3 perlhack - How to hack at the Perl internals
7 This document attempts to explain how Perl development takes place,
8 and ends with some suggestions for people wanting to become bona fide
11 The perl5-porters mailing list is where the Perl standard distribution
12 is maintained and developed. The list can get anywhere from 10 to 150
13 messages a day, depending on the heatedness of the debate. Most days
14 there are two or three patches, extensions, features, or bugs being
17 A searchable archive of the list is at either:
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
23 http://archive.develooper.com/perl5-porters@perl.org/
25 List subscribers (the porters themselves) come in several flavours.
26 Some are quiet curious lurkers, who rarely pitch in and instead watch
27 the ongoing development to ensure they're forewarned of new changes or
28 features in Perl. Some are representatives of vendors, who are there
29 to make sure that Perl continues to compile and work on their
30 platforms. Some patch any reported bug that they know how to fix,
31 some are actively patching their pet area (threads, Win32, the regexp
32 engine), while others seem to do nothing but complain. In other
33 words, it's your usual mix of technical people.
35 Over this group of porters presides Larry Wall. He has the final word
36 in what does and does not change in the Perl language. Various
37 releases of Perl are shepherded by a "pumpking", a porter
38 responsible for gathering patches, deciding on a patch-by-patch,
39 feature-by-feature basis what will and will not go into the release.
40 For instance, Gurusamy Sarathy was the pumpking for the 5.6 release of
41 Perl, and Jarkko Hietaniemi was the pumpking for the 5.8 release, and
42 Rafael Garcia-Suarez holds the pumpking crown for the 5.10 release.
44 In addition, various people are pumpkings for different things. For
45 instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the
46 I<Configure> pumpkin up till the 5.8 release. For the 5.10 release
47 H.Merijn Brand took over.
49 Larry sees Perl development along the lines of the US government:
50 there's the Legislature (the porters), the Executive branch (the
51 pumpkings), and the Supreme Court (Larry). The legislature can
52 discuss and submit patches to the executive branch all they like, but
53 the executive branch is free to veto them. Rarely, the Supreme Court
54 will side with the executive branch over the legislature, or the
55 legislature over the executive branch. Mostly, however, the
56 legislature and the executive branch are supposed to get along and
57 work out their differences without impeachment or court cases.
59 You might sometimes see reference to Rule 1 and Rule 2. Larry's power
60 as Supreme Court is expressed in The Rules:
66 Larry is always by definition right about how Perl should behave.
67 This means he has final veto power on the core functionality.
71 Larry is allowed to change his mind about any matter at a later date,
72 regardless of whether he previously invoked Rule 1.
76 Got that? Larry is always right, even when he was wrong. It's rare
77 to see either Rule exercised, but they are often alluded to.
79 New features and extensions to the language are contentious, because
80 the criteria used by the pumpkings, Larry, and other porters to decide
81 which features should be implemented and incorporated are not codified
82 in a few small design goals as with some other languages. Instead,
83 the heuristics are flexible and often difficult to fathom. Here is
84 one person's list, roughly in decreasing order of importance, of
85 heuristics that new features have to be weighed against:
89 =item Does concept match the general goals of Perl?
91 These haven't been written anywhere in stone, but one approximation
94 1. Keep it fast, simple, and useful.
95 2. Keep features/concepts as orthogonal as possible.
96 3. No arbitrary limits (platforms, data sizes, cultures).
97 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
98 5. Either assimilate new technologies, or build bridges to them.
100 =item Where is the implementation?
102 All the talk in the world is useless without an implementation. In
103 almost every case, the person or people who argue for a new feature
104 will be expected to be the ones who implement it. Porters capable
105 of coding new features have their own agendas, and are not available
106 to implement your (possibly good) idea.
108 =item Backwards compatibility
110 It's a cardinal sin to break existing Perl programs. New warnings are
111 contentious--some say that a program that emits warnings is not
112 broken, while others say it is. Adding keywords has the potential to
113 break programs, changing the meaning of existing token sequences or
114 functions might break programs.
116 =item Could it be a module instead?
118 Perl 5 has extension mechanisms, modules and XS, specifically to avoid
119 the need to keep changing the Perl interpreter. You can write modules
120 that export functions, you can give those functions prototypes so they
121 can be called like built-in functions, you can even write XS code to
122 mess with the runtime data structures of the Perl interpreter if you
123 want to implement really complicated things. If it can be done in a
124 module instead of in the core, it's highly unlikely to be added.
126 =item Is the feature generic enough?
128 Is this something that only the submitter wants added to the language,
129 or would it be broadly useful? Sometimes, instead of adding a feature
130 with a tight focus, the porters might decide to wait until someone
131 implements the more generalized feature. For instance, instead of
132 implementing a "delayed evaluation" feature, the porters are waiting
133 for a macro system that would permit delayed evaluation and much more.
135 =item Does it potentially introduce new bugs?
137 Radical rewrites of large chunks of the Perl interpreter have the
138 potential to introduce new bugs. The smaller and more localized the
141 =item Does it preclude other desirable features?
143 A patch is likely to be rejected if it closes off future avenues of
144 development. For instance, a patch that placed a true and final
145 interpretation on prototypes is likely to be rejected because there
146 are still options for the future of prototypes that haven't been
149 =item Is the implementation robust?
151 Good patches (tight code, complete, correct) stand more chance of
152 going in. Sloppy or incorrect patches might be placed on the back
153 burner until the pumpking has time to fix, or might be discarded
154 altogether without further notice.
156 =item Is the implementation generic enough to be portable?
158 The worst patches make use of a system-specific features. It's highly
159 unlikely that nonportable additions to the Perl language will be
162 =item Is the implementation tested?
164 Patches which change behaviour (fixing bugs or introducing new features)
165 must include regression tests to verify that everything works as expected.
166 Without tests provided by the original author, how can anyone else changing
167 perl in the future be sure that they haven't unwittingly broken the behaviour
168 the patch implements? And without tests, how can the patch's author be
169 confident that his/her hard work put into the patch won't be accidentally
170 thrown away by someone in the future?
172 =item Is there enough documentation?
174 Patches without documentation are probably ill-thought out or
175 incomplete. Nothing can be added without documentation, so submitting
176 a patch for the appropriate manpages as well as the source code is
179 =item Is there another way to do it?
181 Larry said "Although the Perl Slogan is I<There's More Than One Way
182 to Do It>, I hesitate to make 10 ways to do something". This is a
183 tricky heuristic to navigate, though--one man's essential addition is
184 another man's pointless cruft.
186 =item Does it create too much work?
188 Work for the pumpking, work for Perl programmers, work for module
189 authors, ... Perl is supposed to be easy.
191 =item Patches speak louder than words
193 Working code is always preferred to pie-in-the-sky ideas. A patch to
194 add a feature stands a much higher chance of making it to the language
195 than does a random feature request, no matter how fervently argued the
196 request might be. This ties into "Will it be useful?", as the fact
197 that someone took the time to make the patch demonstrates a strong
198 desire for the feature.
202 If you're on the list, you might hear the word "core" bandied
203 around. It refers to the standard distribution. "Hacking on the
204 core" means you're changing the C source code to the Perl
205 interpreter. "A core module" is one that ships with Perl.
207 =head2 Keeping in sync
209 The source code to the Perl interpreter, in its different versions, is
210 kept in a repository managed by a revision control system ( which is
211 currently the Perforce program, see http://perforce.com/ ). The
212 pumpkings and a few others have access to the repository to check in
213 changes. Periodically the pumpking for the development version of Perl
214 will release a new version, so the rest of the porters can see what's
215 changed. The current state of the main trunk of repository, and patches
216 that describe the individual changes that have happened since the last
217 public release are available at this location:
219 http://public.activestate.com/pub/apc/
220 ftp://public.activestate.com/pub/apc/
222 If you're looking for a particular change, or a change that affected
223 a particular set of files, you may find the B<Perl Repository Browser>
226 http://public.activestate.com/cgi-bin/perlbrowse
228 You may also want to subscribe to the perl5-changes mailing list to
229 receive a copy of each patch that gets submitted to the maintenance
230 and development "branches" of the perl repository. See
231 http://lists.perl.org/ for subscription information.
233 If you are a member of the perl5-porters mailing list, it is a good
234 thing to keep in touch with the most recent changes. If not only to
235 verify if what you would have posted as a bug report isn't already
236 solved in the most recent available perl development branch, also
237 known as perl-current, bleading edge perl, bleedperl or bleadperl.
239 Needless to say, the source code in perl-current is usually in a perpetual
240 state of evolution. You should expect it to be very buggy. Do B<not> use
241 it for any purpose other than testing and development.
243 Keeping in sync with the most recent branch can be done in several ways,
244 but the most convenient and reliable way is using B<rsync>, available at
245 ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
248 If you choose to keep in sync using rsync, there are two approaches
253 =item rsync'ing the source tree
255 Presuming you are in the directory where your perl source resides
256 and you have rsync installed and available, you can "upgrade" to
259 # rsync -avz rsync://public.activestate.com/perl-current/ .
261 This takes care of updating every single item in the source tree to
262 the latest applied patch level, creating files that are new (to your
263 distribution) and setting date/time stamps of existing files to
264 reflect the bleadperl status.
266 Note that this will not delete any files that were in '.' before
267 the rsync. Once you are sure that the rsync is running correctly,
268 run it with the --delete and the --dry-run options like this:
270 # rsync -avz --delete --dry-run rsync://public.activestate.com/perl-current/ .
272 This will I<simulate> an rsync run that also deletes files not
273 present in the bleadperl master copy. Observe the results from
274 this run closely. If you are sure that the actual run would delete
275 no files precious to you, you could remove the '--dry-run' option.
277 You can than check what patch was the latest that was applied by
278 looking in the file B<.patch>, which will show the number of the
281 If you have more than one machine to keep in sync, and not all of
282 them have access to the WAN (so you are not able to rsync all the
283 source trees to the real source), there are some ways to get around
288 =item Using rsync over the LAN
290 Set up a local rsync server which makes the rsynced source tree
291 available to the LAN and sync the other machines against this
294 From http://rsync.samba.org/README.html :
296 "Rsync uses rsh or ssh for communication. It does not need to be
297 setuid and requires no special privileges for installation. It
298 does not require an inetd entry or a daemon. You must, however,
299 have a working rsh or ssh system. Using ssh is recommended for
300 its security features."
302 =item Using pushing over the NFS
304 Having the other systems mounted over the NFS, you can take an
305 active pushing approach by checking the just updated tree against
306 the other not-yet synced trees. An example would be
315 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
318 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
320 foreach my $host (keys %remote) {
321 unless (-d $remote{$host}) {
322 print STDERR "Cannot Xsync for host $host\n";
325 foreach my $file (keys %MF) {
326 my $rfile = "$remote{$host}/$file";
327 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
328 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
329 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
330 printf "%4s %-34s %8d %9d %8d %9d\n",
331 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
333 copy ($file, $rfile);
334 utime time, $MF{$file}[2], $rfile;
335 chmod $MF{$file}[0], $rfile;
339 though this is not perfect. It could be improved with checking
340 file checksums before updating. Not all NFS systems support
341 reliable utime support (when used over the NFS).
345 =item rsync'ing the patches
347 The source tree is maintained by the pumpking who applies patches to
348 the files in the tree. These patches are either created by the
349 pumpking himself using C<diff -c> after updating the file manually or
350 by applying patches sent in by posters on the perl5-porters list.
351 These patches are also saved and rsync'able, so you can apply them
352 yourself to the source files.
354 Presuming you are in a directory where your patches reside, you can
355 get them in sync with
357 # rsync -avz rsync://public.activestate.com/perl-current-diffs/ .
359 This makes sure the latest available patch is downloaded to your
362 It's then up to you to apply these patches, using something like
364 # last="`cat ../perl-current/.patch`.gz"
365 # rsync -avz rsync://public.activestate.com/perl-current-diffs/ .
366 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
368 # patch -p1 -N <../perl-current-diffs/blead.patch
370 or, since this is only a hint towards how it works, use CPAN-patchaperl
371 from Andreas König to have better control over the patching process.
