349fd
MB
364 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
365 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
366 # cd ../perl-current
367 # patch -p1 -N <../perl-current-diffs/blead.patch
368
369or, since this is only a hint towards how it works, use CPAN-patchaperl
370from Andreas K├Ânig to have better control over the patching process.
371
372=back
373
f7e1e956 374=head2 Why rsync the source tree
a1f349fd
MB
375
376=over 4
377
10f58044 378=item It's easier to rsync the source tree
a1f349fd
MB
379
380Since you don't have to apply the patches yourself, you are sure all
381files in the source tree are in the right state.
382
a1f349fd
MB
383=item It's more reliable
384
0cfb3454
GS
385While both the rsync-able source and patch areas are automatically
386updated every few minutes, keep in mind that applying patches may
387sometimes mean careful hand-holding, especially if your version of
388the C<patch> program does not understand how to deal with new files,
389files with 8-bit characters, or files without trailing newlines.
a1f349fd
MB
390
391=back
392
f7e1e956 393=head2 Why rsync the patches
a1f349fd
MB
394
395=over 4
396
10f58044 397=item It's easier to rsync the patches
a1f349fd
MB
398
399If you have more than one machine that you want to keep in track with
3e148164 400bleadperl, it's easier to rsync the patches only once and then apply
a1f349fd
MB
401them to all the source trees on the different machines.
402
403In case you try to keep in pace on 5 different machines, for which
404only one of them has access to the WAN, rsync'ing all the source
3e148164 405trees should than be done 5 times over the NFS. Having
a1f349fd 406rsync'ed the patches only once, I can apply them to all the source
3e148164 407trees automatically. Need you say more ;-)
a1f349fd
MB
408
409=item It's a good reference
410
411If you do not only like to have the most recent development branch,
412but also like to B<fix> bugs, or extend features, you want to dive
413into the sources. If you are a seasoned perl core diver, you don't
414need no manuals, tips, roadmaps, perlguts.pod or other aids to find
415your way around. But if you are a starter, the patches may help you
416in finding where you should start and how to change the bits that
417bug you.
418
419The file B<Changes> is updated on occasions the pumpking sees as his
420own little sync points. On those occasions, he releases a tar-ball of
421the current source tree (i.e. perl@7582.tar.gz), which will be an
422excellent point to start with when choosing to use the 'rsync the
423patches' scheme. Starting with perl@7582, which means a set of source
424files on which the latest applied patch is number 7582, you apply all
f18956b7 425succeeding patches available from then on (7583, 7584, ...).
a1f349fd
MB
426
427You can use the patches later as a kind of search archive.
428
429=over 4
430
431=item Finding a start point
432
433If you want to fix/change the behaviour of function/feature Foo, just
434scan the patches for patches that mention Foo either in the subject,
3e148164 435the comments, or the body of the fix. A good chance the patch shows
a1f349fd
MB
436you the files that are affected by that patch which are very likely
437to be the starting point of your journey into the guts of perl.
438
439=item Finding how to fix a bug
440
441If you've found I<where> the function/feature Foo misbehaves, but you
442don't know how to fix it (but you do know the change you want to
443make), you can, again, peruse the patches for similar changes and
444look how others apply the fix.
445
446=item Finding the source of misbehaviour
447
448When you keep in sync with bleadperl, the pumpking would love to
3958b146 449I<see> that the community efforts really work. So after each of his
a1f349fd
MB
450sync points, you are to 'make test' to check if everything is still
451in working order. If it is, you do 'make ok', which will send an OK
452report to perlbug@perl.org. (If you do not have access to a mailer
3e148164 453from the system you just finished successfully 'make test', you can
a1f349fd
MB
454do 'make okfile', which creates the file C<perl.ok>, which you can
455than take to your favourite mailer and mail yourself).
456
3958b146 457But of course, as always, things will not always lead to a success
a1f349fd
MB
458path, and one or more test do not pass the 'make test'. Before
459sending in a bug report (using 'make nok' or 'make nokfile'), check
460the mailing list if someone else has reported the bug already and if
461so, confirm it by replying to that message. If not, you might want to
462trace the source of that misbehaviour B<before> sending in the bug,
463which will help all the other porters in finding the solution.
464
3e148164
JH
465Here the saved patches come in very handy. You can check the list of
466patches to see which patch changed what file and what change caused
467the misbehaviour. If you note that in the bug report, it saves the
468one trying to solve it, looking for that point.
a1f349fd
MB
469
470=back
471
472If searching the patches is too bothersome, you might consider using
473perl's bugtron to find more information about discussions and
474ramblings on posted bugs.
475
3e148164
JH
476If you want to get the best of both worlds, rsync both the source
477tree for convenience, reliability and ease and rsync the patches
478for reference.
479
52315700
RF
480=back
481
482
3fd28c4e 483=head2 Perlbug administration
52315700 484
3fd28c4e
RF
485There is a single remote administrative interface for modifying bug status,
486category, open issues etc. using the B<RT> I<bugtracker> system, maintained
487by I<Robert Spier>. Become an administrator, and close any bugs you can get
488your sticky mitts on:
52315700 489
3fd28c4e 490 http://rt.perl.org
52315700 491
3fd28c4e 492The bugtracker mechanism for B<perl5> bugs in particular is at:
52315700 493
3fd28c4e 494 http://bugs6.perl.org/perlbug
52315700 495
3fd28c4e 496To email the bug system administrators:
52315700 497
3fd28c4e 498 "perlbug-admin" <perlbug-admin@perl.org>
52315700 499
52315700 500
a1f349fd
MB
501=head2 Submitting patches
502
f7e1e956
MS
503Always submit patches to I<perl5-porters@perl.org>. If you're
504patching a core module and there's an author listed, send the author a
505copy (see L<Patching a core module>). This lets other porters review
506your patch, which catches a surprising number of errors in patches.
507Either use the diff program (available in source code form from
f224927c 508ftp://ftp.gnu.org/pub/gnu/ , or use Johan Vromans' I<makepatch>
f7e1e956
MS
509(available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
510but context diffs are accepted. Do not send RCS-style diffs or diffs
511without context lines. More information is given in the
512I<Porting/patching.pod> file in the Perl source distribution. Please
513patch against the latest B<development> version (e.g., if you're
514fixing a bug in the 5.005 track, patch against the latest 5.005_5x
515version). Only patches that survive the heat of the development
516branch get applied to maintenance versions.
517
518Your patch should update the documentation and test suite. See
519L<Writing a test>.
e8cd7eae
GS
520
521To report a bug in Perl, use the program I<perlbug> which comes with
522Perl (if you can't get Perl to work, send mail to the address
f18956b7 523I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
e8cd7eae 524I<perlbug> feeds into the automated bug-tracking system, access to
f224927c 525which is provided through the web at http://bugs.perl.org/ . It
e8cd7eae
GS
526often pays to check the archives of the perl5-porters mailing list to
527see whether the bug you're reporting has been reported before, and if
528so whether it was considered a bug. See above for the location of
529the searchable archives.
530
f224927c 531The CPAN testers ( http://testers.cpan.org/ ) are a group of
ba139f7d
JH
532volunteers who test CPAN modules on a variety of platforms. Perl
533Smokers ( http://archives.develooper.com/daily-build@perl.org/ )
534automatically tests Perl source releases on platforms with various
535configurations. Both efforts welcome volunteers.
e8cd7eae 536
e8cd7eae
GS
537It's a good idea to read and lurk for a while before chipping in.
538That way you'll get to see the dynamic of the conversations, learn the
539personalities of the players, and hopefully be better prepared to make
540a useful contribution when do you speak up.
541
542If after all this you still think you want to join the perl5-porters
f6c51b38
GS
543mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
544unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
e8cd7eae 545
a422fd2d
SC
546To hack on the Perl guts, you'll need to read the following things:
547
548=over 3
549
550=item L<perlguts>
551
552This is of paramount importance, since it's the documentation of what
553goes where in the Perl source. Read it over a couple of times and it
554might start to make sense - don't worry if it doesn't yet, because the
555best way to study it is to read it in conjunction with poking at Perl
556source, and we'll do that later on.
557
558You might also want to look at Gisle Aas's illustrated perlguts -
559there's no guarantee that this will be absolutely up-to-date with the
560latest documentation in the Perl core, but the fundamentals will be
1577cd80 561right. ( http://gisle.aas.no/perl/illguts/ )
a422fd2d
SC
562
563=item L<perlxstut> and L<perlxs>
564
565A working knowledge of XSUB programming is incredibly useful for core
566hacking; XSUBs use techniques drawn from the PP code, the portion of the
567guts that actually executes a Perl program. It's a lot gentler to learn
568those techniques from simple examples and explanation than from the core
569itself.
