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 Hugo van der Sanden and Rafael Garcia-Suarez share the pumpking crown for
45 In addition, various people are pumpkings for different things. For
46 instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the
47 I<Configure> pumpkin up till the 5.8 release. For the 5.10 release
48 H.Merijn Brand took over.
50 Larry sees Perl development along the lines of the US government:
51 there's the Legislature (the porters), the Executive branch (the
52 pumpkings), and the Supreme Court (Larry). The legislature can
53 discuss and submit patches to the executive branch all they like, but
54 the executive branch is free to veto them. Rarely, the Supreme Court
55 will side with the executive branch over the legislature, or the
56 legislature over the executive branch. Mostly, however, the
57 legislature and the executive branch are supposed to get along and
58 work out their differences without impeachment or court cases.
60 You might sometimes see reference to Rule 1 and Rule 2. Larry's power
61 as Supreme Court is expressed in The Rules:
67 Larry is always by definition right about how Perl should behave.
68 This means he has final veto power on the core functionality.
72 Larry is allowed to change his mind about any matter at a later date,
73 regardless of whether he previously invoked Rule 1.
77 Got that? Larry is always right, even when he was wrong. It's rare
78 to see either Rule exercised, but they are often alluded to.
80 New features and extensions to the language are contentious, because
81 the criteria used by the pumpkings, Larry, and other porters to decide
82 which features should be implemented and incorporated are not codified
83 in a few small design goals as with some other languages. Instead,
84 the heuristics are flexible and often difficult to fathom. Here is
85 one person's list, roughly in decreasing order of importance, of
86 heuristics that new features have to be weighed against:
90 =item Does concept match the general goals of Perl?
92 These haven't been written anywhere in stone, but one approximation
95 1. Keep it fast, simple, and useful.
96 2. Keep features/concepts as orthogonal as possible.
97 3. No arbitrary limits (platforms, data sizes, cultures).
98 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
99 5. Either assimilate new technologies, or build bridges to them.
101 =item Where is the implementation?
103 All the talk in the world is useless without an implementation. In
104 almost every case, the person or people who argue for a new feature
105 will be expected to be the ones who implement it. Porters capable
106 of coding new features have their own agendas, and are not available
107 to implement your (possibly good) idea.
109 =item Backwards compatibility
111 It's a cardinal sin to break existing Perl programs. New warnings are
112 contentious--some say that a program that emits warnings is not
113 broken, while others say it is. Adding keywords has the potential to
114 break programs, changing the meaning of existing token sequences or
115 functions might break programs.
117 =item Could it be a module instead?
119 Perl 5 has extension mechanisms, modules and XS, specifically to avoid
120 the need to keep changing the Perl interpreter. You can write modules
121 that export functions, you can give those functions prototypes so they
122 can be called like built-in functions, you can even write XS code to
123 mess with the runtime data structures of the Perl interpreter if you
124 want to implement really complicated things. If it can be done in a
125 module instead of in the core, it's highly unlikely to be added.
127 =item Is the feature generic enough?
129 Is this something that only the submitter wants added to the language,
130 or would it be broadly useful? Sometimes, instead of adding a feature
131 with a tight focus, the porters might decide to wait until someone
132 implements the more generalized feature. For instance, instead of
133 implementing a "delayed evaluation" feature, the porters are waiting
134 for a macro system that would permit delayed evaluation and much more.
136 =item Does it potentially introduce new bugs?
138 Radical rewrites of large chunks of the Perl interpreter have the
139 potential to introduce new bugs. The smaller and more localized the
142 =item Does it preclude other desirable features?
144 A patch is likely to be rejected if it closes off future avenues of
145 development. For instance, a patch that placed a true and final
146 interpretation on prototypes is likely to be rejected because there
147 are still options for the future of prototypes that haven't been
150 =item Is the implementation robust?
152 Good patches (tight code, complete, correct) stand more chance of
153 going in. Sloppy or incorrect patches might be placed on the back
154 burner until the pumpking has time to fix, or might be discarded
155 altogether without further notice.
157 =item Is the implementation generic enough to be portable?
159 The worst patches make use of a system-specific features. It's highly
160 unlikely that nonportable additions to the Perl language will be
163 =item Is the implementation tested?
165 Patches which change behaviour (fixing bugs or introducing new features)
166 must include regression tests to verify that everything works as expected.
167 Without tests provided by the original author, how can anyone else changing
168 perl in the future be sure that they haven't unwittingly broken the behaviour
169 the patch implements? And without tests, how can the patch's author be
170 confident that his/her hard work put into the patch won't be accidentally
171 thrown away by someone in the future?
173 =item Is there enough documentation?
175 Patches without documentation are probably ill-thought out or
176 incomplete. Nothing can be added without documentation, so submitting
177 a patch for the appropriate manpages as well as the source code is
180 =item Is there another way to do it?
182 Larry said "Although the Perl Slogan is I<There's More Than One Way
183 to Do It>, I hesitate to make 10 ways to do something". This is a
184 tricky heuristic to navigate, though--one man's essential addition is
185 another man's pointless cruft.
187 =item Does it create too much work?