375 =head2 Why rsync the source tree
379 =item It's easier to rsync the source tree
381 Since you don't have to apply the patches yourself, you are sure all
382 files in the source tree are in the right state.
384 =item It's more reliable
386 While both the rsync-able source and patch areas are automatically
387 updated every few minutes, keep in mind that applying patches may
388 sometimes mean careful hand-holding, especially if your version of
389 the C<patch> program does not understand how to deal with new files,
390 files with 8-bit characters, or files without trailing newlines.
394 =head2 Why rsync the patches
398 =item It's easier to rsync the patches
400 If you have more than one machine that you want to keep in track with
401 bleadperl, it's easier to rsync the patches only once and then apply
402 them to all the source trees on the different machines.
404 In case you try to keep in pace on 5 different machines, for which
405 only one of them has access to the WAN, rsync'ing all the source
406 trees should than be done 5 times over the NFS. Having
407 rsync'ed the patches only once, I can apply them to all the source
408 trees automatically. Need you say more ;-)
410 =item It's a good reference
412 If you do not only like to have the most recent development branch,
413 but also like to B<fix> bugs, or extend features, you want to dive
414 into the sources. If you are a seasoned perl core diver, you don't
415 need no manuals, tips, roadmaps, perlguts.pod or other aids to find
416 your way around. But if you are a starter, the patches may help you
417 in finding where you should start and how to change the bits that
420 The file B<Changes> is updated on occasions the pumpking sees as his
421 own little sync points. On those occasions, he releases a tar-ball of
422 the current source tree (i.e. perl@7582.tar.gz), which will be an
423 excellent point to start with when choosing to use the 'rsync the
424 patches' scheme. Starting with perl@7582, which means a set of source
425 files on which the latest applied patch is number 7582, you apply all
426 succeeding patches available from then on (7583, 7584, ...).
428 You can use the patches later as a kind of search archive.
432 =item Finding a start point
434 If you want to fix/change the behaviour of function/feature Foo, just
435 scan the patches for patches that mention Foo either in the subject,
436 the comments, or the body of the fix. A good chance the patch shows
437 you the files that are affected by that patch which are very likely
438 to be the starting point of your journey into the guts of perl.
440 =item Finding how to fix a bug
442 If you've found I<where> the function/feature Foo misbehaves, but you
443 don't know how to fix it (but you do know the change you want to
444 make), you can, again, peruse the patches for similar changes and
445 look how others apply the fix.
447 =item Finding the source of misbehaviour
449 When you keep in sync with bleadperl, the pumpking would love to
450 I<see> that the community efforts really work. So after each of his
451 sync points, you are to 'make test' to check if everything is still
452 in working order. If it is, you do 'make ok', which will send an OK
453 report to perlbug@perl.org. (If you do not have access to a mailer
454 from the system you just finished successfully 'make test', you can
455 do 'make okfile', which creates the file C<perl.ok>, which you can
456 than take to your favourite mailer and mail yourself).
458 But of course, as always, things will not always lead to a success
459 path, and one or more test do not pass the 'make test'. Before
460 sending in a bug report (using 'make nok' or 'make nokfile'), check
461 the mailing list if someone else has reported the bug already and if
462 so, confirm it by replying to that message. If not, you might want to
463 trace the source of that misbehaviour B<before> sending in the bug,
464 which will help all the other porters in finding the solution.
466 Here the saved patches come in very handy. You can check the list of
467 patches to see which patch changed what file and what change caused
468 the misbehaviour. If you note that in the bug report, it saves the
469 one trying to solve it, looking for that point.
473 If searching the patches is too bothersome, you might consider using
474 perl's bugtron to find more information about discussions and
475 ramblings on posted bugs.
477 If you want to get the best of both worlds, rsync both the source
478 tree for convenience, reliability and ease and rsync the patches
483 =head2 Working with the source
485 Because you cannot use the Perforce client, you cannot easily generate
486 diffs against the repository, nor will merges occur when you update
487 via rsync. If you edit a file locally and then rsync against the
488 latest source, changes made in the remote copy will I<overwrite> your
491 The best way to deal with this is to maintain a tree of symlinks to
492 the rsync'd source. Then, when you want to edit a file, you remove
493 the symlink, copy the real file into the other tree, and edit it. You
494 can then diff your edited file against the original to generate a
495 patch, and you can safely update the original tree.
497 Perl's F<Configure> script can generate this tree of symlinks for you.
498 The following example assumes that you have used rsync to pull a copy
499 of the Perl source into the F<perl-rsync> directory. In the directory
500 above that one, you can execute the following commands:
504 ../perl-rsync/Configure -Dmksymlinks -Dusedevel -D"optimize=-g"
506 This will start the Perl configuration process. After a few prompts,
507 you should see something like this:
509 Symbolic links are supported.
511 Checking how to test for symbolic links...
512 Your builtin 'test -h' may be broken.
513 Trying external '/usr/bin/test -h'.
514 You can test for symbolic links with '/usr/bin/test -h'.
516 Creating the symbolic links...
517 (First creating the subdirectories...)
518 (Then creating the symlinks...)
520 The specifics may vary based on your operating system, of course.
521 After you see this, you can abort the F<Configure> script, and you
522 will see that the directory you are in has a tree of symlinks to the
523 F<perl-rsync> directories and files.
525 If you plan to do a lot of work with the Perl source, here are some
526 Bourne shell script functions that can make your life easier:
539 if [ -L $1.orig ]; then
545 Replace "vi" with your favorite flavor of editor.
547 Here is another function which will quickly generate a patch for the
548 files which have been edited in your symlink tree:
552 for f in `find . -name '*.orig' | sed s,^\./,,`
554 case `echo $f | sed 's,.orig$,,;s,.*\.,,'` in
556 pod) diffopts='-F^=' ;;
559 diff -du $diffopts $f `echo $f | sed 's,.orig$,,'`
563 This function produces patches which include enough context to make
564 your changes obvious. This makes it easier for the Perl pumpking(s)
565 to review them when you send them to the perl5-porters list, and that
566 means they're more likely to get applied.
568 This function assumed a GNU diff, and may require some tweaking for
571 =head2 Perlbug administration
573 There is a single remote administrative interface for modifying bug status,
574 category, open issues etc. using the B<RT> I<bugtracker> system, maintained
575 by I<Robert Spier>. Become an administrator, and close any bugs you can get
576 your sticky mitts on:
580 The bugtracker mechanism for B<perl5> bugs in particular is at:
582 http://bugs6.perl.org/perlbug
584 To email the bug system administrators:
586 "perlbug-admin" <perlbug-admin@perl.org>
589 =head2 Submitting patches
591 Always submit patches to I<perl5-porters@perl.org>. If you're
592 patching a core module and there's an author listed, send the author a
593 copy (see L<Patching a core module>). This lets other porters review
594 your patch, which catches a surprising number of errors in patches.
595 Either use the diff program (available in source code form from
596 ftp://ftp.gnu.org/pub/gnu/ , or use Johan Vromans' I<makepatch>
597 (available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
598 but context diffs are accepted. Do not send RCS-style diffs or diffs
599 without context lines. More information is given in the
600 I<Porting/patching.pod> file in the Perl source distribution. Please
601 patch against the latest B<development> version. (e.g., even if you're
602 fixing a bug in the 5.8 track, patch against the latest B<development>
603 version rsynced from rsync://public.activestate.com/perl-current/ )
605 If changes are accepted, they are applied to the development branch. Then
606 the 5.8 pumpking decides which of those patches is to be backported to the
607 maint branch. Only patches that survive the heat of the development
608 branch get applied to maintenance versions.
610 Your patch should also update the documentation and test suite. See
613 Patching documentation also follows the same order: if accepted, a patch
614 is first applied to B<development>, and if relevant then it's backported
615 to B<maintenance>. (With an exception for some patches that document
616 behaviour that only appears in the maintenance branch, but which has
617 changed in the development version.)
619 To report a bug in Perl, use the program I<perlbug> which comes with
620 Perl (if you can't get Perl to work, send mail to the address
621 I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
622 I<perlbug> feeds into the automated bug-tracking system, access to
623 which is provided through the web at http://bugs.perl.org/ . It
624 often pays to check the archives of the perl5-porters mailing list to
625 see whether the bug you're reporting has been reported before, and if
626 so whether it was considered a bug. See above for the location of
627 the searchable archives.
629 The CPAN testers ( http://testers.cpan.org/ ) are a group of
630 volunteers who test CPAN modules on a variety of platforms. Perl
631 Smokers ( http://archives.develooper.com/daily-build@perl.org/ )
632 automatically tests Perl source releases on platforms with various
633 configurations. Both efforts welcome volunteers.
635 It's a good idea to read and lurk for a while before chipping in.
636 That way you'll get to see the dynamic of the conversations, learn the
637 personalities of the players, and hopefully be better prepared to make
638 a useful contribution when do you speak up.
640 If after all this you still think you want to join the perl5-porters
641 mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
642 unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
644 To hack on the Perl guts, you'll need to read the following things:
650 This is of paramount importance, since it's the documentation of what
651 goes where in the Perl source. Read it over a couple of times and it
652 might start to make sense - don't worry if it doesn't yet, because the
653 best way to study it is to read it in conjunction with poking at Perl
654 source, and we'll do that later on.
656 You might also want to look at Gisle Aas's illustrated perlguts -
657 there's no guarantee that this will be absolutely up-to-date with the
658 latest documentation in the Perl core, but the fundamentals will be
659 right. ( http://gisle.aas.no/perl/illguts/ )
661 =item L<perlxstut> and L<perlxs>
663 A working knowledge of XSUB programming is incredibly useful for core
664 hacking; XSUBs use techniques drawn from the PP code, the portion of the
665 guts that actually executes a Perl program. It's a lot gentler to learn
666 those techniques from simple examples and explanation than from the core
671 The documentation for the Perl API explains what some of the internal
672 functions do, as well as the many macros used in the source.
674 =item F<Porting/pumpkin.pod>
676 This is a collection of words of wisdom for a Perl porter; some of it is
677 only useful to the pumpkin holder, but most of it applies to anyone
678 wanting to go about Perl development.
680 =item The perl5-porters FAQ
682 This should be available from http://simon-cozens.org/writings/p5p-faq ;
683 alternatively, you can get the FAQ emailed to you by sending mail to
684 C<perl5-porters-faq@perl.org>. It contains hints on reading perl5-porters,
685 information on how perl5-porters works and how Perl development in general
690 =head2 Finding Your Way Around
692 Perl maintenance can be split into a number of areas, and certain people
693 (pumpkins) will have responsibility for each area. These areas sometimes
694 correspond to files or directories in the source kit. Among the areas are:
700 Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
701 subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
702 contains the core XS modules.
706 There are tests for nearly all the modules, built-ins and major bits
707 of functionality. Test files all have a .t suffix. Module tests live
708 in the F<lib/> and F<ext/> directories next to the module being
709 tested. Others live in F<t/>. See L<Writing a test>
713 Documentation maintenance includes looking after everything in the
714 F<pod/> directory, (as well as contributing new documentation) and
715 the documentation to the modules in core.
719 The configure process is the way we make Perl portable across the
720 myriad of operating systems it supports. Responsibility for the
721 configure, build and installation process, as well as the overall
722 portability of the core code rests with the configure pumpkin - others
723 help out with individual operating systems.
725 The files involved are the operating system directories, (F<win32/>,
726 F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
727 and F<Makefile>, as well as the metaconfig files which generate
728 F<Configure>. (metaconfig isn't included in the core distribution.)
732 And of course, there's the core of the Perl interpreter itself. Let's
733 have a look at that in a little more detail.
737 Before we leave looking at the layout, though, don't forget that
738 F<MANIFEST> contains not only the file names in the Perl distribution,
739 but short descriptions of what's in them, too. For an overview of the
740 important files, try this:
742 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
744 =head2 Elements of the interpreter
746 The work of the interpreter has two main stages: compiling the code
747 into the internal representation, or bytecode, and then executing it.