570
571=item L<perlapi>
572
573The documentation for the Perl API explains what some of the internal
574functions do, as well as the many macros used in the source.
575
576=item F<Porting/pumpkin.pod>
577
578This is a collection of words of wisdom for a Perl porter; some of it is
579only useful to the pumpkin holder, but most of it applies to anyone
580wanting to go about Perl development.
581
582=item The perl5-porters FAQ
583
7d7d5695
RGS
584This should be available from http://simon-cozens.org/writings/p5p-faq ;
585alternatively, you can get the FAQ emailed to you by sending mail to
586C<perl5-porters-faq@perl.org>. It contains hints on reading perl5-porters,
587information on how perl5-porters works and how Perl development in general
588works.
a422fd2d
SC
589
590=back
591
592=head2 Finding Your Way Around
593
594Perl maintenance can be split into a number of areas, and certain people
595(pumpkins) will have responsibility for each area. These areas sometimes
596correspond to files or directories in the source kit. Among the areas are:
597
598=over 3
599
600=item Core modules
601
602Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
603subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
604contains the core XS modules.
605
f7e1e956
MS
606=item Tests
607
608There are tests for nearly all the modules, built-ins and major bits
609of functionality. Test files all have a .t suffix. Module tests live
610in the F<lib/> and F<ext/> directories next to the module being
611tested. Others live in F<t/>. See L<Writing a test>
612
a422fd2d
SC
613=item Documentation
614
615Documentation maintenance includes looking after everything in the
616F<pod/> directory, (as well as contributing new documentation) and
617the documentation to the modules in core.
618
619=item Configure
620
621The configure process is the way we make Perl portable across the
622myriad of operating systems it supports. Responsibility for the
623configure, build and installation process, as well as the overall
624portability of the core code rests with the configure pumpkin - others
625help out with individual operating systems.
626
627The files involved are the operating system directories, (F<win32/>,
628F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
629and F<Makefile>, as well as the metaconfig files which generate
630F<Configure>. (metaconfig isn't included in the core distribution.)
631
632=item Interpreter
633
634And of course, there's the core of the Perl interpreter itself. Let's
635have a look at that in a little more detail.
636
637=back
638
639Before we leave looking at the layout, though, don't forget that
640F<MANIFEST> contains not only the file names in the Perl distribution,
641but short descriptions of what's in them, too. For an overview of the
642important files, try this:
643
644 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
645
646=head2 Elements of the interpreter
647
648The work of the interpreter has two main stages: compiling the code
649into the internal representation, or bytecode, and then executing it.
650L<perlguts/Compiled code> explains exactly how the compilation stage
651happens.
652
653Here is a short breakdown of perl's operation:
654
655=over 3
656
657=item Startup
658
659The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
660This is very high-level code, enough to fit on a single screen, and it
661resembles the code found in L<perlembed>; most of the real action takes
662place in F<perl.c>
663
664First, F<perlmain.c> allocates some memory and constructs a Perl
665interpreter:
666
667 1 PERL_SYS_INIT3(&argc,&argv,&env);
668 2
669 3 if (!PL_do_undump) {
670 4 my_perl = perl_alloc();
671 5 if (!my_perl)
672 6 exit(1);
673 7 perl_construct(my_perl);
674 8 PL_perl_destruct_level = 0;
675 9 }
676
677Line 1 is a macro, and its definition is dependent on your operating
678system. Line 3 references C<PL_do_undump>, a global variable - all
679global variables in Perl start with C<PL_>. This tells you whether the
680current running program was created with the C<-u> flag to perl and then
681F<undump>, which means it's going to be false in any sane context.
682
683Line 4 calls a function in F<perl.c> to allocate memory for a Perl
684interpreter. It's quite a simple function, and the guts of it looks like
685this:
686
687 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
688
689Here you see an example of Perl's system abstraction, which we'll see
690later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
691own C<malloc> as defined in F<malloc.c> if you selected that option at
692configure time.
693
694Next, in line 7, we construct the interpreter; this sets up all the
695special variables that Perl needs, the stacks, and so on.
696
697Now we pass Perl the command line options, and tell it to go:
698
699 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
700 if (!exitstatus) {
701 exitstatus = perl_run(my_perl);
702 }
703
704
705C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
706in F<perl.c>, which processes the command line options, sets up any
707statically linked XS modules, opens the program and calls C<yyparse> to
708parse it.
709
710=item Parsing
711
712The aim of this stage is to take the Perl source, and turn it into an op
713tree. We'll see what one of those looks like later. Strictly speaking,
714there's three things going on here.
715
716C<yyparse>, the parser, lives in F<perly.c>, although you're better off
717reading the original YACC input in F<perly.y>. (Yes, Virginia, there
718B<is> a YACC grammar for Perl!) The job of the parser is to take your
719code and `understand' it, splitting it into sentences, deciding which
720operands go with which operators and so on.
721
722The parser is nobly assisted by the lexer, which chunks up your input
723into tokens, and decides what type of thing each token is: a variable
724name, an operator, a bareword, a subroutine, a core function, and so on.
725The main point of entry to the lexer is C<yylex>, and that and its
726associated routines can be found in F<toke.c>. Perl isn't much like
727other computer languages; it's highly context sensitive at times, it can
728be tricky to work out what sort of token something is, or where a token
729ends. As such, there's a lot of interplay between the tokeniser and the
730parser, which can get pretty frightening if you're not used to it.
731
732As the parser understands a Perl program, it builds up a tree of
733operations for the interpreter to perform during execution. The routines
734which construct and link together the various operations are to be found
735in F<op.c>, and will be examined later.
736
737=item Optimization
738
739Now the parsing stage is complete, and the finished tree represents
740the operations that the Perl interpreter needs to perform to execute our
741program. Next, Perl does a dry run over the tree looking for
742optimisations: constant expressions such as C<3 + 4> will be computed
743now, and the optimizer will also see if any multiple operations can be
744replaced with a single one. For instance, to fetch the variable C<$foo>,
745instead of grabbing the glob C<*foo> and looking at the scalar
746component, the optimizer fiddles the op tree to use a function which
747directly looks up the scalar in question. The main optimizer is C<peep>
748in F<op.c>, and many ops have their own optimizing functions.
749
750=item Running
751
752Now we're finally ready to go: we have compiled Perl byte code, and all
753that's left to do is run it. The actual execution is done by the
754C<runops_standard> function in F<run.c>; more specifically, it's done by
755these three innocent looking lines:
756
757 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
758 PERL_ASYNC_CHECK();
759 }
760
761You may be more comfortable with the Perl version of that:
762
763 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
764
765Well, maybe not. Anyway, each op contains a function pointer, which
766stipulates the function which will actually carry out the operation.
767This function will return the next op in the sequence - this allows for
768things like C<if> which choose the next op dynamically at run time.
769The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
770execution if required.
771
772The actual functions called are known as PP code, and they're spread
773between four files: F<pp_hot.c> contains the `hot' code, which is most
774often used and highly optimized, F<pp_sys.c> contains all the
775system-specific functions, F<pp_ctl.c> contains the functions which
776implement control structures (C<if>, C<while> and the like) and F<pp.c>
777contains everything else. These are, if you like, the C code for Perl's
778built-in functions and operators.
779
780=back
781
782=head2 Internal Variable Types
783
784You should by now have had a look at L<perlguts>, which tells you about
785Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
786that now.
787
788These variables are used not only to represent Perl-space variables, but
789also any constants in the code, as well as some structures completely
790internal to Perl. The symbol table, for instance, is an ordinary Perl
791hash. Your code is represented by an SV as it's read into the parser;
792any program files you call are opened via ordinary Perl filehandles, and
793so on.
794
795The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
796Perl program. Let's see, for instance, how Perl treats the constant
797C<"hello">.
798
799 % perl -MDevel::Peek -e 'Dump("hello")'
800 1 SV = PV(0xa041450) at 0xa04ecbc
801 2 REFCNT = 1
802 3 FLAGS = (POK,READONLY,pPOK)
803 4 PV = 0xa0484e0 "hello"\0
804 5 CUR = 5
805 6 LEN = 6
806
807Reading C<Devel::Peek> output takes a bit of practise, so let's go
808through it line by line.
809
810Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
811memory. SVs themselves are very simple structures, but they contain a
812pointer to a more complex structure. In this case, it's a PV, a
813structure which holds a string value, at location C<0xa041450>. Line 2
814is the reference count; there are no other references to this data, so
815it's 1.
816
817Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
818read-only SV (because it's a constant) and the data is a PV internally.
819Next we've got the contents of the string, starting at location
820C<0xa0484e0>.
821
822Line 5 gives us the current length of the string - note that this does
823B<not> include the null terminator. Line 6 is not the length of the
824string, but the length of the currently allocated buffer; as the string
825grows, Perl automatically extends the available storage via a routine
826called C<SvGROW>.