189 Work for the pumpking, work for Perl programmers, work for module
190 authors, ... Perl is supposed to be easy.
192 =item Patches speak louder than words
194 Working code is always preferred to pie-in-the-sky ideas. A patch to
195 add a feature stands a much higher chance of making it to the language
196 than does a random feature request, no matter how fervently argued the
197 request might be. This ties into "Will it be useful?", as the fact
198 that someone took the time to make the patch demonstrates a strong
199 desire for the feature.
203 If you're on the list, you might hear the word "core" bandied
204 around. It refers to the standard distribution. "Hacking on the
205 core" means you're changing the C source code to the Perl
206 interpreter. "A core module" is one that ships with Perl.
208 =head2 Keeping in sync
210 The source code to the Perl interpreter, in its different versions, is
211 kept in a repository managed by a revision control system ( which is
212 currently the Perforce program, see http://perforce.com/ ). The
213 pumpkings and a few others have access to the repository to check in
214 changes. Periodically the pumpking for the development version of Perl
215 will release a new version, so the rest of the porters can see what's
216 changed. The current state of the main trunk of repository, and patches
217 that describe the individual changes that have happened since the last
218 public release are available at this location:
220 http://public.activestate.com/gsar/APC/
221 ftp://ftp.linux.activestate.com/pub/staff/gsar/APC/
223 If you're looking for a particular change, or a change that affected
224 a particular set of files, you may find the B<Perl Repository Browser>
227 http://public.activestate.com/cgi-bin/perlbrowse
229 You may also want to subscribe to the perl5-changes mailing list to
230 receive a copy of each patch that gets submitted to the maintenance
231 and development "branches" of the perl repository. See
232 http://lists.perl.org/ for subscription information.
234 If you are a member of the perl5-porters mailing list, it is a good
235 thing to keep in touch with the most recent changes. If not only to
236 verify if what you would have posted as a bug report isn't already
237 solved in the most recent available perl development branch, also
238 known as perl-current, bleading edge perl, bleedperl or bleadperl.
240 Needless to say, the source code in perl-current is usually in a perpetual
241 state of evolution. You should expect it to be very buggy. Do B<not> use
242 it for any purpose other than testing and development.
244 Keeping in sync with the most recent branch can be done in several ways,
245 but the most convenient and reliable way is using B<rsync>, available at
246 ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
249 If you choose to keep in sync using rsync, there are two approaches
254 =item rsync'ing the source tree
256 Presuming you are in the directory where your perl source resides
257 and you have rsync installed and available, you can "upgrade" to
260 # rsync -avz rsync://ftp.linux.activestate.com/perl-current/ .
262 This takes care of updating every single item in the source tree to
263 the latest applied patch level, creating files that are new (to your
264 distribution) and setting date/time stamps of existing files to
265 reflect the bleadperl status.
267 Note that this will not delete any files that were in '.' before
268 the rsync. Once you are sure that the rsync is running correctly,
269 run it with the --delete and the --dry-run options like this:
271 # rsync -avz --delete --dry-run rsync://ftp.linux.activestate.com/perl-current/ .
273 This will I<simulate> an rsync run that also deletes files not
274 present in the bleadperl master copy. Observe the results from
275 this run closely. If you are sure that the actual run would delete
276 no files precious to you, you could remove the '--dry-run' option.
278 You can than check what patch was the latest that was applied by
279 looking in the file B<.patch>, which will show the number of the
282 If you have more than one machine to keep in sync, and not all of
283 them have access to the WAN (so you are not able to rsync all the
284 source trees to the real source), there are some ways to get around
289 =item Using rsync over the LAN
291 Set up a local rsync server which makes the rsynced source tree
292 available to the LAN and sync the other machines against this
295 From http://rsync.samba.org/README.html :
297 "Rsync uses rsh or ssh for communication. It does not need to be
298 setuid and requires no special privileges for installation. It
299 does not require an inetd entry or a daemon. You must, however,
300 have a working rsh or ssh system. Using ssh is recommended for
301 its security features."
303 =item Using pushing over the NFS
305 Having the other systems mounted over the NFS, you can take an
306 active pushing approach by checking the just updated tree against
307 the other not-yet synced trees. An example would be
316 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
319 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
321 foreach my $host (keys %remote) {
322 unless (-d $remote{$host}) {
323 print STDERR "Cannot Xsync for host $host\n";
326 foreach my $file (keys %MF) {
327 my $rfile = "$remote{$host}/$file";
328 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
329 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
330 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
331 printf "%4s %-34s %8d %9d %8d %9d\n",
332 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
334 copy ($file, $rfile);
335 utime time, $MF{$file}[2], $rfile;
336 chmod $MF{$file}[0], $rfile;
340 though this is not perfect. It could be improved with checking
341 file checksums before updating. Not all NFS systems support
342 reliable utime support (when used over the NFS).
346 =item rsync'ing the patches
348 The source tree is maintained by the pumpking who applies patches to
349 the files in the tree. These patches are either created by the
350 pumpking himself using C<diff -c> after updating the file manually or
351 by applying patches sent in by posters on the perl5-porters list.