748 L<perlguts/Compiled code> explains exactly how the compilation stage
751 Here is a short breakdown of perl's operation:
757 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
758 This is very high-level code, enough to fit on a single screen, and it
759 resembles the code found in L<perlembed>; most of the real action takes
762 First, F<perlmain.c> allocates some memory and constructs a Perl
765 1 PERL_SYS_INIT3(&argc,&argv,&env);
767 3 if (!PL_do_undump) {
768 4 my_perl = perl_alloc();
771 7 perl_construct(my_perl);
772 8 PL_perl_destruct_level = 0;
775 Line 1 is a macro, and its definition is dependent on your operating
776 system. Line 3 references C<PL_do_undump>, a global variable - all
777 global variables in Perl start with C<PL_>. This tells you whether the
778 current running program was created with the C<-u> flag to perl and then
779 F<undump>, which means it's going to be false in any sane context.
781 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
782 interpreter. It's quite a simple function, and the guts of it looks like
785 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
787 Here you see an example of Perl's system abstraction, which we'll see
788 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
789 own C<malloc> as defined in F<malloc.c> if you selected that option at
792 Next, in line 7, we construct the interpreter; this sets up all the
793 special variables that Perl needs, the stacks, and so on.
795 Now we pass Perl the command line options, and tell it to go:
797 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
799 exitstatus = perl_run(my_perl);
803 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
804 in F<perl.c>, which processes the command line options, sets up any
805 statically linked XS modules, opens the program and calls C<yyparse> to
810 The aim of this stage is to take the Perl source, and turn it into an op
811 tree. We'll see what one of those looks like later. Strictly speaking,
812 there's three things going on here.
814 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
815 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
816 B<is> a YACC grammar for Perl!) The job of the parser is to take your
817 code and "understand" it, splitting it into sentences, deciding which
818 operands go with which operators and so on.
820 The parser is nobly assisted by the lexer, which chunks up your input
821 into tokens, and decides what type of thing each token is: a variable
822 name, an operator, a bareword, a subroutine, a core function, and so on.
823 The main point of entry to the lexer is C<yylex>, and that and its
824 associated routines can be found in F<toke.c>. Perl isn't much like
825 other computer languages; it's highly context sensitive at times, it can
826 be tricky to work out what sort of token something is, or where a token
827 ends. As such, there's a lot of interplay between the tokeniser and the
828 parser, which can get pretty frightening if you're not used to it.
830 As the parser understands a Perl program, it builds up a tree of
831 operations for the interpreter to perform during execution. The routines
832 which construct and link together the various operations are to be found
833 in F<op.c>, and will be examined later.
837 Now the parsing stage is complete, and the finished tree represents
838 the operations that the Perl interpreter needs to perform to execute our
839 program. Next, Perl does a dry run over the tree looking for
840 optimisations: constant expressions such as C<3 + 4> will be computed
841 now, and the optimizer will also see if any multiple operations can be
842 replaced with a single one. For instance, to fetch the variable C<$foo>,
843 instead of grabbing the glob C<*foo> and looking at the scalar
844 component, the optimizer fiddles the op tree to use a function which
845 directly looks up the scalar in question. The main optimizer is C<peep>
846 in F<op.c>, and many ops have their own optimizing functions.
850 Now we're finally ready to go: we have compiled Perl byte code, and all
851 that's left to do is run it. The actual execution is done by the
852 C<runops_standard> function in F<run.c>; more specifically, it's done by
853 these three innocent looking lines:
855 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
859 You may be more comfortable with the Perl version of that:
861 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
863 Well, maybe not. Anyway, each op contains a function pointer, which
864 stipulates the function which will actually carry out the operation.
865 This function will return the next op in the sequence - this allows for
866 things like C<if> which choose the next op dynamically at run time.
867 The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
868 execution if required.
870 The actual functions called are known as PP code, and they're spread
871 between four files: F<pp_hot.c> contains the "hot" code, which is most
872 often used and highly optimized, F<pp_sys.c> contains all the
873 system-specific functions, F<pp_ctl.c> contains the functions which
874 implement control structures (C<if>, C<while> and the like) and F<pp.c>
875 contains everything else. These are, if you like, the C code for Perl's
876 built-in functions and operators.
878 Note that each C<pp_> function is expected to return a pointer to the next
879 op. Calls to perl subs (and eval blocks) are handled within the same
880 runops loop, and do not consume extra space on the C stack. For example,
881 C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
882 struct onto the context stack which contain the address of the op
883 following the sub call or eval. They then return the first op of that sub
884 or eval block, and so execution continues of that sub or block. Later, a
885 C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
886 retrieves the return op from it, and returns it.
888 =item Exception handing
890 Perl's exception handing (i.e. C<die> etc) is built on top of the low-level
891 C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
892 way to capture the current PC and SP registers and later restore them; i.e.
893 a C<longjmp()> continues at the point in code where a previous C<setjmp()>
894 was done, with anything further up on the C stack being lost. This is why
895 code should always save values using C<SAVE_FOO> rather than in auto
898 The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
899 C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
900 C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
901 C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
903 At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
904 C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
905 loop or whatever, and handle possible exception returns. For a 2 return,
906 final cleanup is performed, such as popping stacks and calling C<CHECK> or
907 C<END> blocks. Amongst other things, this is how scope cleanup still
908 occurs during an C<exit>.
910 If a C<die> can find a C<CxEVAL> block on the context stack, then the
911 stack is popped to that level and the return op in that block is assigned
912 to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
913 passes control back to the guard. In the case of C<perl_run> and
914 C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
915 loop. The is the normal way that C<die> or C<croak> is handled within an
918 Sometimes ops are executed within an inner runops loop, such as tie, sort
919 or overload code. In this case, something like
921 sub FETCH { eval { die } }
923 would cause a longjmp right back to the guard in C<perl_run>, popping both
924 runops loops, which is clearly incorrect. One way to avoid this is for the
925 tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
926 runops loop, but for efficiency reasons, perl in fact just sets a flag,
927 using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
928 C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
929 which does a C<JMPENV_PUSH> and starts a new runops level to execute the
930 code, rather than doing it on the current loop.
932 As a further optimisation, on exit from the eval block in the C<FETCH>,
933 execution of the code following the block is still carried on in the inner
934 loop. When an exception is raised, C<docatch> compares the C<JMPENV>
935 level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
936 re-throws the exception. In this way any inner loops get popped.
940 1: eval { tie @a, 'A' };
946 To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
947 enters a runops loop. This loop executes the eval and tie ops on line 1,
948 with the eval pushing a C<CxEVAL> onto the context stack.
950 The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
951 to execute the body of C<TIEARRAY>. When it executes the entertry op on
952 line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
953 does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
954 the die op. At this point the C call stack looks like this:
957 Perl_runops # third loop
961 Perl_runops # second loop
965 Perl_runops # first loop
970 and the context and data stacks, as shown by C<-Dstv>, look like:
974 CX 1: EVAL => AV() PV("A"\0)
982 The die pops the first C<CxEVAL> off the context stack, sets
983 C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
984 the top C<docatch>. This then starts another third-level runops level,
985 which executes the nextstate, pushmark and die ops on line 4. At the point
986 that the second C<pp_die> is called, the C call stack looks exactly like
987 that above, even though we are no longer within an inner eval; this is
988 because of the optimization mentioned earlier. However, the context stack
989 now looks like this, ie with the top CxEVAL popped:
993 CX 1: EVAL => AV() PV("A"\0)
999 The die on line 4 pops the context stack back down to the CxEVAL, leaving
1005 As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
1006 C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
1010 Perl_runops # second loop
1014 Perl_runops # first loop
1019 In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
1020 differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
1021 and the C stack unwinds to:
1026 Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
1027 and execution continues.
1031 =head2 Internal Variable Types
1033 You should by now have had a look at L<perlguts>, which tells you about
1034 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
1037 These variables are used not only to represent Perl-space variables, but
1038 also any constants in the code, as well as some structures completely
1039 internal to Perl. The symbol table, for instance, is an ordinary Perl
1040 hash. Your code is represented by an SV as it's read into the parser;
1041 any program files you call are opened via ordinary Perl filehandles, and
1044 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
1045 Perl program. Let's see, for instance, how Perl treats the constant
1048 % perl -MDevel::Peek -e 'Dump("hello")'
1049 1 SV = PV(0xa041450) at 0xa04ecbc
1051 3 FLAGS = (POK,READONLY,pPOK)
1052 4 PV = 0xa0484e0 "hello"\0
1056 Reading C<Devel::Peek> output takes a bit of practise, so let's go
1057 through it line by line.
1059 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
1060 memory. SVs themselves are very simple structures, but they contain a
1061 pointer to a more complex structure. In this case, it's a PV, a
1062 structure which holds a string value, at location C<0xa041450>. Line 2
1063 is the reference count; there are no other references to this data, so
1066 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
1067 read-only SV (because it's a constant) and the data is a PV internally.
1068 Next we've got the contents of the string, starting at location
1071 Line 5 gives us the current length of the string - note that this does
1072 B<not> include the null terminator. Line 6 is not the length of the
1073 string, but the length of the currently allocated buffer; as the string
1074 grows, Perl automatically extends the available storage via a routine
1077 You can get at any of these quantities from C very easily; just add
1078 C<Sv> to the name of the field shown in the snippet, and you've got a
1079 macro which will return the value: C<SvCUR(sv)> returns the current
1080 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
1081 C<SvPV(sv, len)> returns the string itself with its length, and so on.
1082 More macros to manipulate these properties can be found in L<perlguts>.
1084 Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
1087 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
1092 6 junk = SvPV_force(sv, tlen);
1093 7 SvGROW(sv, tlen + len + 1);
1096 10 Move(ptr,SvPVX(sv)+tlen,len,char);
1097 11 SvCUR(sv) += len;
1098 12 *SvEND(sv) = '\0';
1099 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
1103 This is a function which adds a string, C<ptr>, of length C<len> onto
1104 the end of the PV stored in C<sv>. The first thing we do in line 6 is
1105 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
1106 macro to force a PV. As a side effect, C<tlen> gets set to the current
1107 value of the PV, and the PV itself is returned to C<junk>.
1109 In line 7, we make sure that the SV will have enough room to accommodate
1110 the old string, the new string and the null terminator. If C<LEN> isn't
1111 big enough, C<SvGROW> will reallocate space for us.
1113 Now, if C<junk> is the same as the string we're trying to add, we can
1114 grab the string directly from the SV; C<SvPVX> is the address of the PV
1117 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
1118 memory around: we move the string C<ptr> to the end of the PV - that's
1119 the start of the PV plus its current length. We're moving C<len> bytes
1120 of type C<char>. After doing so, we need to tell Perl we've extended the
1121 string, by altering C<CUR> to reflect the new length. C<SvEND> is a
1122 macro which gives us the end of the string, so that needs to be a
1125 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
1126 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1127 want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
1128 version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
1129 and turns on POK. The final C<SvTAINT> is a macro which launders tainted
1130 data if taint mode is turned on.
1132 AVs and HVs are more complicated, but SVs are by far the most common
1133 variable type being thrown around. Having seen something of how we
1134 manipulate these, let's go on and look at how the op tree is
1139 First, what is the op tree, anyway? The op tree is the parsed
1140 representation of your program, as we saw in our section on parsing, and
1141 it's the sequence of operations that Perl goes through to execute your
1142 program, as we saw in L</Running>.
1144 An op is a fundamental operation that Perl can perform: all the built-in
1145 functions and operators are ops, and there are a series of ops which
1146 deal with concepts the interpreter needs internally - entering and
1147 leaving a block, ending a statement, fetching a variable, and so on.
1149 The op tree is connected in two ways: you can imagine that there are two
1150 "routes" through it, two orders in which you can traverse the tree.
1151 First, parse order reflects how the parser understood the code, and
1152 secondly, execution order tells perl what order to perform the
1155 The easiest way to examine the op tree is to stop Perl after it has
1156 finished parsing, and get it to dump out the tree. This is exactly what
1157 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
1158 and L<B::Debug|B::Debug> do.