827
828You can get at any of these quantities from C very easily; just add
829C<Sv> to the name of the field shown in the snippet, and you've got a
830macro which will return the value: C<SvCUR(sv)> returns the current
831length of the string, C<SvREFCOUNT(sv)> returns the reference count,
832C<SvPV(sv, len)> returns the string itself with its length, and so on.
833More macros to manipulate these properties can be found in L<perlguts>.
834
835Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
836
837 1 void
838 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
839 3 {
840 4 STRLEN tlen;
841 5 char *junk;
842
843 6 junk = SvPV_force(sv, tlen);
844 7 SvGROW(sv, tlen + len + 1);
845 8 if (ptr == junk)
846 9 ptr = SvPVX(sv);
847 10 Move(ptr,SvPVX(sv)+tlen,len,char);
848 11 SvCUR(sv) += len;
849 12 *SvEND(sv) = '\0';
850 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
851 14 SvTAINT(sv);
852 15 }
853
854This is a function which adds a string, C<ptr>, of length C<len> onto
855the end of the PV stored in C<sv>. The first thing we do in line 6 is
856make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
857macro to force a PV. As a side effect, C<tlen> gets set to the current
858value of the PV, and the PV itself is returned to C<junk>.
859
b1866b2d 860In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d
SC
861the old string, the new string and the null terminator. If C<LEN> isn't
862big enough, C<SvGROW> will reallocate space for us.
863
864Now, if C<junk> is the same as the string we're trying to add, we can
865grab the string directly from the SV; C<SvPVX> is the address of the PV
866in the SV.
867
868Line 10 does the actual catenation: the C<Move> macro moves a chunk of
869memory around: we move the string C<ptr> to the end of the PV - that's
870the start of the PV plus its current length. We're moving C<len> bytes
871of type C<char>. After doing so, we need to tell Perl we've extended the
872string, by altering C<CUR> to reflect the new length. C<SvEND> is a
873macro which gives us the end of the string, so that needs to be a
874C<"\0">.
875
876Line 13 manipulates the flags; since we've changed the PV, any IV or NV
877values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
878want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF8-aware
879version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
880and turns on POK. The final C<SvTAINT> is a macro which launders tainted
881data if taint mode is turned on.
882
883AVs and HVs are more complicated, but SVs are by far the most common
884variable type being thrown around. Having seen something of how we
885manipulate these, let's go on and look at how the op tree is
886constructed.
887
888=head2 Op Trees
889
890First, what is the op tree, anyway? The op tree is the parsed
891representation of your program, as we saw in our section on parsing, and
892it's the sequence of operations that Perl goes through to execute your
893program, as we saw in L</Running>.
894
895An op is a fundamental operation that Perl can perform: all the built-in
896functions and operators are ops, and there are a series of ops which
897deal with concepts the interpreter needs internally - entering and
898leaving a block, ending a statement, fetching a variable, and so on.
899
900The op tree is connected in two ways: you can imagine that there are two
901"routes" through it, two orders in which you can traverse the tree.
902First, parse order reflects how the parser understood the code, and
903secondly, execution order tells perl what order to perform the
904operations in.
905
906The easiest way to examine the op tree is to stop Perl after it has
907finished parsing, and get it to dump out the tree. This is exactly what
7d7d5695
RGS
908the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
909and L<B::Debug|B::Debug> do.
a422fd2d
SC
910
911Let's have a look at how Perl sees C<$a = $b + $c>:
912
913 % perl -MO=Terse -e '$a=$b+$c'
914 1 LISTOP (0x8179888) leave
915 2 OP (0x81798b0) enter
916 3 COP (0x8179850) nextstate
917 4 BINOP (0x8179828) sassign
918 5 BINOP (0x8179800) add [1]
919 6 UNOP (0x81796e0) null [15]
920 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
921 8 UNOP (0x81797e0) null [15]
922 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
923 10 UNOP (0x816b4f0) null [15]
924 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
925
926Let's start in the middle, at line 4. This is a BINOP, a binary
927operator, which is at location C<0x8179828>. The specific operator in
928question is C<sassign> - scalar assignment - and you can find the code
929which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
930binary operator, it has two children: the add operator, providing the
931result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
932line 10.
933
934Line 10 is the null op: this does exactly nothing. What is that doing
935there? If you see the null op, it's a sign that something has been
936optimized away after parsing. As we mentioned in L</Optimization>,
937the optimization stage sometimes converts two operations into one, for
938example when fetching a scalar variable. When this happens, instead of
939rewriting the op tree and cleaning up the dangling pointers, it's easier
940just to replace the redundant operation with the null op. Originally,
941the tree would have looked like this:
942
943 10 SVOP (0x816b4f0) rv2sv [15]
944 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
945
946That is, fetch the C<a> entry from the main symbol table, and then look
947at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
948happens to do both these things.
949
950The right hand side, starting at line 5 is similar to what we've just
951seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
952two C<gvsv>s.
953
954Now, what's this about?
955
956 1 LISTOP (0x8179888) leave
957 2 OP (0x81798b0) enter
958 3 COP (0x8179850) nextstate
959
960C<enter> and C<leave> are scoping ops, and their job is to perform any
961housekeeping every time you enter and leave a block: lexical variables
962are tidied up, unreferenced variables are destroyed, and so on. Every
963program will have those first three lines: C<leave> is a list, and its
964children are all the statements in the block. Statements are delimited
965by C<nextstate>, so a block is a collection of C<nextstate> ops, with
966the ops to be performed for each statement being the children of
967C<nextstate>. C<enter> is a single op which functions as a marker.
968
969That's how Perl parsed the program, from top to bottom:
970
971 Program
972 |
973 Statement
974 |
975 =
976 / \
977 / \
978 $a +
979 / \
980 $b $c
981
982However, it's impossible to B<perform> the operations in this order:
983you have to find the values of C<$b> and C<$c> before you add them
984together, for instance. So, the other thread that runs through the op
985tree is the execution order: each op has a field C<op_next> which points
986to the next op to be run, so following these pointers tells us how perl
987executes the code. We can traverse the tree in this order using
988the C<exec> option to C<B::Terse>:
989
990 % perl -MO=Terse,exec -e '$a=$b+$c'
991 1 OP (0x8179928) enter
992 2 COP (0x81798c8) nextstate
993 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
994 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
995 5 BINOP (0x8179878) add [1]
996 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
997 7 BINOP (0x81798a0) sassign
998 8 LISTOP (0x8179900) leave
999
1000This probably makes more sense for a human: enter a block, start a
1001statement. Get the values of C<$b> and C<$c>, and add them together.
1002Find C<$a>, and assign one to the other. Then leave.
1003
1004The way Perl builds up these op trees in the parsing process can be
1005unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1006piece we need to construct the tree for C<$a = $b + $c>
1007
1008 1 term : term ASSIGNOP term
1009 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1010 3 | term ADDOP term
1011 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1012
1013If you're not used to reading BNF grammars, this is how it works: You're
1014fed certain things by the tokeniser, which generally end up in upper
1015case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1016code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
1017`terminal symbols', because you can't get any simpler than them.
1018
1019The grammar, lines one and three of the snippet above, tells you how to
1020build up more complex forms. These complex forms, `non-terminal symbols'
1021are generally placed in lower case. C<term> here is a non-terminal
1022symbol, representing a single expression.
1023
1024The grammar gives you the following rule: you can make the thing on the
1025left of the colon if you see all the things on the right in sequence.
1026This is called a "reduction", and the aim of parsing is to completely
1027reduce the input. There are several different ways you can perform a
1028reduction, separated by vertical bars: so, C<term> followed by C<=>
1029followed by C<term> makes a C<term>, and C<term> followed by C<+>
1030followed by C<term> can also make a C<term>.
1031
1032So, if you see two terms with an C<=> or C<+>, between them, you can
1033turn them into a single expression. When you do this, you execute the
1034code in the block on the next line: if you see C<=>, you'll do the code
1035in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1036which contributes to the op tree.
1037
1038 | term ADDOP term
1039 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1040
1041What this does is creates a new binary op, and feeds it a number of
1042variables. The variables refer to the tokens: C<$1> is the first token in
1043the input, C<$2> the second, and so on - think regular expression
1044backreferences. C<$$> is the op returned from this reduction. So, we
1045call C<newBINOP> to create a new binary operator. The first parameter to
1046C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1047operator, so we want the type to be C<ADDOP>. We could specify this
1048directly, but it's right there as the second token in the input, so we
1049use C<$2>. The second parameter is the op's flags: 0 means `nothing
1050special'. Then the things to add: the left and right hand side of our
1051expression, in scalar context.
1052
1053=head2 Stacks
1054
1055When perl executes something like C<addop>, how does it pass on its
1056results to the next op? The answer is, through the use of stacks. Perl
1057has a number of stacks to store things it's currently working on, and
1058we'll look at the three most important ones here.