352 These patches are also saved and rsync'able, so you can apply them
353 yourself to the source files.
355 Presuming you are in a directory where your patches reside, you can
356 get them in sync with
358 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
360 This makes sure the latest available patch is downloaded to your
363 It's then up to you to apply these patches, using something like
365 # last=`ls -t *.gz | sed q`
366 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
367 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
369 # patch -p1 -N <../perl-current-diffs/blead.patch
371 or, since this is only a hint towards how it works, use CPAN-patchaperl
372 from Andreas König to have better control over the patching process.
376 =head2 Why rsync the source tree
380 =item It's easier to rsync the source tree
382 Since you don't have to apply the patches yourself, you are sure all
383 files in the source tree are in the right state.
385 =item It's more reliable
387 While both the rsync-able source and patch areas are automatically
388 updated every few minutes, keep in mind that applying patches may
389 sometimes mean careful hand-holding, especially if your version of
390 the C<patch> program does not understand how to deal with new files,
391 files with 8-bit characters, or files without trailing newlines.
395 =head2 Why rsync the patches
399 =item It's easier to rsync the patches
401 If you have more than one machine that you want to keep in track with
402 bleadperl, it's easier to rsync the patches only once and then apply
403 them to all the source trees on the different machines.
405 In case you try to keep in pace on 5 different machines, for which
406 only one of them has access to the WAN, rsync'ing all the source
407 trees should than be done 5 times over the NFS. Having
408 rsync'ed the patches only once, I can apply them to all the source
409 trees automatically. Need you say more ;-)
411 =item It's a good reference
413 If you do not only like to have the most recent development branch,
414 but also like to B<fix> bugs, or extend features, you want to dive
415 into the sources. If you are a seasoned perl core diver, you don't
416 need no manuals, tips, roadmaps, perlguts.pod or other aids to find
417 your way around. But if you are a starter, the patches may help you
418 in finding where you should start and how to change the bits that
421 The file B<Changes> is updated on occasions the pumpking sees as his
422 own little sync points. On those occasions, he releases a tar-ball of
423 the current source tree (i.e. perl@7582.tar.gz), which will be an
424 excellent point to start with when choosing to use the 'rsync the
425 patches' scheme. Starting with perl@7582, which means a set of source
426 files on which the latest applied patch is number 7582, you apply all
427 succeeding patches available from then on (7583, 7584, ...).
429 You can use the patches later as a kind of search archive.
433 =item Finding a start point
435 If you want to fix/change the behaviour of function/feature Foo, just
436 scan the patches for patches that mention Foo either in the subject,
437 the comments, or the body of the fix. A good chance the patch shows
438 you the files that are affected by that patch which are very likely
439 to be the starting point of your journey into the guts of perl.
441 =item Finding how to fix a bug
443 If you've found I<where> the function/feature Foo misbehaves, but you
444 don't know how to fix it (but you do know the change you want to
445 make), you can, again, peruse the patches for similar changes and
446 look how others apply the fix.
448 =item Finding the source of misbehaviour
450 When you keep in sync with bleadperl, the pumpking would love to
451 I<see> that the community efforts really work. So after each of his
452 sync points, you are to 'make test' to check if everything is still
453 in working order. If it is, you do 'make ok', which will send an OK
454 report to perlbug@perl.org. (If you do not have access to a mailer
455 from the system you just finished successfully 'make test', you can
456 do 'make okfile', which creates the file C<perl.ok>, which you can
457 than take to your favourite mailer and mail yourself).
459 But of course, as always, things will not always lead to a success
460 path, and one or more test do not pass the 'make test'. Before
461 sending in a bug report (using 'make nok' or 'make nokfile'), check
462 the mailing list if someone else has reported the bug already and if
463 so, confirm it by replying to that message. If not, you might want to
464 trace the source of that misbehaviour B<before> sending in the bug,
465 which will help all the other porters in finding the solution.
467 Here the saved patches come in very handy. You can check the list of
468 patches to see which patch changed what file and what change caused
469 the misbehaviour. If you note that in the bug report, it saves the
470 one trying to solve it, looking for that point.
474 If searching the patches is too bothersome, you might consider using
475 perl's bugtron to find more information about discussions and
476 ramblings on posted bugs.
478 If you want to get the best of both worlds, rsync both the source
479 tree for convenience, reliability and ease and rsync the patches
484 =head2 Working with the source
486 Because you cannot use the Perforce client, you cannot easily generate
487 diffs against the repository, nor will merges occur when you update
488 via rsync. If you edit a file locally and then rsync against the
489 latest source, changes made in the remote copy will I<overwrite> your
492 The best way to deal with this is to maintain a tree of symlinks to
493 the rsync'd source. Then, when you want to edit a file, you remove
494 the symlink, copy the real file into the other tree, and edit it. You
495 can then diff your edited file against the original to generate a
496 patch, and you can safely update the original tree.