1160 Let's have a look at how Perl sees C<$a = $b + $c>:
1162 % perl -MO=Terse -e '$a=$b+$c'
1163 1 LISTOP (0x8179888) leave
1164 2 OP (0x81798b0) enter
1165 3 COP (0x8179850) nextstate
1166 4 BINOP (0x8179828) sassign
1167 5 BINOP (0x8179800) add [1]
1168 6 UNOP (0x81796e0) null [15]
1169 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
1170 8 UNOP (0x81797e0) null [15]
1171 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
1172 10 UNOP (0x816b4f0) null [15]
1173 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
1175 Let's start in the middle, at line 4. This is a BINOP, a binary
1176 operator, which is at location C<0x8179828>. The specific operator in
1177 question is C<sassign> - scalar assignment - and you can find the code
1178 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
1179 binary operator, it has two children: the add operator, providing the
1180 result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
1183 Line 10 is the null op: this does exactly nothing. What is that doing
1184 there? If you see the null op, it's a sign that something has been
1185 optimized away after parsing. As we mentioned in L</Optimization>,
1186 the optimization stage sometimes converts two operations into one, for
1187 example when fetching a scalar variable. When this happens, instead of
1188 rewriting the op tree and cleaning up the dangling pointers, it's easier
1189 just to replace the redundant operation with the null op. Originally,
1190 the tree would have looked like this:
1192 10 SVOP (0x816b4f0) rv2sv [15]
1193 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
1195 That is, fetch the C<a> entry from the main symbol table, and then look
1196 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
1197 happens to do both these things.
1199 The right hand side, starting at line 5 is similar to what we've just
1200 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
1203 Now, what's this about?
1205 1 LISTOP (0x8179888) leave
1206 2 OP (0x81798b0) enter
1207 3 COP (0x8179850) nextstate
1209 C<enter> and C<leave> are scoping ops, and their job is to perform any
1210 housekeeping every time you enter and leave a block: lexical variables
1211 are tidied up, unreferenced variables are destroyed, and so on. Every
1212 program will have those first three lines: C<leave> is a list, and its
1213 children are all the statements in the block. Statements are delimited
1214 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
1215 the ops to be performed for each statement being the children of
1216 C<nextstate>. C<enter> is a single op which functions as a marker.
1218 That's how Perl parsed the program, from top to bottom:
1231 However, it's impossible to B<perform> the operations in this order:
1232 you have to find the values of C<$b> and C<$c> before you add them
1233 together, for instance. So, the other thread that runs through the op
1234 tree is the execution order: each op has a field C<op_next> which points
1235 to the next op to be run, so following these pointers tells us how perl
1236 executes the code. We can traverse the tree in this order using
1237 the C<exec> option to C<B::Terse>:
1239 % perl -MO=Terse,exec -e '$a=$b+$c'
1240 1 OP (0x8179928) enter
1241 2 COP (0x81798c8) nextstate
1242 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
1243 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
1244 5 BINOP (0x8179878) add [1]
1245 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
1246 7 BINOP (0x81798a0) sassign
1247 8 LISTOP (0x8179900) leave
1249 This probably makes more sense for a human: enter a block, start a
1250 statement. Get the values of C<$b> and C<$c>, and add them together.
1251 Find C<$a>, and assign one to the other. Then leave.
1253 The way Perl builds up these op trees in the parsing process can be
1254 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1255 piece we need to construct the tree for C<$a = $b + $c>
1257 1 term : term ASSIGNOP term
1258 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1260 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1262 If you're not used to reading BNF grammars, this is how it works: You're
1263 fed certain things by the tokeniser, which generally end up in upper
1264 case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1265 code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
1266 "terminal symbols", because you can't get any simpler than them.
1268 The grammar, lines one and three of the snippet above, tells you how to
1269 build up more complex forms. These complex forms, "non-terminal symbols"
1270 are generally placed in lower case. C<term> here is a non-terminal
1271 symbol, representing a single expression.
1273 The grammar gives you the following rule: you can make the thing on the
1274 left of the colon if you see all the things on the right in sequence.
1275 This is called a "reduction", and the aim of parsing is to completely
1276 reduce the input. There are several different ways you can perform a
1277 reduction, separated by vertical bars: so, C<term> followed by C<=>
1278 followed by C<term> makes a C<term>, and C<term> followed by C<+>
1279 followed by C<term> can also make a C<term>.
1281 So, if you see two terms with an C<=> or C<+>, between them, you can
1282 turn them into a single expression. When you do this, you execute the
1283 code in the block on the next line: if you see C<=>, you'll do the code
1284 in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1285 which contributes to the op tree.
1288 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1290 What this does is creates a new binary op, and feeds it a number of
1291 variables. The variables refer to the tokens: C<$1> is the first token in
1292 the input, C<$2> the second, and so on - think regular expression
1293 backreferences. C<$$> is the op returned from this reduction. So, we
1294 call C<newBINOP> to create a new binary operator. The first parameter to
1295 C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1296 operator, so we want the type to be C<ADDOP>. We could specify this
1297 directly, but it's right there as the second token in the input, so we
1298 use C<$2>. The second parameter is the op's flags: 0 means "nothing
1299 special". Then the things to add: the left and right hand side of our
1300 expression, in scalar context.
1304 When perl executes something like C<addop>, how does it pass on its
1305 results to the next op? The answer is, through the use of stacks. Perl
1306 has a number of stacks to store things it's currently working on, and
1307 we'll look at the three most important ones here.
1311 =item Argument stack
1313 Arguments are passed to PP code and returned from PP code using the
1314 argument stack, C<ST>. The typical way to handle arguments is to pop
1315 them off the stack, deal with them how you wish, and then push the result
1316 back onto the stack. This is how, for instance, the cosine operator
1321 value = Perl_cos(value);
1324 We'll see a more tricky example of this when we consider Perl's macros
1325 below. C<POPn> gives you the NV (floating point value) of the top SV on
1326 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1327 the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1328 should be extended if necessary - it can't be necessary here, because we
1329 know there's room for one more item on the stack, since we've just
1330 removed one! The C<XPUSH*> macros at least guarantee safety.
1332 Alternatively, you can fiddle with the stack directly: C<SP> gives you
1333 the first element in your portion of the stack, and C<TOP*> gives you
1334 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1335 negation of an integer:
1339 Just set the integer value of the top stack entry to its negation.
1341 Argument stack manipulation in the core is exactly the same as it is in
1342 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1343 description of the macros used in stack manipulation.
1347 I say "your portion of the stack" above because PP code doesn't
1348 necessarily get the whole stack to itself: if your function calls
1349 another function, you'll only want to expose the arguments aimed for the
1350 called function, and not (necessarily) let it get at your own data. The
1351 way we do this is to have a "virtual" bottom-of-stack, exposed to each
1352 function. The mark stack keeps bookmarks to locations in the argument
1353 stack usable by each function. For instance, when dealing with a tied
1354 variable, (internally, something with "P" magic) Perl has to call
1355 methods for accesses to the tied variables. However, we need to separate
1356 the arguments exposed to the method to the argument exposed to the
1357 original function - the store or fetch or whatever it may be. Here's
1358 roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
1362 3 PUSHs(SvTIED_obj((SV*)av, mg));
1366 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1369 Let's examine the whole implementation, for practice:
1373 Push the current state of the stack pointer onto the mark stack. This is
1374 so that when we've finished adding items to the argument stack, Perl
1375 knows how many things we've added recently.
1378 3 PUSHs(SvTIED_obj((SV*)av, mg));
1381 We're going to add two more items onto the argument stack: when you have
1382 a tied array, the C<PUSH> subroutine receives the object and the value
1383 to be pushed, and that's exactly what we have here - the tied object,
1384 retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1388 Next we tell Perl to update the global stack pointer from our internal
1389 variable: C<dSP> only gave us a local copy, not a reference to the global.
1392 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1395 C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1396 variables are tidied up, everything that has been localised gets
1397 its previous value returned, and so on. Think of them as the C<{> and
1398 C<}> of a Perl block.
1400 To actually do the magic method call, we have to call a subroutine in
1401 Perl space: C<call_method> takes care of that, and it's described in
1402 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1403 going to discard its return value. The call_method() function
1404 removes the top element of the mark stack, so there is nothing for
1405 the caller to clean up.
1409 C doesn't have a concept of local scope, so perl provides one. We've
1410 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1411 stack implements the C equivalent of, for example:
1418 See L<perlguts/Localising Changes> for how to use the save stack.
1422 =head2 Millions of Macros
1424 One thing you'll notice about the Perl source is that it's full of
1425 macros. Some have called the pervasive use of macros the hardest thing
1426 to understand, others find it adds to clarity. Let's take an example,
1427 the code which implements the addition operator:
1431 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1434 6 SETn( left + right );
1439 Every line here (apart from the braces, of course) contains a macro. The
1440 first line sets up the function declaration as Perl expects for PP code;
1441 line 3 sets up variable declarations for the argument stack and the
1442 target, the return value of the operation. Finally, it tries to see if
1443 the addition operation is overloaded; if so, the appropriate subroutine
1446 Line 5 is another variable declaration - all variable declarations start
1447 with C<d> - which pops from the top of the argument stack two NVs (hence
1448 C<nn>) and puts them into the variables C<right> and C<left>, hence the
1449 C<rl>. These are the two operands to the addition operator. Next, we
1450 call C<SETn> to set the NV of the return value to the result of adding
1451 the two values. This done, we return - the C<RETURN> macro makes sure
1452 that our return value is properly handled, and we pass the next operator
1453 to run back to the main run loop.
1455 Most of these macros are explained in L<perlapi>, and some of the more
1456 important ones are explained in L<perlxs> as well. Pay special attention
1457 to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1458 the C<[pad]THX_?> macros.
1460 =head2 The .i Targets
1462 You can expand the macros in a F<foo.c> file by saying
1466 which will expand the macros using cpp. Don't be scared by the results.
1468 =head1 SOURCE CODE STATIC ANALYSIS
1470 Various tools exist for analysing C source code B<statically>, as
1471 opposed to B<dynamically>, that is, without executing the code.
1472 It is possible to detect resource leaks, undefined behaviour, type
1473 mismatches, portability problems, code paths that would cause illegal
1474 memory accesses, and other similar problems by just parsing the C code
1475 and looking at the resulting graph, what does it tell about the
1476 execution and data flows. As a matter of fact, this is exactly
1477 how C compilers know to give warnings about dubious code.
1481 The good old C code quality inspector, C<lint>, is available in
1482 several platforms, but please be aware that there are several
1483 different implementations of it by different vendors, which means that
1484 the flags are not identical across different platforms.
1486 There is a lint variant called C<splint> (Secure Programming Lint)
1487 available from http://www.splint.org/ that should compile on any
1490 There are C<lint> and <splint> targets in Makefile, but you may have
1491 to diddle with the flags (see above).
1495 Coverity (http://www.coverity.com/) is a product similar to lint and
1496 as a testbed for their product they periodically check several open
1497 source projects, and they give out accounts to open source developers
1498 to the defect databases.
1500 =head2 cpd (cut-and-paste detector)
1502 The cpd tool detects cut-and-paste coding. If one instance of the
1503 cut-and-pasted code changes, all the other spots should probably be
1504 changed, too. Therefore such code should probably be turned into a
1505 subroutine or a macro.
1507 cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1508 (http://pmd.sourceforge.net/). pmd was originally written for static
1509 analysis of Java code, but later the cpd part of it was extended to
1510 parse also C and C++.
1512 Download the pmd-X.y.jar from the SourceForge site, and then run
1513 it on source code thusly:
1515 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1517 You may run into memory limits, in which case you should use the -Xmx option:
1523 Though much can be written about the inconsistency and coverage
1524 problems of gcc warnings (like C<-Wall> not meaning "all the
1525 warnings", or some common portability problems not being covered by
1526 C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1527 collection of warnings, and so forth), gcc is still a useful tool in
1528 keeping our coding nose clean.
1530 The C<-Wall> is by default on.
1532 The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1533 always, but unfortunately they are not safe on all platforms, they can
1534 for example cause fatal conflicts with the system headers (Solaris
1535 being a prime example). If Configure C<-Dgccansipedantic> is used,
1536 the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1537 where they are known to be safe.