1059
1060=over 3
1061
1062=item Argument stack
1063
1064Arguments are passed to PP code and returned from PP code using the
1065argument stack, C<ST>. The typical way to handle arguments is to pop
1066them off the stack, deal with them how you wish, and then push the result
1067back onto the stack. This is how, for instance, the cosine operator
1068works:
1069
1070 NV value;
1071 value = POPn;
1072 value = Perl_cos(value);
1073 XPUSHn(value);
1074
1075We'll see a more tricky example of this when we consider Perl's macros
1076below. C<POPn> gives you the NV (floating point value) of the top SV on
1077the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1078the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1079should be extended if necessary - it can't be necessary here, because we
1080know there's room for one more item on the stack, since we've just
1081removed one! The C<XPUSH*> macros at least guarantee safety.
1082
1083Alternatively, you can fiddle with the stack directly: C<SP> gives you
1084the first element in your portion of the stack, and C<TOP*> gives you
1085the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1086negation of an integer:
1087
1088 SETi(-TOPi);
1089
1090Just set the integer value of the top stack entry to its negation.
1091
1092Argument stack manipulation in the core is exactly the same as it is in
1093XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1094description of the macros used in stack manipulation.
1095
1096=item Mark stack
1097
1098I say `your portion of the stack' above because PP code doesn't
1099necessarily get the whole stack to itself: if your function calls
1100another function, you'll only want to expose the arguments aimed for the
1101called function, and not (necessarily) let it get at your own data. The
1102way we do this is to have a `virtual' bottom-of-stack, exposed to each
1103function. The mark stack keeps bookmarks to locations in the argument
1104stack usable by each function. For instance, when dealing with a tied
1105variable, (internally, something with `P' magic) Perl has to call
1106methods for accesses to the tied variables. However, we need to separate
1107the arguments exposed to the method to the argument exposed to the
1108original function - the store or fetch or whatever it may be. Here's how
1109the tied C<push> is implemented; see C<av_push> in F<av.c>:
1110
1111 1 PUSHMARK(SP);
1112 2 EXTEND(SP,2);
1113 3 PUSHs(SvTIED_obj((SV*)av, mg));
1114 4 PUSHs(val);
1115 5 PUTBACK;
1116 6 ENTER;
1117 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1118 8 LEAVE;
1119 9 POPSTACK;
13a2d996 1120
a422fd2d
SC
1121The lines which concern the mark stack are the first, fifth and last
1122lines: they save away, restore and remove the current position of the
1123argument stack.
1124
1125Let's examine the whole implementation, for practice:
1126
1127 1 PUSHMARK(SP);
1128
1129Push the current state of the stack pointer onto the mark stack. This is
1130so that when we've finished adding items to the argument stack, Perl
1131knows how many things we've added recently.
1132
1133 2 EXTEND(SP,2);
1134 3 PUSHs(SvTIED_obj((SV*)av, mg));
1135 4 PUSHs(val);
1136
1137We're going to add two more items onto the argument stack: when you have
1138a tied array, the C<PUSH> subroutine receives the object and the value
1139to be pushed, and that's exactly what we have here - the tied object,
1140retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1141
1142 5 PUTBACK;
1143
1144Next we tell Perl to make the change to the global stack pointer: C<dSP>
1145only gave us a local copy, not a reference to the global.
1146
1147 6 ENTER;
1148 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1149 8 LEAVE;
1150
1151C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1152variables are tidied up, everything that has been localised gets
1153its previous value returned, and so on. Think of them as the C<{> and
1154C<}> of a Perl block.
1155
1156To actually do the magic method call, we have to call a subroutine in
1157Perl space: C<call_method> takes care of that, and it's described in
1158L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1159going to discard its return value.
1160
1161 9 POPSTACK;
1162
1163Finally, we remove the value we placed on the mark stack, since we
1164don't need it any more.
1165
1166=item Save stack
1167
1168C doesn't have a concept of local scope, so perl provides one. We've
1169seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1170stack implements the C equivalent of, for example:
1171
1172 {
1173 local $foo = 42;
1174 ...
1175 }
1176
1177See L<perlguts/Localising Changes> for how to use the save stack.
1178
1179=back
1180
1181=head2 Millions of Macros
1182
1183One thing you'll notice about the Perl source is that it's full of
1184macros. Some have called the pervasive use of macros the hardest thing
1185to understand, others find it adds to clarity. Let's take an example,
1186the code which implements the addition operator:
1187
1188 1 PP(pp_add)
1189 2 {
39644a26 1190 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1191 4 {
1192 5 dPOPTOPnnrl_ul;
1193 6 SETn( left + right );
1194 7 RETURN;
1195 8 }
1196 9 }
1197
1198Every line here (apart from the braces, of course) contains a macro. The
1199first line sets up the function declaration as Perl expects for PP code;
1200line 3 sets up variable declarations for the argument stack and the
1201target, the return value of the operation. Finally, it tries to see if
1202the addition operation is overloaded; if so, the appropriate subroutine
1203is called.
1204
1205Line 5 is another variable declaration - all variable declarations start
1206with C<d> - which pops from the top of the argument stack two NVs (hence
1207C<nn>) and puts them into the variables C<right> and C<left>, hence the
1208C<rl>. These are the two operands to the addition operator. Next, we
1209call C<SETn> to set the NV of the return value to the result of adding
1210the two values. This done, we return - the C<RETURN> macro makes sure
1211that our return value is properly handled, and we pass the next operator
1212to run back to the main run loop.
1213
1214Most of these macros are explained in L<perlapi>, and some of the more
1215important ones are explained in L<perlxs> as well. Pay special attention
1216to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1217the C<[pad]THX_?> macros.
1218
a422fd2d
SC
1219=head2 Poking at Perl
1220
1221To really poke around with Perl, you'll probably want to build Perl for
1222debugging, like this:
1223
1224 ./Configure -d -D optimize=-g
1225 make
1226
1227C<-g> is a flag to the C compiler to have it produce debugging
1228information which will allow us to step through a running program.
1229F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1230enables all the internal debugging code in Perl. There are a whole bunch
1231of things you can debug with this: L<perlrun> lists them all, and the
1232best way to find out about them is to play about with them. The most
1233useful options are probably
1234
1235 l Context (loop) stack processing
1236 t Trace execution
1237 o Method and overloading resolution
1238 c String/numeric conversions
1239
1240Some of the functionality of the debugging code can be achieved using XS
1241modules.
13a2d996 1242
a422fd2d
SC
1243 -Dr => use re 'debug'
1244 -Dx => use O 'Debug'
1245
1246=head2 Using a source-level debugger
1247
1248If the debugging output of C<-D> doesn't help you, it's time to step
1249through perl's execution with a source-level debugger.
1250
1251=over 3
1252
1253=item *
1254
1255We'll use C<gdb> for our examples here; the principles will apply to any
1256debugger, but check the manual of the one you're using.
1257
1258=back
1259
1260To fire up the debugger, type
1261
1262 gdb ./perl
1263
1264You'll want to do that in your Perl source tree so the debugger can read
1265the source code. You should see the copyright message, followed by the
1266prompt.
1267
1268 (gdb)
1269
1270C<help> will get you into the documentation, but here are the most
1271useful commands:
1272
1273=over 3
1274
1275=item run [args]
1276
1277Run the program with the given arguments.
1278
1279=item break function_name
1280
1281=item break source.c:xxx
1282
1283Tells the debugger that we'll want to pause execution when we reach
cea6626f 1284either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d
SC
1285line in the named source file.
1286
1287=item step
1288
1289Steps through the program a line at a time.
1290
1291=item next
1292
1293Steps through the program a line at a time, without descending into
1294functions.
1295
1296=item continue
1297
1298Run until the next breakpoint.
1299
1300=item finish
1301
1302Run until the end of the current function, then stop again.
1303
13a2d996 1304=item 'enter'
a422fd2d
SC
1305
1306Just pressing Enter will do the most recent operation again - it's a
1307blessing when stepping through miles of source code.
1308
1309=item print
1310
1311Execute the given C code and print its results. B<WARNING>: Perl makes
1312heavy use of macros, and F<gdb> is not aware of macros. You'll have to
1313substitute them yourself. So, for instance, you can't say
1314
1315 print SvPV_nolen(sv)
1316
1317but you have to say
1318
1319 print Perl_sv_2pv_nolen(sv)
1320
1321You may find it helpful to have a "macro dictionary", which you can
1322produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1323recursively apply the macros for you.