498 Perl's F<Configure> script can generate this tree of symlinks for you.
499 The following example assumes that you have used rsync to pull a copy
500 of the Perl source into the F<perl-rsync> directory. In the directory
501 above that one, you can execute the following commands:
505 ../perl-rsync/Configure -Dmksymlinks -Dusedevel -D"optimize=-g"
507 This will start the Perl configuration process. After a few prompts,
508 you should see something like this:
510 Symbolic links are supported.
512 Checking how to test for symbolic links...
513 Your builtin 'test -h' may be broken.
514 Trying external '/usr/bin/test -h'.
515 You can test for symbolic links with '/usr/bin/test -h'.
517 Creating the symbolic links...
518 (First creating the subdirectories...)
519 (Then creating the symlinks...)
521 The specifics may vary based on your operating system, of course.
522 After you see this, you can abort the F<Configure> script, and you
523 will see that the directory you are in has a tree of symlinks to the
524 F<perl-rsync> directories and files.
526 If you plan to do a lot of work with the Perl source, here are some
527 Bourne shell script functions that can make your life easier:
540 if [ -L $1.orig ]; then
546 Replace "vi" with your favorite flavor of editor.
548 Here is another function which will quickly generate a patch for the
549 files which have been edited in your symlink tree:
553 for f in `find . -name '*.orig' | sed s,^\./,,`
555 case `echo $f | sed 's,.orig$,,;s,.*\.,,'` in
557 pod) diffopts='-F^=' ;;
560 diff -du $diffopts $f `echo $f | sed 's,.orig$,,'`
564 This function produces patches which include enough context to make
565 your changes obvious. This makes it easier for the Perl pumpking(s)
566 to review them when you send them to the perl5-porters list, and that
567 means they're more likely to get applied.
569 This function assumed a GNU diff, and may require some tweaking for
572 =head2 Perlbug administration
574 There is a single remote administrative interface for modifying bug status,
575 category, open issues etc. using the B<RT> I<bugtracker> system, maintained
576 by I<Robert Spier>. Become an administrator, and close any bugs you can get
577 your sticky mitts on:
581 The bugtracker mechanism for B<perl5> bugs in particular is at:
583 http://bugs6.perl.org/perlbug
585 To email the bug system administrators:
587 "perlbug-admin" <perlbug-admin@perl.org>
590 =head2 Submitting patches
592 Always submit patches to I<perl5-porters@perl.org>. If you're
593 patching a core module and there's an author listed, send the author a
594 copy (see L<Patching a core module>). This lets other porters review
595 your patch, which catches a surprising number of errors in patches.
596 Either use the diff program (available in source code form from
597 ftp://ftp.gnu.org/pub/gnu/ , or use Johan Vromans' I<makepatch>
598 (available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
599 but context diffs are accepted. Do not send RCS-style diffs or diffs
600 without context lines. More information is given in the
601 I<Porting/patching.pod> file in the Perl source distribution. Please
602 patch against the latest B<development> version (e.g., if you're
603 fixing a bug in the 5.005 track, patch against the latest 5.005_5x
604 version). Only patches that survive the heat of the development
605 branch get applied to maintenance versions.
607 Your patch should update the documentation and test suite. See
610 To report a bug in Perl, use the program I<perlbug> which comes with
611 Perl (if you can't get Perl to work, send mail to the address
612 I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
613 I<perlbug> feeds into the automated bug-tracking system, access to
614 which is provided through the web at http://bugs.perl.org/ . It
615 often pays to check the archives of the perl5-porters mailing list to
616 see whether the bug you're reporting has been reported before, and if
617 so whether it was considered a bug. See above for the location of
618 the searchable archives.
620 The CPAN testers ( http://testers.cpan.org/ ) are a group of
621 volunteers who test CPAN modules on a variety of platforms. Perl
622 Smokers ( http://archives.develooper.com/daily-build@perl.org/ )
623 automatically tests Perl source releases on platforms with various
624 configurations. Both efforts welcome volunteers.
626 It's a good idea to read and lurk for a while before chipping in.
627 That way you'll get to see the dynamic of the conversations, learn the
628 personalities of the players, and hopefully be better prepared to make
629 a useful contribution when do you speak up.
631 If after all this you still think you want to join the perl5-porters
632 mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
633 unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
635 To hack on the Perl guts, you'll need to read the following things:
641 This is of paramount importance, since it's the documentation of what
642 goes where in the Perl source. Read it over a couple of times and it
643 might start to make sense - don't worry if it doesn't yet, because the
644 best way to study it is to read it in conjunction with poking at Perl
645 source, and we'll do that later on.
647 You might also want to look at Gisle Aas's illustrated perlguts -
648 there's no guarantee that this will be absolutely up-to-date with the
649 latest documentation in the Perl core, but the fundamentals will be
650 right. ( http://gisle.aas.no/perl/illguts/ )
652 =item L<perlxstut> and L<perlxs>
654 A working knowledge of XSUB programming is incredibly useful for core
655 hacking; XSUBs use techniques drawn from the PP code, the portion of the
656 guts that actually executes a Perl program. It's a lot gentler to learn
657 those techniques from simple examples and explanation than from the core
662 The documentation for the Perl API explains what some of the internal
663 functions do, as well as the many macros used in the source.
665 =item F<Porting/pumpkin.pod>
667 This is a collection of words of wisdom for a Perl porter; some of it is
668 only useful to the pumpkin holder, but most of it applies to anyone
669 wanting to go about Perl development.