1539 Starting from Perl 5.9.4 the following extra flags are added:
1553 C<-Wdeclaration-after-statement>
1557 The following flags would be nice to have but they would first need
1558 their own Stygian stablemaster:
1572 C<-Wstrict-prototypes>
1576 The C<-Wtraditional> is another example of the annoying tendency of
1577 gcc to bundle a lot of warnings under one switch -- it would be
1578 impossible to deploy in practice because it would complain a lot -- but
1579 it does contain some warnings that would be beneficial to have available
1580 on their own, such as the warning about string constants inside macros
1581 containing the macro arguments: this behaved differently pre-ANSI
1582 than it does in ANSI, and some C compilers are still in transition,
1583 AIX being an example.
1585 =head2 Warnings of other C compilers
1587 Other C compilers (yes, there B<are> other C compilers than gcc) often
1588 have their "strict ANSI" or "strict ANSI with some portability extensions"
1589 modes on, like for example the Sun Workshop has its C<-Xa> mode on
1590 (though implicitly), or the DEC (these days, HP...) has its C<-std1>
1595 You can compile a special debugging version of Perl, which allows you
1596 to use the C<-D> option of Perl to tell more about what Perl is doing.
1597 But sometimes there is no alternative than to dive in with a debugger,
1598 either to see the stack trace of a core dump (very useful in a bug
1599 report), or trying to figure out what went wrong before the core dump
1600 happened, or how did we end up having wrong or unexpected results.
1602 =head2 Poking at Perl
1604 To really poke around with Perl, you'll probably want to build Perl for
1605 debugging, like this:
1607 ./Configure -d -D optimize=-g
1610 C<-g> is a flag to the C compiler to have it produce debugging
1611 information which will allow us to step through a running program,
1612 and to see in which C function we are at (without the debugging
1613 information we might see only the numerical addresses of the functions,
1614 which is not very helpful).
1616 F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1617 enables all the internal debugging code in Perl. There are a whole bunch
1618 of things you can debug with this: L<perlrun> lists them all, and the
1619 best way to find out about them is to play about with them. The most
1620 useful options are probably
1622 l Context (loop) stack processing
1624 o Method and overloading resolution
1625 c String/numeric conversions
1627 Some of the functionality of the debugging code can be achieved using XS
1630 -Dr => use re 'debug'
1631 -Dx => use O 'Debug'
1633 =head2 Using a source-level debugger
1635 If the debugging output of C<-D> doesn't help you, it's time to step
1636 through perl's execution with a source-level debugger.
1642 We'll use C<gdb> for our examples here; the principles will apply to
1643 any debugger (many vendors call their debugger C<dbx>), but check the
1644 manual of the one you're using.
1648 To fire up the debugger, type
1652 Or if you have a core dump:
1656 You'll want to do that in your Perl source tree so the debugger can read
1657 the source code. You should see the copyright message, followed by the
1662 C<help> will get you into the documentation, but here are the most
1669 Run the program with the given arguments.
1671 =item break function_name
1673 =item break source.c:xxx
1675 Tells the debugger that we'll want to pause execution when we reach
1676 either the named function (but see L<perlguts/Internal Functions>!) or the given
1677 line in the named source file.
1681 Steps through the program a line at a time.
1685 Steps through the program a line at a time, without descending into
1690 Run until the next breakpoint.
1694 Run until the end of the current function, then stop again.
1698 Just pressing Enter will do the most recent operation again - it's a
1699 blessing when stepping through miles of source code.
1703 Execute the given C code and print its results. B<WARNING>: Perl makes
1704 heavy use of macros, and F<gdb> does not necessarily support macros
1705 (see later L</"gdb macro support">). You'll have to substitute them
1706 yourself, or to invoke cpp on the source code files
1707 (see L</"The .i Targets">)
1708 So, for instance, you can't say
1710 print SvPV_nolen(sv)
1714 print Perl_sv_2pv_nolen(sv)
1718 You may find it helpful to have a "macro dictionary", which you can
1719 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1720 recursively apply those macros for you.
1722 =head2 gdb macro support
1724 Recent versions of F<gdb> have fairly good macro support, but
1725 in order to use it you'll need to compile perl with macro definitions
1726 included in the debugging information. Using F<gcc> version 3.1, this
1727 means configuring with C<-Doptimize=-g3>. Other compilers might use a
1728 different switch (if they support debugging macros at all).
1730 =head2 Dumping Perl Data Structures
1732 One way to get around this macro hell is to use the dumping functions in
1733 F<dump.c>; these work a little like an internal
1734 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1735 that you can't get at from Perl. Let's take an example. We'll use the
1736 C<$a = $b + $c> we used before, but give it a bit of context:
1737 C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1739 What about C<pp_add>, the function we examined earlier to implement the
1742 (gdb) break Perl_pp_add
1743 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1745 Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
1746 With the breakpoint in place, we can run our program:
1748 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1750 Lots of junk will go past as gdb reads in the relevant source files and
1751 libraries, and then:
1753 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
1754 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1759 We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1760 arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1763 #define dPOPTOPnnrl_ul NV right = POPn; \
1764 SV *leftsv = TOPs; \
1765 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1767 C<POPn> takes the SV from the top of the stack and obtains its NV either
1768 directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1769 C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1770 C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1771 C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1773 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1774 convert it. If we step again, we'll find ourselves there:
1776 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1780 We can now use C<Perl_sv_dump> to investigate the SV:
1782 SV = PV(0xa057cc0) at 0xa0675d0
1785 PV = 0xa06a510 "6XXXX"\0
1790 We know we're going to get C<6> from this, so let's finish the
1794 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1795 0x462669 in Perl_pp_add () at pp_hot.c:311
1798 We can also dump out this op: the current op is always stored in
1799 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1800 similar output to L<B::Debug|B::Debug>.
1803 13 TYPE = add ===> 14
1805 FLAGS = (SCALAR,KIDS)
1807 TYPE = null ===> (12)
1809 FLAGS = (SCALAR,KIDS)
1811 11 TYPE = gvsv ===> 12
1817 # finish this later #
1821 All right, we've now had a look at how to navigate the Perl sources and
1822 some things you'll need to know when fiddling with them. Let's now get
1823 on and create a simple patch. Here's something Larry suggested: if a
1824 C<U> is the first active format during a C<pack>, (for example,
1825 C<pack "U3C8", @stuff>) then the resulting string should be treated as
1828 How do we prepare to fix this up? First we locate the code in question -
1829 the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1830 files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1831 altering this file, let's copy it to F<pp.c~>.
1833 [Well, it was in F<pp.c> when this tutorial was written. It has now been
1834 split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1836 Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1837 loop over the pattern, taking each format character in turn into
1838 C<datum_type>. Then for each possible format character, we swallow up
1839 the other arguments in the pattern (a field width, an asterisk, and so
1840 on) and convert the next chunk input into the specified format, adding
1841 it onto the output SV C<cat>.
1843 How do we know if the C<U> is the first format in the C<pat>? Well, if
1844 we have a pointer to the start of C<pat> then, if we see a C<U> we can
1845 test whether we're still at the start of the string. So, here's where
1849 register char *pat = SvPVx(*++MARK, fromlen);
1850 register char *patend = pat + fromlen;
1855 We'll have another string pointer in there:
1858 register char *pat = SvPVx(*++MARK, fromlen);
1859 register char *patend = pat + fromlen;
1865 And just before we start the loop, we'll set C<patcopy> to be the start
1870 sv_setpvn(cat, "", 0);
1872 while (pat < patend) {
1874 Now if we see a C<U> which was at the start of the string, we turn on
1875 the C<UTF8> flag for the output SV, C<cat>:
1877 + if (datumtype == 'U' && pat==patcopy+1)
1879 if (datumtype == '#') {
1880 while (pat < patend && *pat != '\n')
1883 Remember that it has to be C<patcopy+1> because the first character of
1884 the string is the C<U> which has been swallowed into C<datumtype!>
1886 Oops, we forgot one thing: what if there are spaces at the start of the
1887 pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1888 character, even though it's not the first thing in the pattern. In this
1889 case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1891 if (isSPACE(datumtype))
1896 if (isSPACE(datumtype)) {
1901 OK. That's the C part done. Now we must do two additional things before
1902 this patch is ready to go: we've changed the behaviour of Perl, and so
1903 we must document that change. We must also provide some more regression
1904 tests to make sure our patch works and doesn't create a bug somewhere
1905 else along the line.
1907 The regression tests for each operator live in F<t/op/>, and so we
1908 make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1909 tests to the end. First, we'll test that the C<U> does indeed create
1912 t/op/pack.t has a sensible ok() function, but if it didn't we could
1913 use the one from t/test.pl.
1915 require './test.pl';
1916 plan( tests => 159 );
1920 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1921 print "ok $test\n"; $test++;
1923 we can write the more sensible (see L<Test::More> for a full
1924 explanation of is() and other testing functions).
1926 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
1927 "U* produces unicode" );
1929 Now we'll test that we got that space-at-the-beginning business right:
1931 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
1932 " with spaces at the beginning" );
1934 And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1935 the first active format:
1937 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
1938 "U* not first isn't unicode" );
1940 Mustn't forget to change the number of tests which appears at the top,
1941 or else the automated tester will get confused. This will either look
1948 plan( tests => 156 );
1950 We now compile up Perl, and run it through the test suite. Our new
1953 Finally, the documentation. The job is never done until the paperwork is
1954 over, so let's describe the change we've just made. The relevant place
1955 is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1956 this text in the description of C<pack>:
1960 If the pattern begins with a C<U>, the resulting string will be treated
1961 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1962 with an initial C<U0>, and the bytes that follow will be interpreted as
1963 Unicode characters. If you don't want this to happen, you can begin your
1964 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
1965 string, and then follow this with a C<U*> somewhere in your pattern.
1967 All done. Now let's create the patch. F<Porting/patching.pod> tells us
1968 that if we're making major changes, we should copy the entire directory
1969 to somewhere safe before we begin fiddling, and then do
1971 diff -ruN old new > patch
1973 However, we know which files we've changed, and we can simply do this:
1975 diff -u pp.c~ pp.c > patch
1976 diff -u t/op/pack.t~ t/op/pack.t >> patch
1977 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1979 We end up with a patch looking a little like this:
1981 --- pp.c~ Fri Jun 02 04:34:10 2000
1982 +++ pp.c Fri Jun 16 11:37:25 2000
1983 @@ -4375,6 +4375,7 @@
1986 register char *pat = SvPVx(*++MARK, fromlen);
1988 register char *patend = pat + fromlen;
1991 @@ -4405,6 +4406,7 @@
1994 And finally, we submit it, with our rationale, to perl5-porters. Job
1997 =head2 Patching a core module
1999 This works just like patching anything else, with an extra
2000 consideration. Many core modules also live on CPAN. If this is so,
2001 patch the CPAN version instead of the core and send the patch off to
2002 the module maintainer (with a copy to p5p). This will help the module
2003 maintainer keep the CPAN version in sync with the core version without
2004 constantly scanning p5p.
2006 The list of maintainers of core modules is usefully documented in
2007 F<Porting/Maintainers.pl>.
2009 =head2 Adding a new function to the core
2011 If, as part of a patch to fix a bug, or just because you have an
2012 especially good idea, you decide to add a new function to the core,
2013 discuss your ideas on p5p well before you start work. It may be that
2014 someone else has already attempted to do what you are considering and
2015 can give lots of good advice or even provide you with bits of code
2016 that they already started (but never finished).
2018 You have to follow all of the advice given above for patching. It is
2019 extremely important to test any addition thoroughly and add new tests
2020 to explore all boundary conditions that your new function is expected
2021 to handle. If your new function is used only by one module (e.g. toke),
2022 then it should probably be named S_your_function (for static); on the
2023 other hand, if you expect it to accessible from other functions in
2024 Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
2027 The location of any new code is also an important consideration. Don't
2028 just create a new top level .c file and put your code there; you would
2029 have to make changes to Configure (so the Makefile is created properly),
2030 as well as possibly lots of include files. This is strictly pumpking
2033 It is better to add your function to one of the existing top level
2034 source code files, but your choice is complicated by the nature of
2035 the Perl distribution. Only the files that are marked as compiled
2036 static are located in the perl executable. Everything else is located
2037 in the shared library (or DLL if you are running under WIN32). So,
2038 for example, if a function was only used by functions located in
2039 toke.c, then your code can go in toke.c. If, however, you want to call
2040 the function from universal.c, then you should put your code in another
2041 location, for example util.c.