1324
1325=back
1326
1327=head2 Dumping Perl Data Structures
1328
1329One way to get around this macro hell is to use the dumping functions in
1330F<dump.c>; these work a little like an internal
1331L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1332that you can't get at from Perl. Let's take an example. We'll use the
1333C<$a = $b + $c> we used before, but give it a bit of context:
1334C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1335
1336What about C<pp_add>, the function we examined earlier to implement the
1337C<+> operator:
1338
1339 (gdb) break Perl_pp_add
1340 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1341
cea6626f 1342Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d
SC
1343With the breakpoint in place, we can run our program:
1344
1345 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1346
1347Lots of junk will go past as gdb reads in the relevant source files and
1348libraries, and then:
1349
1350 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1351 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1352 (gdb) step
1353 311 dPOPTOPnnrl_ul;
1354 (gdb)
1355
1356We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1357arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1358slightly expand it:
1359
1360 #define dPOPTOPnnrl_ul NV right = POPn; \
1361 SV *leftsv = TOPs; \
1362 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1363
1364C<POPn> takes the SV from the top of the stack and obtains its NV either
1365directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1366C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1367C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1368C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1369
1370Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1371convert it. If we step again, we'll find ourselves there:
1372
1373 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1374 1669 if (!sv)
1375 (gdb)
1376
1377We can now use C<Perl_sv_dump> to investigate the SV:
1378
1379 SV = PV(0xa057cc0) at 0xa0675d0
1380 REFCNT = 1
1381 FLAGS = (POK,pPOK)
1382 PV = 0xa06a510 "6XXXX"\0
1383 CUR = 5
1384 LEN = 6
1385 $1 = void
1386
1387We know we're going to get C<6> from this, so let's finish the
1388subroutine:
1389
1390 (gdb) finish
1391 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1392 0x462669 in Perl_pp_add () at pp_hot.c:311
1393 311 dPOPTOPnnrl_ul;
1394
1395We can also dump out this op: the current op is always stored in
1396C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1397similar output to L<B::Debug|B::Debug>.
1398
1399 {
1400 13 TYPE = add ===> 14
1401 TARG = 1
1402 FLAGS = (SCALAR,KIDS)
1403 {
1404 TYPE = null ===> (12)
1405 (was rv2sv)
1406 FLAGS = (SCALAR,KIDS)
1407 {
1408 11 TYPE = gvsv ===> 12
1409 FLAGS = (SCALAR)
1410 GV = main::b
1411 }
1412 }
1413
10f58044 1414# finish this later #
a422fd2d
SC
1415
1416=head2 Patching
1417
1418All right, we've now had a look at how to navigate the Perl sources and
1419some things you'll need to know when fiddling with them. Let's now get
1420on and create a simple patch. Here's something Larry suggested: if a
1421C<U> is the first active format during a C<pack>, (for example,
1422C<pack "U3C8", @stuff>) then the resulting string should be treated as
1423UTF8 encoded.
1424
1425How do we prepare to fix this up? First we locate the code in question -
1426the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1427files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1428altering this file, let's copy it to F<pp.c~>.
1429
a6ec74c1
JH
1430[Well, it was in F<pp.c> when this tutorial was written. It has now been
1431split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1432
a422fd2d
SC
1433Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1434loop over the pattern, taking each format character in turn into
1435C<datum_type>. Then for each possible format character, we swallow up
1436the other arguments in the pattern (a field width, an asterisk, and so
1437on) and convert the next chunk input into the specified format, adding
1438it onto the output SV C<cat>.
1439
1440How do we know if the C<U> is the first format in the C<pat>? Well, if
1441we have a pointer to the start of C<pat> then, if we see a C<U> we can
1442test whether we're still at the start of the string. So, here's where
1443C<pat> is set up:
1444
1445 STRLEN fromlen;
1446 register char *pat = SvPVx(*++MARK, fromlen);
1447 register char *patend = pat + fromlen;
1448 register I32 len;
1449 I32 datumtype;
1450 SV *fromstr;
1451
1452We'll have another string pointer in there:
1453
1454 STRLEN fromlen;
1455 register char *pat = SvPVx(*++MARK, fromlen);
1456 register char *patend = pat + fromlen;
1457 + char *patcopy;
1458 register I32 len;
1459 I32 datumtype;
1460 SV *fromstr;
1461
1462And just before we start the loop, we'll set C<patcopy> to be the start
1463of C<pat>:
1464
1465 items = SP - MARK;
1466 MARK++;
1467 sv_setpvn(cat, "", 0);
1468 + patcopy = pat;
1469 while (pat < patend) {
1470
1471Now if we see a C<U> which was at the start of the string, we turn on
1472the UTF8 flag for the output SV, C<cat>:
1473
1474 + if (datumtype == 'U' && pat==patcopy+1)
1475 + SvUTF8_on(cat);
1476 if (datumtype == '#') {
1477 while (pat < patend && *pat != '\n')
1478 pat++;
1479
1480Remember that it has to be C<patcopy+1> because the first character of
1481the string is the C<U> which has been swallowed into C<datumtype!>
1482
1483Oops, we forgot one thing: what if there are spaces at the start of the
1484pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1485character, even though it's not the first thing in the pattern. In this
1486case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1487
1488 if (isSPACE(datumtype))
1489 continue;
1490
1491needs to become
1492
1493 if (isSPACE(datumtype)) {
1494 patcopy++;
1495 continue;
1496 }
1497
1498OK. That's the C part done. Now we must do two additional things before
1499this patch is ready to go: we've changed the behaviour of Perl, and so
1500we must document that change. We must also provide some more regression
1501tests to make sure our patch works and doesn't create a bug somewhere
1502else along the line.
1503
b23b8711
MS
1504The regression tests for each operator live in F<t/op/>, and so we
1505make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1506tests to the end. First, we'll test that the C<U> does indeed create
1507Unicode strings.
1508
1509t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1510use the one from t/test.pl.
b23b8711 1511
35c336e6
MS
1512 require './test.pl';
1513 plan( tests => 159 );
b23b8711
MS
1514
1515so instead of this:
a422fd2d
SC
1516
1517 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1518 print "ok $test\n"; $test++;
1519
35c336e6
MS
1520we can write the more sensible (see L<Test::More> for a full
1521explanation of is() and other testing functions).
b23b8711 1522
35c336e6 1523 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
812f5127 1524 "U* produces unicode" );
b23b8711 1525
a422fd2d
SC
1526Now we'll test that we got that space-at-the-beginning business right:
1527
35c336e6 1528 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1529 " with spaces at the beginning" );
a422fd2d
SC
1530
1531And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1532the first active format:
1533
35c336e6 1534 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
812f5127 1535 "U* not first isn't unicode" );
a422fd2d 1536
35c336e6
MS
1537Mustn't forget to change the number of tests which appears at the top,
1538or else the automated tester will get confused. This will either look
1539like this:
a422fd2d 1540
35c336e6
MS
1541 print "1..156\n";
1542
1543or this:
1544
1545 plan( tests => 156 );
a422fd2d
SC
1546
1547We now compile up Perl, and run it through the test suite. Our new
1548tests pass, hooray!
1549
1550Finally, the documentation. The job is never done until the paperwork is
1551over, so let's describe the change we've just made. The relevant place
1552is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1553this text in the description of C<pack>:
1554
1555 =item *
1556
1557 If the pattern begins with a C<U>, the resulting string will be treated
1558 as Unicode-encoded. You can force UTF8 encoding on in a string with an
1559 initial C<U0>, and the bytes that follow will be interpreted as Unicode
1560 characters. If you don't want this to happen, you can begin your pattern
1561 with C<C0> (or anything else) to force Perl not to UTF8 encode your
1562 string, and then follow this with a C<U*> somewhere in your pattern.
1563
1564All done. Now let's create the patch. F<Porting/patching.pod> tells us
1565that if we're making major changes, we should copy the entire directory
1566to somewhere safe before we begin fiddling, and then do
13a2d996 1567
a422fd2d
SC
1568 diff -ruN old new > patch
1569
1570However, we know which files we've changed, and we can simply do this:
1571
1572 diff -u pp.c~ pp.c > patch
1573 diff -u t/op/pack.t~ t/op/pack.t >> patch
1574 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1575
1576We end up with a patch looking a little like this:
1577
1578 --- pp.c~ Fri Jun 02 04:34:10 2000
1579 +++ pp.c Fri Jun 16 11:37:25 2000
1580 @@ -4375,6 +4375,7 @@
1581 register I32 items;
1582 STRLEN fromlen;
1583 register char *pat = SvPVx(*++MARK, fromlen);
1584 + char *patcopy;
1585 register char *patend = pat + fromlen;
1586 register I32 len;
1587 I32 datumtype;
1588 @@ -4405,6 +4406,7 @@
1589 ...
1590
1591And finally, we submit it, with our rationale, to perl5-porters. Job
1592done!