671 =item The perl5-porters FAQ
673 This should be available from http://simon-cozens.org/writings/p5p-faq ;
674 alternatively, you can get the FAQ emailed to you by sending mail to
675 C<perl5-porters-faq@perl.org>. It contains hints on reading perl5-porters,
676 information on how perl5-porters works and how Perl development in general
681 =head2 Finding Your Way Around
683 Perl maintenance can be split into a number of areas, and certain people
684 (pumpkins) will have responsibility for each area. These areas sometimes
685 correspond to files or directories in the source kit. Among the areas are:
691 Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
692 subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
693 contains the core XS modules.
697 There are tests for nearly all the modules, built-ins and major bits
698 of functionality. Test files all have a .t suffix. Module tests live
699 in the F<lib/> and F<ext/> directories next to the module being
700 tested. Others live in F<t/>. See L<Writing a test>
704 Documentation maintenance includes looking after everything in the
705 F<pod/> directory, (as well as contributing new documentation) and
706 the documentation to the modules in core.
710 The configure process is the way we make Perl portable across the
711 myriad of operating systems it supports. Responsibility for the
712 configure, build and installation process, as well as the overall
713 portability of the core code rests with the configure pumpkin - others
714 help out with individual operating systems.
716 The files involved are the operating system directories, (F<win32/>,
717 F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
718 and F<Makefile>, as well as the metaconfig files which generate
719 F<Configure>. (metaconfig isn't included in the core distribution.)
723 And of course, there's the core of the Perl interpreter itself. Let's
724 have a look at that in a little more detail.
728 Before we leave looking at the layout, though, don't forget that
729 F<MANIFEST> contains not only the file names in the Perl distribution,
730 but short descriptions of what's in them, too. For an overview of the
731 important files, try this:
733 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
735 =head2 Elements of the interpreter
737 The work of the interpreter has two main stages: compiling the code
738 into the internal representation, or bytecode, and then executing it.
739 L<perlguts/Compiled code> explains exactly how the compilation stage
742 Here is a short breakdown of perl's operation:
748 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
749 This is very high-level code, enough to fit on a single screen, and it
750 resembles the code found in L<perlembed>; most of the real action takes
753 First, F<perlmain.c> allocates some memory and constructs a Perl
756 1 PERL_SYS_INIT3(&argc,&argv,&env);
758 3 if (!PL_do_undump) {
759 4 my_perl = perl_alloc();
762 7 perl_construct(my_perl);
763 8 PL_perl_destruct_level = 0;
766 Line 1 is a macro, and its definition is dependent on your operating
767 system. Line 3 references C<PL_do_undump>, a global variable - all
768 global variables in Perl start with C<PL_>. This tells you whether the
769 current running program was created with the C<-u> flag to perl and then
770 F<undump>, which means it's going to be false in any sane context.
772 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
773 interpreter. It's quite a simple function, and the guts of it looks like
776 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
778 Here you see an example of Perl's system abstraction, which we'll see
779 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
780 own C<malloc> as defined in F<malloc.c> if you selected that option at
783 Next, in line 7, we construct the interpreter; this sets up all the
784 special variables that Perl needs, the stacks, and so on.
786 Now we pass Perl the command line options, and tell it to go:
788 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
790 exitstatus = perl_run(my_perl);
794 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
795 in F<perl.c>, which processes the command line options, sets up any
796 statically linked XS modules, opens the program and calls C<yyparse> to
801 The aim of this stage is to take the Perl source, and turn it into an op
802 tree. We'll see what one of those looks like later. Strictly speaking,
803 there's three things going on here.
805 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
806 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
807 B<is> a YACC grammar for Perl!) The job of the parser is to take your
808 code and "understand" it, splitting it into sentences, deciding which
809 operands go with which operators and so on.
811 The parser is nobly assisted by the lexer, which chunks up your input
812 into tokens, and decides what type of thing each token is: a variable
813 name, an operator, a bareword, a subroutine, a core function, and so on.
814 The main point of entry to the lexer is C<yylex>, and that and its
815 associated routines can be found in F<toke.c>. Perl isn't much like
816 other computer languages; it's highly context sensitive at times, it can
817 be tricky to work out what sort of token something is, or where a token
818 ends. As such, there's a lot of interplay between the tokeniser and the
819 parser, which can get pretty frightening if you're not used to it.
821 As the parser understands a Perl program, it builds up a tree of
822 operations for the interpreter to perform during execution. The routines
823 which construct and link together the various operations are to be found
824 in F<op.c>, and will be examined later.
828 Now the parsing stage is complete, and the finished tree represents
829 the operations that the Perl interpreter needs to perform to execute our
830 program. Next, Perl does a dry run over the tree looking for
831 optimisations: constant expressions such as C<3 + 4> will be computed
832 now, and the optimizer will also see if any multiple operations can be
833 replaced with a single one. For instance, to fetch the variable C<$foo>,
834 instead of grabbing the glob C<*foo> and looking at the scalar
835 component, the optimizer fiddles the op tree to use a function which
836 directly looks up the scalar in question. The main optimizer is C<peep>
837 in F<op.c>, and many ops have their own optimizing functions.