2043 In addition to writing your c-code, you will need to create an
2044 appropriate entry in embed.pl describing your function, then run
2045 'make regen_headers' to create the entries in the numerous header
2046 files that perl needs to compile correctly. See L<perlguts/Internal Functions>
2047 for information on the various options that you can set in embed.pl.
2048 You will forget to do this a few (or many) times and you will get
2049 warnings during the compilation phase. Make sure that you mention
2050 this when you post your patch to P5P; the pumpking needs to know this.
2052 When you write your new code, please be conscious of existing code
2053 conventions used in the perl source files. See L<perlstyle> for
2054 details. Although most of the guidelines discussed seem to focus on
2055 Perl code, rather than c, they all apply (except when they don't ;).
2056 See also I<Porting/patching.pod> file in the Perl source distribution
2057 for lots of details about both formatting and submitting patches of
2060 Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
2061 Test on as many platforms as you can find. Test as many perl
2062 Configure options as you can (e.g. MULTIPLICITY). If you have
2063 profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
2064 below for how to use them to further test your code. Remember that
2065 most of the people on P5P are doing this on their own time and
2066 don't have the time to debug your code.
2068 =head2 Writing a test
2070 Every module and built-in function has an associated test file (or
2071 should...). If you add or change functionality, you have to write a
2072 test. If you fix a bug, you have to write a test so that bug never
2073 comes back. If you alter the docs, it would be nice to test what the
2074 new documentation says.
2076 In short, if you submit a patch you probably also have to patch the
2079 For modules, the test file is right next to the module itself.
2080 F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
2081 so there are some snags (and it would be wonderful for you to brush
2082 them out), but it basically works that way. Everything else lives in
2089 Testing of the absolute basic functionality of Perl. Things like
2090 C<if>, basic file reads and writes, simple regexes, etc. These are
2091 run first in the test suite and if any of them fail, something is
2096 These test the basic control structures, C<if/else>, C<while>,
2101 Tests basic issues of how Perl parses and compiles itself.
2105 Tests for built-in IO functions, including command line arguments.
2109 The old home for the module tests, you shouldn't put anything new in
2110 here. There are still some bits and pieces hanging around in here
2111 that need to be moved. Perhaps you could move them? Thanks!
2115 Tests for perl's built in functions that don't fit into any of the
2120 Tests for POD directives. There are still some tests for the Pod
2121 modules hanging around in here that need to be moved out into F<lib/>.
2125 Testing features of how perl actually runs, including exit codes and
2126 handling of PERL* environment variables.
2130 Tests for the core support of Unicode.
2134 Windows-specific tests.
2138 A test suite for the s2p converter.
2142 The core uses the same testing style as the rest of Perl, a simple
2143 "ok/not ok" run through Test::Harness, but there are a few special
2146 There are three ways to write a test in the core. Test::More,
2147 t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
2148 decision of which to use depends on what part of the test suite you're
2149 working on. This is a measure to prevent a high-level failure (such
2150 as Config.pm breaking) from causing basic functionality tests to fail.
2156 Since we don't know if require works, or even subroutines, use ad hoc
2157 tests for these two. Step carefully to avoid using the feature being
2160 =item t/cmd t/run t/io t/op
2162 Now that basic require() and subroutines are tested, you can use the
2163 t/test.pl library which emulates the important features of Test::More
2164 while using a minimum of core features.
2166 You can also conditionally use certain libraries like Config, but be
2167 sure to skip the test gracefully if it's not there.
2171 Now that the core of Perl is tested, Test::More can be used. You can
2172 also use the full suite of core modules in the tests.
2176 When you say "make test" Perl uses the F<t/TEST> program to run the
2177 test suite (except under Win32 where it uses F<t/harness> instead.)
2178 All tests are run from the F<t/> directory, B<not> the directory
2179 which contains the test. This causes some problems with the tests
2180 in F<lib/>, so here's some opportunity for some patching.
2182 You must be triply conscious of cross-platform concerns. This usually
2183 boils down to using File::Spec and avoiding things like C<fork()> and
2184 C<system()> unless absolutely necessary.
2186 =head2 Special Make Test Targets
2188 There are various special make targets that can be used to test Perl
2189 slightly differently than the standard "test" target. Not all them
2190 are expected to give a 100% success rate. Many of them have several
2191 aliases, and many of them are not available on certain operating
2198 Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
2200 (Not available on Win32)
2204 Run all the tests through B::Deparse. Not all tests will succeed.
2206 (Not available on Win32)
2208 =item test.taintwarn
2210 Run all tests with the B<-t> command-line switch. Not all tests
2211 are expected to succeed (until they're specifically fixed, of course).
2213 (Not available on Win32)
2217 Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
2218 F<t/op>, and F<t/uni> tests.
2220 =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
2222 (Only in Linux) Run all the tests using the memory leak + naughty
2223 memory access tool "valgrind". The log files will be named
2224 F<testname.valgrind>.
2226 =item test.third check.third utest.third ucheck.third
2228 (Only in Tru64) Run all the tests using the memory leak + naughty
2229 memory access tool "Third Degree". The log files will be named
2230 F<perl.3log.testname>.
2232 =item test.torture torturetest
2234 Run all the usual tests and some extra tests. As of Perl 5.8.0 the
2235 only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
2237 You can also run the torture test with F<t/harness> by giving
2238 C<-torture> argument to F<t/harness>.
2240 =item utest ucheck test.utf8 check.utf8
2242 Run all the tests with -Mutf8. Not all tests will succeed.
2244 (Not available on Win32)
2246 =item minitest.utf16 test.utf16
2248 Runs the tests with UTF-16 encoded scripts, encoded with different
2249 versions of this encoding.
2251 C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
2252 C<-utf16> arguments to F<t/TEST>.
2254 (Not available on Win32)
2258 Run the test suite with the F<t/harness> controlling program, instead of
2259 F<t/TEST>. F<t/harness> is more sophisticated, and uses the
2260 L<Test::Harness> module, thus using this test target supposes that perl
2261 mostly works. The main advantage for our purposes is that it prints a
2262 detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
2263 doesn't redirect stderr to stdout.
2265 Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
2266 there is no special "test_harness" target.
2268 Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
2269 environment variables to control the behaviour of F<t/harness>. This means
2272 nmake test TEST_FILES="op/*.t"
2273 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
2275 =item test-notty test_notty
2277 Sets PERL_SKIP_TTY_TEST to true before running normal test.
2281 =head2 Running tests by hand
2283 You can run part of the test suite by hand by using one the following
2284 commands from the F<t/> directory :
2286 ./perl -I../lib TEST list-of-.t-files
2290 ./perl -I../lib harness list-of-.t-files
2292 (if you don't specify test scripts, the whole test suite will be run.)
2294 =head3 Using t/harness for testing
2296 If you use C<harness> for testing you have several command line options
2297 available to you. The arguments are as follows, and are in the order
2298 that they must appear if used together.
2300 harness -v -torture -re=pattern LIST OF FILES TO TEST
2301 harness -v -torture -re LIST OF PATTERNS TO MATCH
2303 If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
2304 the manifest. The file list may include shell wildcards which will be
2311 Run the tests under verbose mode so you can see what tests were run,
2316 Run the torture tests as well as the normal set.
2320 Filter the file list so that all the test files run match PATTERN.
2321 Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2322 in that it allows the file list to be provided as well.
2324 =item -re LIST OF PATTERNS
2326 Filter the file list so that all the test files run match
2327 /(LIST|OF|PATTERNS)/. Note that with this form the patterns
2328 are joined by '|' and you cannot supply a list of files, instead
2329 the test files are obtained from the MANIFEST.
2333 You can run an individual test by a command similar to
2335 ./perl -I../lib patho/to/foo.t
2337 except that the harnesses set up some environment variables that may
2338 affect the execution of the test :
2344 indicates that we're running this test part of the perl core test suite.
2345 This is useful for modules that have a dual life on CPAN.
2347 =item PERL_DESTRUCT_LEVEL=2
2349 is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2353 (used only by F<t/TEST>) if set, overrides the path to the perl executable
2354 that should be used to run the tests (the default being F<./perl>).
2356 =item PERL_SKIP_TTY_TEST
2358 if set, tells to skip the tests that need a terminal. It's actually set
2359 automatically by the Makefile, but can also be forced artificially by
2360 running 'make test_notty'.
2364 =head2 Common problems when patching Perl source code
2366 Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2367 some cases we have to take pre-ANSI requirements into consideration.
2368 You don't care about some particular platform having broken Perl?
2369 I hear there is still a strong demand for J2EE programmers.
2371 =head2 Perl environment problems
2377 Not compiling with threading
2379 Compiling with threading (-Duseithreads) completely rewrites
2380 the function prototypes of Perl. You better try your changes
2381 with that. Related to this is the difference between "Perl_-less"
2382 and "Perl_-ly" APIs, for example:
2384 Perl_sv_setiv(aTHX_ ...);
2387 The first one explicitly passes in the context, which is needed for e.g.
2388 threaded builds. The second one does that implicitly; do not get them
2389 mixed. If you are not passing in a aTHX_, you will need to do a dTHX as
2390 the first thing in the function.
2392 See L<perlguts/"How multiple interpreters and concurrency are supported">
2393 for further discussion about context.
2397 Not compiling with -DDEBUGGING
2399 The DEBUGGING define exposes more code to the compiler,
2400 therefore more ways for things to go wrong. You should try it.
2404 Introducing (non-read-only) globals
2406 Do not introduce any modifiable globals, truly global or file static.
2407 They are bad form and complicate multithreading and other forms of
2408 concurrency. The right way is to introduce them as new interpreter
2409 variables, see F<intrpvar.h> (at the very end for binary compatibility).
2411 Introducing read-only (const) globals is okay, as long as you verify
2412 with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2413 BSD-style output) that the data you added really is read-only.
2414 (If it is, it shouldn't show up in the output of that command.)
2416 If you want to have static strings, make them constant:
2418 static const char etc[] = "...";
2420 If you want to have arrays of constant strings, note carefully
2421 the right combination of C<const>s:
2423 static const char * const yippee[] =
2424 {"hi", "ho", "silver"};
2426 There is a way to completely hide any modifiable globals (they are all
2427 moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2428 It is not normally used, but can be used for testing, read more
2429 about it in L<perlhack>.
2433 Not exporting your new function
2435 Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2436 function that is part of the public API (the shared Perl library)
2437 to be explicitly marked as exported. See the discussion about
2438 F<embed.pl> in L<perlguts>.
2442 Exporting your new function
2444 The new shiny result of either genuine new functionality or your
2445 arduous refactoring is now ready and correctly exported. So what
2446 could possibly be wrong?
2448 Maybe simply that your function did not need to be exported in the
2449 first place. Perl has a long and not so glorious history of exporting
2450 functions that it should not have.
2452 If the function is used only inside one source code file, make it
2453 static. See the discussion about F<embed.pl> in L<perlguts>.
2455 If the function is used across several files, but intended only for
2456 Perl's internal use (and this should be the common case), do not
2457 export it to the public API. See the discussion about F<embed.pl>
2462 =head2 Portability problems
2464 The following are common causes of compilation and/or execution
2465 failures, not common to Perl as such. The C FAQ is good bedtime
2466 reading. Please test your changes with as many C compilers and
2467 platforms as possible -- we will, anyway, and it's nice to save
2468 oneself from public embarrassment.
2470 If using gcc, you can add the C<-std=c89> option which will hopefully
2471 catch most of these unportabilities. (However it might also catch
2472 incompatibilities in your system's header files.)
2474 Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2475 C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2477 If using the C<gcc -Wall> note that not all the possible
2478 warnings are given unless you also compile with C<-O>.
2480 Also study L<perlport> carefully to avoid any bad assumptions
2481 about the operating system, filesystem, and so forth.
2483 You may once in a while try a "make miniperl" to see whether we
2484 can still compile Perl with just the bare minimum of interfaces.
2486 Do not assume an operating system indicates a certain compiler.
2492 Casting pointers to integers or casting integers to pointers
2494 void castaway(U8* p)
2500 void castaway(U8* p)
2504 Both are bad, and broken, and unportable. Use the PTR2IV()
2505 macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2506 INT2PTR(), and NUM2PTR().)