1593
f7e1e956
MS
1594=head2 Patching a core module
1595
1596This works just like patching anything else, with an extra
1597consideration. Many core modules also live on CPAN. If this is so,
1598patch the CPAN version instead of the core and send the patch off to
1599the module maintainer (with a copy to p5p). This will help the module
1600maintainer keep the CPAN version in sync with the core version without
1601constantly scanning p5p.
1602
acbe17fc
JP
1603=head2 Adding a new function to the core
1604
1605If, as part of a patch to fix a bug, or just because you have an
1606especially good idea, you decide to add a new function to the core,
1607discuss your ideas on p5p well before you start work. It may be that
1608someone else has already attempted to do what you are considering and
1609can give lots of good advice or even provide you with bits of code
1610that they already started (but never finished).
1611
1612You have to follow all of the advice given above for patching. It is
1613extremely important to test any addition thoroughly and add new tests
1614to explore all boundary conditions that your new function is expected
1615to handle. If your new function is used only by one module (e.g. toke),
1616then it should probably be named S_your_function (for static); on the
210b36aa 1617other hand, if you expect it to accessible from other functions in
acbe17fc
JP
1618Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1619for more details.
1620
1621The location of any new code is also an important consideration. Don't
1622just create a new top level .c file and put your code there; you would
1623have to make changes to Configure (so the Makefile is created properly),
1624as well as possibly lots of include files. This is strictly pumpking
1625business.
1626
1627It is better to add your function to one of the existing top level
1628source code files, but your choice is complicated by the nature of
1629the Perl distribution. Only the files that are marked as compiled
1630static are located in the perl executable. Everything else is located
1631in the shared library (or DLL if you are running under WIN32). So,
1632for example, if a function was only used by functions located in
1633toke.c, then your code can go in toke.c. If, however, you want to call
1634the function from universal.c, then you should put your code in another
1635location, for example util.c.
1636
1637In addition to writing your c-code, you will need to create an
1638appropriate entry in embed.pl describing your function, then run
1639'make regen_headers' to create the entries in the numerous header
1640files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1641for information on the various options that you can set in embed.pl.
1642You will forget to do this a few (or many) times and you will get
1643warnings during the compilation phase. Make sure that you mention
1644this when you post your patch to P5P; the pumpking needs to know this.
1645
1646When you write your new code, please be conscious of existing code
884bad00 1647conventions used in the perl source files. See L<perlstyle> for
acbe17fc
JP
1648details. Although most of the guidelines discussed seem to focus on
1649Perl code, rather than c, they all apply (except when they don't ;).
1650See also I<Porting/patching.pod> file in the Perl source distribution
1651for lots of details about both formatting and submitting patches of
1652your changes.
1653
1654Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1655Test on as many platforms as you can find. Test as many perl
1656Configure options as you can (e.g. MULTIPLICITY). If you have
1657profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 1658below for how to use them to further test your code. Remember that
acbe17fc
JP
1659most of the people on P5P are doing this on their own time and
1660don't have the time to debug your code.
f7e1e956
MS
1661
1662=head2 Writing a test
1663
1664Every module and built-in function has an associated test file (or
1665should...). If you add or change functionality, you have to write a
1666test. If you fix a bug, you have to write a test so that bug never
1667comes back. If you alter the docs, it would be nice to test what the
1668new documentation says.
1669
1670In short, if you submit a patch you probably also have to patch the
1671tests.
1672
1673For modules, the test file is right next to the module itself.
1674F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1675so there are some snags (and it would be wonderful for you to brush
1676them out), but it basically works that way. Everything else lives in
1677F<t/>.
1678
1679=over 3
1680
1681=item F<t/base/>
1682
1683Testing of the absolute basic functionality of Perl. Things like
1684C<if>, basic file reads and writes, simple regexes, etc. These are
1685run first in the test suite and if any of them fail, something is
1686I<really> broken.
1687
1688=item F<t/cmd/>
1689
1690These test the basic control structures, C<if/else>, C<while>,
35c336e6 1691subroutines, etc.
f7e1e956
MS
1692
1693=item F<t/comp/>
1694
1695Tests basic issues of how Perl parses and compiles itself.
1696
1697=item F<t/io/>
1698
1699Tests for built-in IO functions, including command line arguments.
1700
1701=item F<t/lib/>
1702
1703The old home for the module tests, you shouldn't put anything new in
1704here. There are still some bits and pieces hanging around in here
1705that need to be moved. Perhaps you could move them? Thanks!
1706
1707=item F<t/op/>
1708
1709Tests for perl's built in functions that don't fit into any of the
1710other directories.
1711
1712=item F<t/pod/>
1713
1714Tests for POD directives. There are still some tests for the Pod
1715modules hanging around in here that need to be moved out into F<lib/>.
1716
1717=item F<t/run/>
1718
1719Testing features of how perl actually runs, including exit codes and
1720handling of PERL* environment variables.
1721
1722=back
1723
1724The core uses the same testing style as the rest of Perl, a simple
1725"ok/not ok" run through Test::Harness, but there are a few special
1726considerations.
1727
35c336e6
MS
1728There are three ways to write a test in the core. Test::More,
1729t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1730decision of which to use depends on what part of the test suite you're
1731working on. This is a measure to prevent a high-level failure (such
1732as Config.pm breaking) from causing basic functionality tests to fail.
1733
1734=over 4
1735
1736=item t/base t/comp
1737
1738Since we don't know if require works, or even subroutines, use ad hoc
1739tests for these two. Step carefully to avoid using the feature being
1740tested.
1741
1742=item t/cmd t/run t/io t/op
1743
1744Now that basic require() and subroutines are tested, you can use the
1745t/test.pl library which emulates the important features of Test::More
1746while using a minimum of core features.
1747
1748You can also conditionally use certain libraries like Config, but be
1749sure to skip the test gracefully if it's not there.
1750
1751=item t/lib ext lib
1752
1753Now that the core of Perl is tested, Test::More can be used. You can
1754also use the full suite of core modules in the tests.
1755
1756=back
f7e1e956
MS
1757
1758When you say "make test" Perl uses the F<t/TEST> program to run the
1759test suite. All tests are run from the F<t/> directory, B<not> the
1760directory which contains the test. This causes some problems with the
1761tests in F<lib/>, so here's some opportunity for some patching.
1762
1763You must be triply conscious of cross-platform concerns. This usually
1764boils down to using File::Spec and avoiding things like C<fork()> and
1765C<system()> unless absolutely necessary.
1766
e018f8be
JH
1767=head2 Special Make Test Targets
1768
1769There are various special make targets that can be used to test Perl
1770slightly differently than the standard "test" target. Not all them
1771are expected to give a 100% success rate. Many of them have several
1772aliases.
1773
1774=over 4
1775
1776=item coretest
1777
7d7d5695 1778Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
e018f8be
JH
1779
1780=item test.deparse
1781
1782Run all the tests through the B::Deparse. Not all tests will succeed.
1783
1784=item minitest
1785
1786Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
1787F<t/op>, and F<t/uni> tests.
1788
1789=item test.third check.third utest.third ucheck.third
1790
1791(Only in Tru64) Run all the tests using the memory leak + naughty
1792memory access tool "Third Degree". The log files will be named
1793F<perl3.log.testname>.
1794
1795=item test.torture torturetest
1796
1797Run all the usual tests and some extra tests. As of Perl 5.8.0 the
1798only extra tests are Abigail's JAPHs, t/japh/abigail.t.
1799
1800You can also run the torture test with F<t/harness> by giving
1801C<-torture> argument to F<t/harness>.
1802
1803=item utest ucheck test.utf8 check.utf8
1804
1805Run all the tests with -Mutf8. Not all tests will succeed.
1806
1807=back
f7e1e956 1808
902b9dbf
MF
1809=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
1810
1811Sometimes it helps to use external tools while debugging and
1812testing Perl. This section tries to guide you through using
1813some common testing and debugging tools with Perl. This is
1814meant as a guide to interfacing these tools with Perl, not
1815as any kind of guide to the use of the tools themselves.
1816
1817=head2 Rational Software's Purify
1818
1819Purify is a commercial tool that is helpful in identifying
1820memory overruns, wild pointers, memory leaks and other such
1821badness. Perl must be compiled in a specific way for
1822optimal testing with Purify. Purify is available under
1823Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
1824
1825The only currently known leaks happen when there are
1826compile-time errors within eval or require. (Fixing these
1827is non-trivial, unfortunately, but they must be fixed
1828eventually.)
1829
1830=head2 Purify on Unix
1831
1832On Unix, Purify creates a new Perl binary. To get the most
1833benefit out of Purify, you should create the perl to Purify
1834using:
1835
1836 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
1837 -Uusemymalloc -Dusemultiplicity
1838
1839where these arguments mean:
1840
1841=over 4
1842
1843=item -Accflags=-DPURIFY
1844
1845Disables Perl's arena memory allocation functions, as well as
1846forcing use of memory allocation functions derived from the
1847system malloc.