841 Now we're finally ready to go: we have compiled Perl byte code, and all
842 that's left to do is run it. The actual execution is done by the
843 C<runops_standard> function in F<run.c>; more specifically, it's done by
844 these three innocent looking lines:
846 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
850 You may be more comfortable with the Perl version of that:
852 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
854 Well, maybe not. Anyway, each op contains a function pointer, which
855 stipulates the function which will actually carry out the operation.
856 This function will return the next op in the sequence - this allows for
857 things like C<if> which choose the next op dynamically at run time.
858 The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
859 execution if required.
861 The actual functions called are known as PP code, and they're spread
862 between four files: F<pp_hot.c> contains the "hot" code, which is most
863 often used and highly optimized, F<pp_sys.c> contains all the
864 system-specific functions, F<pp_ctl.c> contains the functions which
865 implement control structures (C<if>, C<while> and the like) and F<pp.c>
866 contains everything else. These are, if you like, the C code for Perl's
867 built-in functions and operators.
869 Note that each C<pp_> function is expected to return a pointer to the next
870 op. Calls to perl subs (and eval blocks) are handled within the same
871 runops loop, and do not consume extra space on the C stack. For example,
872 C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
873 struct onto the context stack which contain the address of the op
874 following the sub call or eval. They then return the first op of that sub
875 or eval block, and so execution continues of that sub or block. Later, a
876 C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
877 retrieves the return op from it, and returns it.
879 =item Exception handing
881 Perl's exception handing (ie C<die> etc) is built on top of the low-level
882 C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
883 way to capture the current PC and SP registers and later restore them; ie
884 a C<longjmp()> continues at the point in code where a previous C<setjmp()>
885 was done, with anything further up on the C stack being lost. This is why
886 code should always save values using C<SAVE_FOO> rather than in auto
889 The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
890 C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
891 C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
892 C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
894 At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
895 C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
896 loop or whatever, and handle possible exception returns. For a 2 return,
897 final cleanup is performed, such as popping stacks and calling C<CHECK> or
898 C<END> blocks. Amongst other things, this is how scope cleanup still
899 occurs during an C<exit>.
901 If a C<die> can find a C<CxEVAL> block on the context stack, then the
902 stack is popped to that level and the return op in that block is assigned
903 to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
904 passes control back to the guard. In the case of C<perl_run> and
905 C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
906 loop. The is the normal way that C<die> or C<croak> is handled within an
909 Sometimes ops are executed within an inner runops loop, such as tie, sort
910 or overload code. In this case, something like
912 sub FETCH { eval { die } }
914 would cause a longjmp right back to the guard in C<perl_run>, popping both
915 runops loops, which is clearly incorrect. One way to avoid this is for the
916 tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
917 runops loop, but for efficiency reasons, perl in fact just sets a flag,
918 using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
919 C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
920 which does a C<JMPENV_PUSH> and starts a new runops level to execute the
921 code, rather than doing it on the current loop.
923 As a further optimisation, on exit from the eval block in the C<FETCH>,
924 execution of the code following the block is still carried on in the inner
925 loop. When an exception is raised, C<docatch> compares the C<JMPENV>
926 level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
927 re-throws the exception. In this way any inner loops get popped.
931 1: eval { tie @a, 'A' };
937 To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
938 enters a runops loop. This loop executes the eval and tie ops on line 1,
939 with the eval pushing a C<CxEVAL> onto the context stack.
941 The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
942 to execute the body of C<TIEARRAY>. When it executes the entertry op on
943 line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
944 does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
945 the die op. At this point the C call stack looks like this:
948 Perl_runops # third loop
952 Perl_runops # second loop
956 Perl_runops # first loop
961 and the context and data stacks, as shown by C<-Dstv>, look like:
965 CX 1: EVAL => AV() PV("A"\0)
973 The die pops the first C<CxEVAL> off the context stack, sets
974 C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
975 the top C<docatch>. This then starts another third-level runops level,
976 which executes the nextstate, pushmark and die ops on line 4. At the point
977 that the second C<pp_die> is called, the C call stack looks exactly like
978 that above, even though we are no longer within an inner eval; this is
979 because of the optimization mentioned earlier. However, the context stack
980 now looks like this, ie with the top CxEVAL popped:
984 CX 1: EVAL => AV() PV("A"\0)
990 The die on line 4 pops the context stack back down to the CxEVAL, leaving
996 As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
997 C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
1001 Perl_runops # second loop
1005 Perl_runops # first loop
1010 In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
1011 differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
1012 and the C stack unwinds to:
1017 Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
1018 and execution continues.