2510 Casting between data function pointers and data pointers
2512 Technically speaking casting between function pointers and data
2513 pointers is unportable and undefined, but practically speaking
2514 it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
2515 macros. Sometimes you can also play games with unions.
2519 Assuming sizeof(int) == sizeof(long)
2521 There are platforms where longs are 64 bits, and platforms where ints
2522 are 64 bits, and while we are out to shock you, even platforms where
2523 shorts are 64 bits. This is all legal according to the C standard.
2524 (In other words, "long long" is not a portable way to specify 64 bits,
2525 and "long long" is not even guaranteed to be any wider than "long".)
2527 Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2528 Avoid things like I32 because they are B<not> guaranteed to be
2529 I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2530 guaranteed to be B<int> or B<long>. If you really explicitly need
2531 64-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
2535 Assuming one can dereference any type of pointer for any type of data
2540 Many platforms, quite rightly so, will give you a core dump instead
2541 of a pony if the p happens not be correctly aligned.
2549 Simply not portable. Get your lvalue to be of the right type,
2550 or maybe use temporary variables, or dirty tricks with unions.
2554 Assume B<anything> about structs
2560 That a certain field exists in a struct
2564 That no other fields exist besides the ones you know of know of
2568 That a field is a certain signedness, sizeof, or type
2572 That the fields are in a certain order
2576 That the sizeof(struct) is the same everywhere
2580 That there is no padding between the fields
2584 That there are no alignment requirements for the fields
2590 Mixing #define and #ifdef
2592 #define BURGLE(x) ... \
2593 #ifdef BURGLE_OLD_STYLE
2594 ... do it the old way ... \
2596 ... do it the new way ... \
2599 You cannot portably "stack" cpp directives. For example in the above
2600 you need two separate BURGLE() #defines, one for each #ifdef branch.
2604 Adding stuff after #endif or #else
2612 The #endif and #else cannot portably have anything after them. If you
2613 want to document what is going (which is a good idea especially if the
2614 branches are long), use (C) comments:
2622 The gcc option C<-Wendif-labels> warns about the bad variant
2623 (by default on starting from Perl 5.9.4).
2627 Having a comma after the last element of an enum list
2632 CINNABAR, /* Right here. */
2635 is not portable. Leave out the last comma.
2637 Also note that whether enums are implicitly morphable to ints
2638 varies between compilers, you might need to (int).
2644 // This function bamfoodles the zorklator.
2646 That is C99 or C++. Perl is C89. Using the //-comments is silently
2647 allowed by many C compilers but cranking up the ANSI C89 strictness
2648 (which we like to do) causes the compilation to fail.
2652 Mixing declarations and code
2660 That is C99 or C++. Some C compilers allow that, but you shouldn't.
2662 The gcc option C<-Wdeclaration-after-statements> scans for such problems
2663 (by default on starting from Perl 5.9.4).
2667 Introducing variables inside for()
2669 for(int i = ...; ...; ...)
2671 That is C99 or C++. While it would indeed be awfully nice to have that
2672 also in C89, to limit the scope of the loop variable, alas, we cannot.
2676 Mixing signed char pointers with unsigned char pointers
2678 int foo(char *s) { ... }
2680 unsigned char *t = ...; /* Or U8* t = ... */
2683 While this is legal practice, it is certainly dubious, and downright
2684 fatal in at least one platform: for example VMS cc considers this a
2685 fatal error. One cause for people often making this mistake is that
2686 a "naked char" and therefore dereferencing a "naked char pointer" have
2687 an undefined signedness: it depends on the compiler and the platform
2688 whether the result is signed or unsigned. For this very same reason
2689 using a 'char' as an array index is bad.
2693 Macros that have string constants and their arguments as substrings of
2694 the string constants
2696 #define FOO(n) printf("number = %d\n", n)
2699 Pre-ANSI semantics for that was equivalent to
2701 printf("10umber = %d\10");
2703 which is probably not what you were expecting. Unfortunately at least
2704 one reasonably common and modern C compiler does "real backward
2705 compatibility" here, in AIX that is what still happens even though the
2706 rest of the AIX compiler is very happily C89.
2710 Using printf formats for non-basic C types
2713 printf("i = %d\n", i);
2715 While this might by accident work in some platform (where IV happens
2716 to be an C<int>), in general it cannot. IV might be something larger.
2717 Even worse the situation is with more specific types (defined by Perl's
2718 configuration step in F<config.h>):
2721 printf("who = %d\n", who);
2723 The problem here is that Uid_t might be not only not C<int>-wide
2724 but it might also be unsigned, in which case large uids would be
2725 printed as negative values.
2727 There is no simple solution to this because of printf()'s limited
2728 intelligence, but for many types the right format is available as
2729 with either 'f' or '_f' suffix, for example:
2731 IVdf /* IV in decimal */
2732 UVxf /* UV is hexadecimal */
2734 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2736 Uid_t_f /* Uid_t in decimal */
2738 printf("who = %"Uid_t_f"\n", who);
2740 Or you can try casting to a "wide enough" type:
2742 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2744 Also remember that the C<%p> format really does require a void pointer:
2747 printf("p = %p\n", (void*)p);
2749 The gcc option C<-Wformat> scans for such problems.
2753 Blindly using variadic macros
2755 gcc has had them for a while with its own syntax, and C99 brought
2756 them with a standardized syntax. Don't use the former, and use
2757 the latter only if the HAS_C99_VARIADIC_MACROS is defined.
2761 Blindly passing va_list
2763 Not all platforms support passing va_list to further varargs (stdarg)
2764 functions. The right thing to do is to copy the va_list using the
2765 Perl_va_copy() if the NEED_VA_COPY is defined.
2769 Using gcc brace groups
2771 val = ({...;...;...});
2773 While a nice extension, it's not portable.
2777 Binding together several statements
2779 Use the macros STMT_START and STMT_END.
2787 Testing for operating systems or versions when should be testing for features
2793 Unless you know with 100% certainty that quux() is only ever available
2794 for the "Foonix" operating system B<and> that is available B<and>
2795 correctly working for B<all> past, present, B<and> future versions of
2796 "Foonix", the above is very wrong. This is more correct (though still
2797 not perfect, because the below is a compile-time check):
2803 How does the HAS_QUUX become defined where it needs to be? Well, if
2804 Foonix happens to be UNIXy enought to be able to run the Configure
2805 script, and Configure has been taught about detecting and testing
2806 quux(), the HAS_QUUX will be correctly defined. In other platforms,
2807 the corresponding configuration step will hopefully do the same.
2809 In a pinch, if you cannot wait for Configure to be educated,
2810 or if you have a good hunch of where quux() might be available,
2811 you can temporarily try the following:
2813 #if (defined(__FOONIX__) || defined(__BARNIX__))
2823 But in any case, try to keep the features and operating systems separate.
2827 =head2 Security problems
2829 Last but not least, here are various tips for safer coding.
2837 Or we will publicly ridicule you. Seriously.
2841 Do not use strcpy() or strcat() or strncpy() or strncat()
2843 Use my_strlcpy() and my_strlcat() instead: they either use the native
2844 implementation, or Perl's own implementation (borrowed from the public
2845 domain implementation of INN).
2849 Do not use sprintf() or vsprintf()
2851 If you really want just plain byte strings, use my_snprintf()
2852 and my_vsnprintf() instead, which will try to use snprintf() and
2853 vsnprintf() if those safer APIs are available. If you want something
2854 fancier than a plain byte string, use SVs and Perl_sv_catpvf().
2858 =head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2860 Sometimes it helps to use external tools while debugging and
2861 testing Perl. This section tries to guide you through using
2862 some common testing and debugging tools with Perl. This is
2863 meant as a guide to interfacing these tools with Perl, not
2864 as any kind of guide to the use of the tools themselves.
2866 B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2867 Third Degree greatly slows down the execution: seconds become minutes,
2868 minutes become hours. For example as of Perl 5.8.1, the
2869 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2870 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2871 than six hours, even on a snappy computer-- the said test must be
2872 doing something that is quite unfriendly for memory debuggers. If you
2873 don't feel like waiting, that you can simply kill away the perl
2876 B<NOTE 2>: To minimize the number of memory leak false alarms (see
2877 L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2878 environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2879 and harness scripts do that automatically. But if you are running
2880 some of the tests manually-- for csh-like shells:
2882 setenv PERL_DESTRUCT_LEVEL 2
2884 and for Bourne-type shells:
2886 PERL_DESTRUCT_LEVEL=2
2887 export PERL_DESTRUCT_LEVEL
2889 or in UNIXy environments you can also use the C<env> command:
2891 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
2893 B<NOTE 3>: There are known memory leaks when there are compile-time
2894 errors within eval or require, seeing C<S_doeval> in the call stack
2895 is a good sign of these. Fixing these leaks is non-trivial,
2896 unfortunately, but they must be fixed eventually.
2898 =head2 Rational Software's Purify
2900 Purify is a commercial tool that is helpful in identifying
2901 memory overruns, wild pointers, memory leaks and other such
2902 badness. Perl must be compiled in a specific way for
2903 optimal testing with Purify. Purify is available under
2904 Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2906 =head2 Purify on Unix
2908 On Unix, Purify creates a new Perl binary. To get the most
2909 benefit out of Purify, you should create the perl to Purify
2912 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2913 -Uusemymalloc -Dusemultiplicity
2915 where these arguments mean:
2919 =item -Accflags=-DPURIFY
2921 Disables Perl's arena memory allocation functions, as well as
2922 forcing use of memory allocation functions derived from the
2925 =item -Doptimize='-g'
2927 Adds debugging information so that you see the exact source
2928 statements where the problem occurs. Without this flag, all
2929 you will see is the source filename of where the error occurred.
2933 Disable Perl's malloc so that Purify can more closely monitor
2934 allocations and leaks. Using Perl's malloc will make Purify
2935 report most leaks in the "potential" leaks category.
2937 =item -Dusemultiplicity
2939 Enabling the multiplicity option allows perl to clean up
2940 thoroughly when the interpreter shuts down, which reduces the
2941 number of bogus leak reports from Purify.
2945 Once you've compiled a perl suitable for Purify'ing, then you
2950 which creates a binary named 'pureperl' that has been Purify'ed.
2951 This binary is used in place of the standard 'perl' binary
2952 when you want to debug Perl memory problems.
2954 As an example, to show any memory leaks produced during the
2955 standard Perl testset you would create and run the Purify'ed
2960 ../pureperl -I../lib harness
2962 which would run Perl on test.pl and report any memory problems.
2964 Purify outputs messages in "Viewer" windows by default. If
2965 you don't have a windowing environment or if you simply
2966 want the Purify output to unobtrusively go to a log file
2967 instead of to the interactive window, use these following
2968 options to output to the log file "perl.log":
2970 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2971 -log-file=perl.log -append-logfile=yes"
2973 If you plan to use the "Viewer" windows, then you only need this option:
2975 setenv PURIFYOPTIONS "-chain-length=25"
2977 In Bourne-type shells:
2980 export PURIFYOPTIONS
2982 or if you have the "env" utility:
2984 env PURIFYOPTIONS="..." ../pureperl ...
2988 Purify on Windows NT instruments the Perl binary 'perl.exe'
2989 on the fly. There are several options in the makefile you
2990 should change to get the most use out of Purify:
2996 You should add -DPURIFY to the DEFINES line so the DEFINES
2997 line looks something like:
2999 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
3001 to disable Perl's arena memory allocation functions, as
3002 well as to force use of memory allocation functions derived
3003 from the system malloc.
3005 =item USE_MULTI = define
3007 Enabling the multiplicity option allows perl to clean up
3008 thoroughly when the interpreter shuts down, which reduces the
3009 number of bogus leak reports from Purify.
3011 =item #PERL_MALLOC = define
3013 Disable Perl's malloc so that Purify can more closely monitor
3014 allocations and leaks. Using Perl's malloc will make Purify
3015 report most leaks in the "potential" leaks category.
3019 Adds debugging information so that you see the exact source
3020 statements where the problem occurs. Without this flag, all
3021 you will see is the source filename of where the error occurred.