1848
1849=item -Doptimize='-g'
1850
1851Adds debugging information so that you see the exact source
1852statements where the problem occurs. Without this flag, all
1853you will see is the source filename of where the error occurred.
1854
1855=item -Uusemymalloc
1856
1857Disable Perl's malloc so that Purify can more closely monitor
1858allocations and leaks. Using Perl's malloc will make Purify
1859report most leaks in the "potential" leaks category.
1860
1861=item -Dusemultiplicity
1862
1863Enabling the multiplicity option allows perl to clean up
1864thoroughly when the interpreter shuts down, which reduces the
1865number of bogus leak reports from Purify.
1866
1867=back
1868
1869Once you've compiled a perl suitable for Purify'ing, then you
1870can just:
1871
1872 make pureperl
1873
1874which creates a binary named 'pureperl' that has been Purify'ed.
1875This binary is used in place of the standard 'perl' binary
1876when you want to debug Perl memory problems.
1877
1f56d61a
JH
1878To minimize the number of memory leak false alarms
1879(see L</PERL_DESTRUCT_LEVEL>), set environment variable
1880PERL_DESTRUCT_LEVEL to 2.
1881
1882 setenv PERL_DESTRUCT_LEVEL 2
1883
1884In Bourne-type shells:
1885
1886 PERL_DESTRUCT_LEVEL=2
1887 export PERL_DESTRUCT_LEVEL
1888
902b9dbf
MF
1889As an example, to show any memory leaks produced during the
1890standard Perl testset you would create and run the Purify'ed
1891perl as:
1892
1893 make pureperl
1894 cd t
1895 ../pureperl -I../lib harness
1896
1897which would run Perl on test.pl and report any memory problems.
1898
1899Purify outputs messages in "Viewer" windows by default. If
1900you don't have a windowing environment or if you simply
1901want the Purify output to unobtrusively go to a log file
1902instead of to the interactive window, use these following
1903options to output to the log file "perl.log":
1904
1905 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
1906 -log-file=perl.log -append-logfile=yes"
1907
1908If you plan to use the "Viewer" windows, then you only need this option:
1909
1910 setenv PURIFYOPTIONS "-chain-length=25"
1911
c406981e
JH
1912In Bourne-type shells:
1913
98631ff8
JL
1914 PURIFYOPTIONS="..."
1915 export PURIFYOPTIONS
c406981e
JH
1916
1917or if you have the "env" utility:
1918
98631ff8 1919 env PURIFYOPTIONS="..." ../pureperl ...
c406981e 1920
902b9dbf
MF
1921=head2 Purify on NT
1922
1923Purify on Windows NT instruments the Perl binary 'perl.exe'
1924on the fly. There are several options in the makefile you
1925should change to get the most use out of Purify:
1926
1927=over 4
1928
1929=item DEFINES
1930
1931You should add -DPURIFY to the DEFINES line so the DEFINES
1932line looks something like:
1933
1934 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
1935
1936to disable Perl's arena memory allocation functions, as
1937well as to force use of memory allocation functions derived
1938from the system malloc.
1939
1940=item USE_MULTI = define
1941
1942Enabling the multiplicity option allows perl to clean up
1943thoroughly when the interpreter shuts down, which reduces the
1944number of bogus leak reports from Purify.
1945
1946=item #PERL_MALLOC = define
1947
1948Disable Perl's malloc so that Purify can more closely monitor
1949allocations and leaks. Using Perl's malloc will make Purify
1950report most leaks in the "potential" leaks category.
1951
1952=item CFG = Debug
1953
1954Adds debugging information so that you see the exact source
1955statements where the problem occurs. Without this flag, all
1956you will see is the source filename of where the error occurred.
1957
1958=back
1959
1960As an example, to show any memory leaks produced during the
1961standard Perl testset you would create and run Purify as:
1962
1963 cd win32
1964 make
1965 cd ../t
1966 purify ../perl -I../lib harness
1967
1968which would instrument Perl in memory, run Perl on test.pl,
1969then finally report any memory problems.
1970
f134cc4e
JH
1971B<NOTE>: as of Perl 5.8.0, the ext/Encode/t/Unicode.t takes
1972extraordinarily long (hours?) to complete under Purify. It has been
1973theorized that it would eventually finish, but nobody has so far been
1974patient enough :-) (This same extreme slowdown has been seen also with
1975the Third Degree tool, so the said test must be doing something that
1976is quite unfriendly for memory debuggers.) It is suggested that you
1977simply kill away that testing process.
1978
1979=head2 Compaq's/Digital's/HP's Third Degree
09187cb1
JH
1980
1981Third Degree is a tool for memory leak detection and memory access checks.
1982It is one of the many tools in the ATOM toolkit. The toolkit is only
1983available on Tru64 (formerly known as Digital UNIX formerly known as
1984DEC OSF/1).
1985
1986When building Perl, you must first run Configure with -Doptimize=-g
1987and -Uusemymalloc flags, after that you can use the make targets
51a35ef1
JH
1988"perl.third" and "test.third". (What is required is that Perl must be
1989compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 1990
64cea5fd 1991The short story is that with "atom" you can instrument the Perl
83f0ef60 1992executable to create a new executable called F<perl.third>. When the
4ae3d70a 1993instrumented executable is run, it creates a log of dubious memory
83f0ef60 1994traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a
JH
1995third for more information. The most extensive Third Degree
1996documentation is available in the Compaq "Tru64 UNIX Programmer's
1997Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 1998
9c54ecba 1999The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
64cea5fd
JH
2000subdirectory. There is a problem with these files: Third Degree is so
2001effective that it finds problems also in the system libraries.
9c54ecba
JH
2002Therefore you should used the Porting/thirdclean script to cleanup
2003the F<*.3log> files.
64cea5fd
JH
2004
2005There are also leaks that for given certain definition of a leak,
2006aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2007
2008=head2 PERL_DESTRUCT_LEVEL
2009
2010If you want to run any of the tests yourself manually using the
2011pureperl or perl.third executables, please note that by default
2012perl B<does not> explicitly cleanup all the memory it has allocated
2013(such as global memory arenas) but instead lets the exit() of
2014the whole program "take care" of such allocations, also known
2015as "global destruction of objects".
2016
2017There is a way to tell perl to do complete cleanup: set the
2018environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2019The t/TEST wrapper does set this to 2, and this is what you
2020need to do too, if you don't want to see the "global leaks":
1f56d61a 2021For example, for "third-degreed" Perl:
64cea5fd 2022
1f56d61a 2023 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
09187cb1 2024
414f2397
RGS
2025(Note: the mod_perl apache module uses also this environment variable
2026for its own purposes and extended its semantics. Refer to the mod_perl
287a822c
RGS
2027documentation for more information. Also, spawned threads do the
2028equivalent of setting this variable to the value 1.)
5a6c59ef
DM
2029
2030If, at the end of a run you get the message I<N scalars leaked>, you can
2031recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause
2032the addresses of all those leaked SVs to be dumped; it also converts
2033C<new_SV()> from a macro into a real function, so you can use your
2034favourite debugger to discover where those pesky SVs were allocated.
414f2397 2035
51a35ef1
JH
2036=head2 Profiling
2037
2038Depending on your platform there are various of profiling Perl.
2039
2040There are two commonly used techniques of profiling executables:
10f58044 2041I<statistical time-sampling> and I<basic-block counting>.
51a35ef1
JH
2042
2043The first method takes periodically samples of the CPU program
2044counter, and since the program counter can be correlated with the code
2045generated for functions, we get a statistical view of in which
2046functions the program is spending its time. The caveats are that very
2047small/fast functions have lower probability of showing up in the
2048profile, and that periodically interrupting the program (this is
2049usually done rather frequently, in the scale of milliseconds) imposes
2050an additional overhead that may skew the results. The first problem
2051can be alleviated by running the code for longer (in general this is a
2052good idea for profiling), the second problem is usually kept in guard
2053by the profiling tools themselves.
2054
10f58044 2055The second method divides up the generated code into I<basic blocks>.
51a35ef1
JH
2056Basic blocks are sections of code that are entered only in the
2057beginning and exited only at the end. For example, a conditional jump
2058starts a basic block. Basic block profiling usually works by
10f58044 2059I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1
JH
2060book-keeping code to the generated code. During the execution of the
2061code the basic block counters are then updated appropriately. The
2062caveat is that the added extra code can skew the results: again, the
2063profiling tools usually try to factor their own effects out of the
2064results.
2065
83f0ef60
JH
2066=head2 Gprof Profiling
2067
51a35ef1
JH
2068gprof is a profiling tool available in many UNIX platforms,
2069it uses F<statistical time-sampling>.