1022 =head2 Internal Variable Types
1024 You should by now have had a look at L<perlguts>, which tells you about
1025 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
1028 These variables are used not only to represent Perl-space variables, but
1029 also any constants in the code, as well as some structures completely
1030 internal to Perl. The symbol table, for instance, is an ordinary Perl
1031 hash. Your code is represented by an SV as it's read into the parser;
1032 any program files you call are opened via ordinary Perl filehandles, and
1035 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
1036 Perl program. Let's see, for instance, how Perl treats the constant
1039 % perl -MDevel::Peek -e 'Dump("hello")'
1040 1 SV = PV(0xa041450) at 0xa04ecbc
1042 3 FLAGS = (POK,READONLY,pPOK)
1043 4 PV = 0xa0484e0 "hello"\0
1047 Reading C<Devel::Peek> output takes a bit of practise, so let's go
1048 through it line by line.
1050 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
1051 memory. SVs themselves are very simple structures, but they contain a
1052 pointer to a more complex structure. In this case, it's a PV, a
1053 structure which holds a string value, at location C<0xa041450>. Line 2
1054 is the reference count; there are no other references to this data, so
1057 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
1058 read-only SV (because it's a constant) and the data is a PV internally.
1059 Next we've got the contents of the string, starting at location
1062 Line 5 gives us the current length of the string - note that this does
1063 B<not> include the null terminator. Line 6 is not the length of the
1064 string, but the length of the currently allocated buffer; as the string
1065 grows, Perl automatically extends the available storage via a routine
1068 You can get at any of these quantities from C very easily; just add
1069 C<Sv> to the name of the field shown in the snippet, and you've got a
1070 macro which will return the value: C<SvCUR(sv)> returns the current
1071 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
1072 C<SvPV(sv, len)> returns the string itself with its length, and so on.
1073 More macros to manipulate these properties can be found in L<perlguts>.
1075 Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
1078 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
1083 6 junk = SvPV_force(sv, tlen);
1084 7 SvGROW(sv, tlen + len + 1);
1087 10 Move(ptr,SvPVX(sv)+tlen,len,char);
1088 11 SvCUR(sv) += len;
1089 12 *SvEND(sv) = '\0';
1090 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
1094 This is a function which adds a string, C<ptr>, of length C<len> onto
1095 the end of the PV stored in C<sv>. The first thing we do in line 6 is
1096 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
1097 macro to force a PV. As a side effect, C<tlen> gets set to the current
1098 value of the PV, and the PV itself is returned to C<junk>.
1100 In line 7, we make sure that the SV will have enough room to accommodate
1101 the old string, the new string and the null terminator. If C<LEN> isn't
1102 big enough, C<SvGROW> will reallocate space for us.
1104 Now, if C<junk> is the same as the string we're trying to add, we can
1105 grab the string directly from the SV; C<SvPVX> is the address of the PV
1108 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
1109 memory around: we move the string C<ptr> to the end of the PV - that's
1110 the start of the PV plus its current length. We're moving C<len> bytes
1111 of type C<char>. After doing so, we need to tell Perl we've extended the
1112 string, by altering C<CUR> to reflect the new length. C<SvEND> is a
1113 macro which gives us the end of the string, so that needs to be a
1116 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
1117 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1118 want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
1119 version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
1120 and turns on POK. The final C<SvTAINT> is a macro which launders tainted
1121 data if taint mode is turned on.
1123 AVs and HVs are more complicated, but SVs are by far the most common
1124 variable type being thrown around. Having seen something of how we
1125 manipulate these, let's go on and look at how the op tree is
1130 First, what is the op tree, anyway? The op tree is the parsed
1131 representation of your program, as we saw in our section on parsing, and
1132 it's the sequence of operations that Perl goes through to execute your
1133 program, as we saw in L</Running>.
1135 An op is a fundamental operation that Perl can perform: all the built-in
1136 functions and operators are ops, and there are a series of ops which
1137 deal with concepts the interpreter needs internally - entering and
1138 leaving a block, ending a statement, fetching a variable, and so on.
1140 The op tree is connected in two ways: you can imagine that there are two
1141 "routes" through it, two orders in which you can traverse the tree.
1142 First, parse order reflects how the parser understood the code, and
1143 secondly, execution order tells perl what order to perform the
1146 The easiest way to examine the op tree is to stop Perl after it has
1147 finished parsing, and get it to dump out the tree. This is exactly what
1148 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
1149 and L<B::Debug|B::Debug> do.
1151 Let's have a look at how Perl sees C<$a = $b + $c>:
1153 % perl -MO=Terse -e '$a=$b+$c'
1154 1 LISTOP (0x8179888) leave
1155 2 OP (0x81798b0) enter
1156 3 COP (0x8179850) nextstate
1157 4 BINOP (0x8179828) sassign
1158 5 BINOP (0x8179800) add [1]
1159 6 UNOP (0x81796e0) null [15]
1160 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
1161 8 UNOP (0x81797e0) null [15]
1162 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
1163 10 UNOP (0x816b4f0) null [15]
1164 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
1166 Let's start in the middle, at line 4. This is a BINOP, a binary
1167 operator, which is at location C<0x8179828>. The specific operator in
1168 question is C<sassign> - scalar assignment - and you can find the code
1169 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
1170 binary operator, it has two children: the add operator, providing the
1171 result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
1174 Line 10 is the null op: this does exactly nothing. What is that doing
1175 there? If you see the null op, it's a sign that something has been
1176 optimized away after parsing. As we mentioned in L</Optimization>,
1177 the optimization stage sometimes converts two operations into one, for
1178 example when fetching a scalar variable. When this happens, instead of
1179 rewriting the op tree and cleaning up the dangling pointers, it's easier
1180 just to replace the redundant operation with the null op. Originally,
1181 the tree would have looked like this:
1183 10 SVOP (0x816b4f0) rv2sv [15]
1184 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
1186 That is, fetch the C<a> entry from the main symbol table, and then look
1187 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
1188 happens to do both these things.