3025 As an example, to show any memory leaks produced during the
3026 standard Perl testset you would create and run Purify as:
3031 purify ../perl -I../lib harness
3033 which would instrument Perl in memory, run Perl on test.pl,
3034 then finally report any memory problems.
3038 The excellent valgrind tool can be used to find out both memory leaks
3039 and illegal memory accesses. As of August 2003 it unfortunately works
3040 only on x86 (ELF) Linux. The special "test.valgrind" target can be used
3041 to run the tests under valgrind. Found errors and memory leaks are
3042 logged in files named F<test.valgrind>.
3044 As system libraries (most notably glibc) are also triggering errors,
3045 valgrind allows to suppress such errors using suppression files. The
3046 default suppression file that comes with valgrind already catches a lot
3047 of them. Some additional suppressions are defined in F<t/perl.supp>.
3049 To get valgrind and for more information see
3051 http://developer.kde.org/~sewardj/
3053 =head2 Compaq's/Digital's/HP's Third Degree
3055 Third Degree is a tool for memory leak detection and memory access checks.
3056 It is one of the many tools in the ATOM toolkit. The toolkit is only
3057 available on Tru64 (formerly known as Digital UNIX formerly known as
3060 When building Perl, you must first run Configure with -Doptimize=-g
3061 and -Uusemymalloc flags, after that you can use the make targets
3062 "perl.third" and "test.third". (What is required is that Perl must be
3063 compiled using the C<-g> flag, you may need to re-Configure.)
3065 The short story is that with "atom" you can instrument the Perl
3066 executable to create a new executable called F<perl.third>. When the
3067 instrumented executable is run, it creates a log of dubious memory
3068 traffic in file called F<perl.3log>. See the manual pages of atom and
3069 third for more information. The most extensive Third Degree
3070 documentation is available in the Compaq "Tru64 UNIX Programmer's
3071 Guide", chapter "Debugging Programs with Third Degree".
3073 The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
3074 subdirectory. There is a problem with these files: Third Degree is so
3075 effective that it finds problems also in the system libraries.
3076 Therefore you should used the Porting/thirdclean script to cleanup
3077 the F<*.3log> files.
3079 There are also leaks that for given certain definition of a leak,
3080 aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
3082 =head2 PERL_DESTRUCT_LEVEL
3084 If you want to run any of the tests yourself manually using e.g.
3085 valgrind, or the pureperl or perl.third executables, please note that
3086 by default perl B<does not> explicitly cleanup all the memory it has
3087 allocated (such as global memory arenas) but instead lets the exit()
3088 of the whole program "take care" of such allocations, also known as
3089 "global destruction of objects".
3091 There is a way to tell perl to do complete cleanup: set the
3092 environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
3093 The t/TEST wrapper does set this to 2, and this is what you
3094 need to do too, if you don't want to see the "global leaks":
3095 For example, for "third-degreed" Perl:
3097 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
3099 (Note: the mod_perl apache module uses also this environment variable
3100 for its own purposes and extended its semantics. Refer to the mod_perl
3101 documentation for more information. Also, spawned threads do the
3102 equivalent of setting this variable to the value 1.)
3104 If, at the end of a run you get the message I<N scalars leaked>, you can
3105 recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
3106 of all those leaked SVs to be dumped along with details as to where each
3107 SV was originally allocated. This information is also displayed by
3108 Devel::Peek. Note that the extra details recorded with each SV increases
3109 memory usage, so it shouldn't be used in production environments. It also
3110 converts C<new_SV()> from a macro into a real function, so you can use
3111 your favourite debugger to discover where those pesky SVs were allocated.
3115 If compiled with C<-DPERL_MEM_LOG>, all Newx() and Renew() allocations
3116 and Safefree() in the Perl core go through logging functions, which is
3117 handy for breakpoint setting. If also compiled with C<-DPERL_MEM_LOG_STDERR>,
3118 the allocations and frees are logged to STDERR (or more precisely, to the
3119 file descriptor 2) in these logging functions, with the calling source code
3120 file and line number (and C function name, if supported by the C compiler).
3122 This logging is somewhat similar to C<-Dm> but independent of C<-DDEBUGGING>,
3123 and at a higher level (the C<-Dm> is directly at the point of C<malloc()>,
3124 while the C<PERL_MEM_LOG> is at the level of C<New()>).
3128 Depending on your platform there are various of profiling Perl.
3130 There are two commonly used techniques of profiling executables:
3131 I<statistical time-sampling> and I<basic-block counting>.
3133 The first method takes periodically samples of the CPU program
3134 counter, and since the program counter can be correlated with the code
3135 generated for functions, we get a statistical view of in which
3136 functions the program is spending its time. The caveats are that very
3137 small/fast functions have lower probability of showing up in the
3138 profile, and that periodically interrupting the program (this is
3139 usually done rather frequently, in the scale of milliseconds) imposes
3140 an additional overhead that may skew the results. The first problem
3141 can be alleviated by running the code for longer (in general this is a
3142 good idea for profiling), the second problem is usually kept in guard
3143 by the profiling tools themselves.
3145 The second method divides up the generated code into I<basic blocks>.
3146 Basic blocks are sections of code that are entered only in the
3147 beginning and exited only at the end. For example, a conditional jump
3148 starts a basic block. Basic block profiling usually works by
3149 I<instrumenting> the code by adding I<enter basic block #nnnn>
3150 book-keeping code to the generated code. During the execution of the
3151 code the basic block counters are then updated appropriately. The
3152 caveat is that the added extra code can skew the results: again, the
3153 profiling tools usually try to factor their own effects out of the
3156 =head2 Gprof Profiling
3158 gprof is a profiling tool available in many UNIX platforms,
3159 it uses F<statistical time-sampling>.
3161 You can build a profiled version of perl called "perl.gprof" by
3162 invoking the make target "perl.gprof" (What is required is that Perl
3163 must be compiled using the C<-pg> flag, you may need to re-Configure).
3164 Running the profiled version of Perl will create an output file called
3165 F<gmon.out> is created which contains the profiling data collected
3166 during the execution.
3168 The gprof tool can then display the collected data in various ways.
3169 Usually gprof understands the following options:
3175 Suppress statically defined functions from the profile.
3179 Suppress the verbose descriptions in the profile.
3183 Exclude the given routine and its descendants from the profile.
3187 Display only the given routine and its descendants in the profile.
3191 Generate a summary file called F<gmon.sum> which then may be given
3192 to subsequent gprof runs to accumulate data over several runs.
3196 Display routines that have zero usage.
3200 For more detailed explanation of the available commands and output
3201 formats, see your own local documentation of gprof.
3203 =head2 GCC gcov Profiling
3205 Starting from GCC 3.0 I<basic block profiling> is officially available
3208 You can build a profiled version of perl called F<perl.gcov> by
3209 invoking the make target "perl.gcov" (what is required that Perl must
3210 be compiled using gcc with the flags C<-fprofile-arcs
3211 -ftest-coverage>, you may need to re-Configure).
3213 Running the profiled version of Perl will cause profile output to be
3214 generated. For each source file an accompanying ".da" file will be
3217 To display the results you use the "gcov" utility (which should
3218 be installed if you have gcc 3.0 or newer installed). F<gcov> is
3219 run on source code files, like this
3223 which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3224 contain the source code annotated with relative frequencies of
3225 execution indicated by "#" markers.
3227 Useful options of F<gcov> include C<-b> which will summarise the
3228 basic block, branch, and function call coverage, and C<-c> which
3229 instead of relative frequencies will use the actual counts. For
3230 more information on the use of F<gcov> and basic block profiling
3231 with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3233 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3235 and its section titled "8. gcov: a Test Coverage Program"
3237 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3239 =head2 Pixie Profiling
3241 Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3242 UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
3243 I<basic-block counting>.
3245 You can build a profiled version of perl called F<perl.pixie> by
3246 invoking the make target "perl.pixie" (what is required is that Perl
3247 must be compiled using the C<-g> flag, you may need to re-Configure).
3249 In Tru64 a file called F<perl.Addrs> will also be silently created,
3250 this file contains the addresses of the basic blocks. Running the
3251 profiled version of Perl will create a new file called "perl.Counts"
3252 which contains the counts for the basic block for that particular
3255 To display the results you use the F<prof> utility. The exact
3256 incantation depends on your operating system, "prof perl.Counts" in
3257 IRIX, and "prof -pixie -all -L. perl" in Tru64.
3259 In IRIX the following prof options are available:
3265 Reports the most heavily used lines in descending order of use.
3266 Useful for finding the hotspot lines.
3270 Groups lines by procedure, with procedures sorted in descending order of use.
3271 Within a procedure, lines are listed in source order.
3272 Useful for finding the hotspots of procedures.
3276 In Tru64 the following options are available:
3282 Procedures sorted in descending order by the number of cycles executed
3283 in each procedure. Useful for finding the hotspot procedures.
3284 (This is the default option.)
3288 Lines sorted in descending order by the number of cycles executed in
3289 each line. Useful for finding the hotspot lines.
3291 =item -i[nvocations]
3293 The called procedures are sorted in descending order by number of calls
3294 made to the procedures. Useful for finding the most used procedures.
3298 Grouped by procedure, sorted by cycles executed per procedure.
3299 Useful for finding the hotspots of procedures.
3303 The compiler emitted code for these lines, but the code was unexecuted.
3307 Unexecuted procedures.
3311 For further information, see your system's manual pages for pixie and prof.
3313 =head2 Miscellaneous tricks
3319 Those debugging perl with the DDD frontend over gdb may find the
3322 You can extend the data conversion shortcuts menu, so for example you
3323 can display an SV's IV value with one click, without doing any typing.
3324 To do that simply edit ~/.ddd/init file and add after:
3326 ! Display shortcuts.
3327 Ddd*gdbDisplayShortcuts: \
3328 /t () // Convert to Bin\n\
3329 /d () // Convert to Dec\n\
3330 /x () // Convert to Hex\n\
3331 /o () // Convert to Oct(\n\
3333 the following two lines:
3335 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3336 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3338 so now you can do ivx and pvx lookups or you can plug there the
3339 sv_peek "conversion":
3341 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3343 (The my_perl is for threaded builds.)
3344 Just remember that every line, but the last one, should end with \n\
3346 Alternatively edit the init file interactively via:
3347 3rd mouse button -> New Display -> Edit Menu
3349 Note: you can define up to 20 conversion shortcuts in the gdb
3354 If you see in a debugger a memory area mysteriously full of 0xABABABAB
3355 or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3362 We've had a brief look around the Perl source, how to maintain quality
3363 of the source code, an overview of the stages F<perl> goes through
3364 when it's running your code, how to use debuggers to poke at the Perl
3365 guts, and finally how to analyse the execution of Perl. We took a very
3366 simple problem and demonstrated how to solve it fully - with
3367 documentation, regression tests, and finally a patch for submission to
3368 p5p. Finally, we talked about how to use external tools to debug and
3371 I'd now suggest you read over those references again, and then, as soon
3372 as possible, get your hands dirty. The best way to learn is by doing,
3379 Subscribe to perl5-porters, follow the patches and try and understand
3380 them; don't be afraid to ask if there's a portion you're not clear on -
3381 who knows, you may unearth a bug in the patch...
3385 Keep up to date with the bleeding edge Perl distributions and get
3386 familiar with the changes. Try and get an idea of what areas people are
3387 working on and the changes they're making.
3391 Do read the README associated with your operating system, e.g. README.aix
3392 on the IBM AIX OS. Don't hesitate to supply patches to that README if
3393 you find anything missing or changed over a new OS release.
3397 Find an area of Perl that seems interesting to you, and see if you can
3398 work out how it works. Scan through the source, and step over it in the
3399 debugger. Play, poke, investigate, fiddle! You'll probably get to
3400 understand not just your chosen area but a much wider range of F<perl>'s
3401 activity as well, and probably sooner than you'd think.
3407 =item I<The Road goes ever on and on, down from the door where it began.>
3411 If you can do these things, you've started on the long road to Perl porting.
3412 Thanks for wanting to help make Perl better - and happy hacking!
3416 This document was written by Nathan Torkington, and is maintained by
3417 the perl5-porters mailing list.