83f0ef60
JH
2070
2071You can build a profiled version of perl called "perl.gprof" by
51a35ef1
JH
2072invoking the make target "perl.gprof" (What is required is that Perl
2073must be compiled using the C<-pg> flag, you may need to re-Configure).
2074Running the profiled version of Perl will create an output file called
2075F<gmon.out> is created which contains the profiling data collected
2076during the execution.
83f0ef60
JH
2077
2078The gprof tool can then display the collected data in various ways.
2079Usually gprof understands the following options:
2080
2081=over 4
2082
2083=item -a
2084
2085Suppress statically defined functions from the profile.
2086
2087=item -b
2088
2089Suppress the verbose descriptions in the profile.
2090
2091=item -e routine
2092
2093Exclude the given routine and its descendants from the profile.
2094
2095=item -f routine
2096
2097Display only the given routine and its descendants in the profile.
2098
2099=item -s
2100
2101Generate a summary file called F<gmon.sum> which then may be given
2102to subsequent gprof runs to accumulate data over several runs.
2103
2104=item -z
2105
2106Display routines that have zero usage.
2107
2108=back
2109
2110For more detailed explanation of the available commands and output
2111formats, see your own local documentation of gprof.
2112
51a35ef1
JH
2113=head2 GCC gcov Profiling
2114
10f58044 2115Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1
JH
2116for the GNU CC.
2117
2118You can build a profiled version of perl called F<perl.gcov> by
2119invoking the make target "perl.gcov" (what is required that Perl must
2120be compiled using gcc with the flags C<-fprofile-arcs
2121-ftest-coverage>, you may need to re-Configure).
2122
2123Running the profiled version of Perl will cause profile output to be
2124generated. For each source file an accompanying ".da" file will be
2125created.
2126
2127To display the results you use the "gcov" utility (which should
2128be installed if you have gcc 3.0 or newer installed). F<gcov> is
2129run on source code files, like this
2130
2131 gcov sv.c
2132
2133which will cause F<sv.c.gcov> to be created. The F<.gcov> files
2134contain the source code annotated with relative frequencies of
2135execution indicated by "#" markers.
2136
2137Useful options of F<gcov> include C<-b> which will summarise the
2138basic block, branch, and function call coverage, and C<-c> which
2139instead of relative frequencies will use the actual counts. For
2140more information on the use of F<gcov> and basic block profiling
2141with gcc, see the latest GNU CC manual, as of GCC 3.0 see
2142
2143 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
2144
2145and its section titled "8. gcov: a Test Coverage Program"
2146
2147 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
2148
4ae3d70a
JH
2149=head2 Pixie Profiling
2150
51a35ef1
JH
2151Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
2152UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 2153I<basic-block counting>.
4ae3d70a 2154
83f0ef60 2155You can build a profiled version of perl called F<perl.pixie> by
51a35ef1
JH
2156invoking the make target "perl.pixie" (what is required is that Perl
2157must be compiled using the C<-g> flag, you may need to re-Configure).
2158
2159In Tru64 a file called F<perl.Addrs> will also be silently created,
2160this file contains the addresses of the basic blocks. Running the
2161profiled version of Perl will create a new file called "perl.Counts"
2162which contains the counts for the basic block for that particular
2163program execution.
4ae3d70a 2164
51a35ef1 2165To display the results you use the F<prof> utility. The exact
4ae3d70a
JH
2166incantation depends on your operating system, "prof perl.Counts" in
2167IRIX, and "prof -pixie -all -L. perl" in Tru64.
2168
6c41479b
JH
2169In IRIX the following prof options are available:
2170
2171=over 4
2172
2173=item -h
2174
2175Reports the most heavily used lines in descending order of use.
6e36760b 2176Useful for finding the hotspot lines.
6c41479b
JH
2177
2178=item -l
2179
2180Groups lines by procedure, with procedures sorted in descending order of use.
2181Within a procedure, lines are listed in source order.
6e36760b 2182Useful for finding the hotspots of procedures.
6c41479b
JH
2183
2184=back
2185
2186In Tru64 the following options are available:
2187
2188=over 4
2189
3958b146 2190=item -p[rocedures]
6c41479b 2191
3958b146 2192Procedures sorted in descending order by the number of cycles executed
6e36760b 2193in each procedure. Useful for finding the hotspot procedures.
6c41479b
JH
2194(This is the default option.)
2195
24000d2f 2196=item -h[eavy]
6c41479b 2197
6e36760b
JH
2198Lines sorted in descending order by the number of cycles executed in
2199each line. Useful for finding the hotspot lines.
6c41479b 2200
24000d2f 2201=item -i[nvocations]
6c41479b 2202
6e36760b
JH
2203The called procedures are sorted in descending order by number of calls
2204made to the procedures. Useful for finding the most used procedures.
6c41479b 2205
24000d2f 2206=item -l[ines]
6c41479b
JH
2207
2208Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 2209Useful for finding the hotspots of procedures.
6c41479b
JH
2210
2211=item -testcoverage
2212
2213The compiler emitted code for these lines, but the code was unexecuted.
2214
24000d2f 2215=item -z[ero]
6c41479b
JH
2216
2217Unexecuted procedures.
2218
aa500c9e 2219=back
6c41479b
JH
2220
2221For further information, see your system's manual pages for pixie and prof.
4ae3d70a 2222
b8ddf6b3
SB
2223=head2 Miscellaneous tricks
2224
2225=over 4
2226
2227=item *
2228
cc177e1a 2229Those debugging perl with the DDD frontend over gdb may find the
b8ddf6b3
SB
2230following useful:
2231
2232You can extend the data conversion shortcuts menu, so for example you
2233can display an SV's IV value with one click, without doing any typing.
2234To do that simply edit ~/.ddd/init file and add after:
2235
2236 ! Display shortcuts.
2237 Ddd*gdbDisplayShortcuts: \
2238 /t () // Convert to Bin\n\
2239 /d () // Convert to Dec\n\
2240 /x () // Convert to Hex\n\
2241 /o () // Convert to Oct(\n\
2242
2243the following two lines:
2244
2245 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
2246 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
2247
2248so now you can do ivx and pvx lookups or you can plug there the
2249sv_peek "conversion":
2250
2251 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
2252
2253(The my_perl is for threaded builds.)
2254Just remember that every line, but the last one, should end with \n\
2255
2256Alternatively edit the init file interactively via:
22573rd mouse button -> New Display -> Edit Menu
2258
2259Note: you can define up to 20 conversion shortcuts in the gdb
2260section.
2261
9965345d
JH
2262=item *
2263
2264If you see in a debugger a memory area mysteriously full of 0xabababab,
2265you may be seeing the effect of the Poison() macro, see L<perlclib>.
2266
b8ddf6b3
SB
2267=back
2268
a422fd2d
SC
2269=head2 CONCLUSION
2270
2271We've had a brief look around the Perl source, an overview of the stages
2272F<perl> goes through when it's running your code, and how to use a
902b9dbf
MF
2273debugger to poke at the Perl guts. We took a very simple problem and
2274demonstrated how to solve it fully - with documentation, regression
2275tests, and finally a patch for submission to p5p. Finally, we talked
2276about how to use external tools to debug and test Perl.
a422fd2d
SC
2277
2278I'd now suggest you read over those references again, and then, as soon
2279as possible, get your hands dirty. The best way to learn is by doing,
2280so:
2281
2282=over 3
2283
2284=item *
2285
2286Subscribe to perl5-porters, follow the patches and try and understand
2287them; don't be afraid to ask if there's a portion you're not clear on -
2288who knows, you may unearth a bug in the patch...
2289
2290=item *
2291
2292Keep up to date with the bleeding edge Perl distributions and get
2293familiar with the changes. Try and get an idea of what areas people are
2294working on and the changes they're making.
2295
2296=item *
2297
3e148164 2298Do read the README associated with your operating system, e.g. README.aix
a1f349fd
MB
2299on the IBM AIX OS. Don't hesitate to supply patches to that README if
2300you find anything missing or changed over a new OS release.
2301
2302=item *
2303
a422fd2d
SC
2304Find an area of Perl that seems interesting to you, and see if you can
2305work out how it works. Scan through the source, and step over it in the
2306debugger. Play, poke, investigate, fiddle! You'll probably get to
2307understand not just your chosen area but a much wider range of F<perl>'s
2308activity as well, and probably sooner than you'd think.
2309
2310=back
2311
2312=over 3
2313
2314=item I<The Road goes ever on and on, down from the door where it began.>
2315
2316=back
2317
2318If you can do these things, you've started on the long road to Perl porting.
2319Thanks for wanting to help make Perl better - and happy hacking!
2320
e8cd7eae
GS
2321=head1 AUTHOR
2322
2323This document was written by Nathan Torkington, and is maintained by
2324the perl5-porters mailing list.
2325