1190 The right hand side, starting at line 5 is similar to what we've just
1191 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
1194 Now, what's this about?
1196 1 LISTOP (0x8179888) leave
1197 2 OP (0x81798b0) enter
1198 3 COP (0x8179850) nextstate
1200 C<enter> and C<leave> are scoping ops, and their job is to perform any
1201 housekeeping every time you enter and leave a block: lexical variables
1202 are tidied up, unreferenced variables are destroyed, and so on. Every
1203 program will have those first three lines: C<leave> is a list, and its
1204 children are all the statements in the block. Statements are delimited
1205 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
1206 the ops to be performed for each statement being the children of
1207 C<nextstate>. C<enter> is a single op which functions as a marker.
1209 That's how Perl parsed the program, from top to bottom:
1222 However, it's impossible to B<perform> the operations in this order:
1223 you have to find the values of C<$b> and C<$c> before you add them
1224 together, for instance. So, the other thread that runs through the op
1225 tree is the execution order: each op has a field C<op_next> which points
1226 to the next op to be run, so following these pointers tells us how perl
1227 executes the code. We can traverse the tree in this order using
1228 the C<exec> option to C<B::Terse>:
1230 % perl -MO=Terse,exec -e '$a=$b+$c'
1231 1 OP (0x8179928) enter
1232 2 COP (0x81798c8) nextstate
1233 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
1234 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
1235 5 BINOP (0x8179878) add [1]
1236 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
1237 7 BINOP (0x81798a0) sassign
1238 8 LISTOP (0x8179900) leave
1240 This probably makes more sense for a human: enter a block, start a
1241 statement. Get the values of C<$b> and C<$c>, and add them together.
1242 Find C<$a>, and assign one to the other. Then leave.
1244 The way Perl builds up these op trees in the parsing process can be
1245 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1246 piece we need to construct the tree for C<$a = $b + $c>
1248 1 term : term ASSIGNOP term
1249 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1251 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1253 If you're not used to reading BNF grammars, this is how it works: You're
1254 fed certain things by the tokeniser, which generally end up in upper
1255 case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1256 code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
1257 "terminal symbols", because you can't get any simpler than them.
1259 The grammar, lines one and three of the snippet above, tells you how to
1260 build up more complex forms. These complex forms, "non-terminal symbols"
1261 are generally placed in lower case. C<term> here is a non-terminal
1262 symbol, representing a single expression.
1264 The grammar gives you the following rule: you can make the thing on the
1265 left of the colon if you see all the things on the right in sequence.
1266 This is called a "reduction", and the aim of parsing is to completely
1267 reduce the input. There are several different ways you can perform a
1268 reduction, separated by vertical bars: so, C<term> followed by C<=>
1269 followed by C<term> makes a C<term>, and C<term> followed by C<+>
1270 followed by C<term> can also make a C<term>.
1272 So, if you see two terms with an C<=> or C<+>, between them, you can
1273 turn them into a single expression. When you do this, you execute the
1274 code in the block on the next line: if you see C<=>, you'll do the code
1275 in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1276 which contributes to the op tree.
1279 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1281 What this does is creates a new binary op, and feeds it a number of
1282 variables. The variables refer to the tokens: C<$1> is the first token in
1283 the input, C<$2> the second, and so on - think regular expression
1284 backreferences. C<$$> is the op returned from this reduction. So, we
1285 call C<newBINOP> to create a new binary operator. The first parameter to
1286 C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1287 operator, so we want the type to be C<ADDOP>. We could specify this
1288 directly, but it's right there as the second token in the input, so we
1289 use C<$2>. The second parameter is the op's flags: 0 means "nothing
1290 special". Then the things to add: the left and right hand side of our
1291 expression, in scalar context.
1295 When perl executes something like C<addop>, how does it pass on its
1296 results to the next op? The answer is, through the use of stacks. Perl
1297 has a number of stacks to store things it's currently working on, and
1298 we'll look at the three most important ones here.
1302 =item Argument stack
1304 Arguments are passed to PP code and returned from PP code using the
1305 argument stack, C<ST>. The typical way to handle arguments is to pop
1306 them off the stack, deal with them how you wish, and then push the result
1307 back onto the stack. This is how, for instance, the cosine operator
1312 value = Perl_cos(value);
1315 We'll see a more tricky example of this when we consider Perl's macros
1316 below. C<POPn> gives you the NV (floating point value) of the top SV on
1317 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1318 the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1319 should be extended if necessary - it can't be necessary here, because we
1320 know there's room for one more item on the stack, since we've just
1321 removed one! The C<XPUSH*> macros at least guarantee safety.
1323 Alternatively, you can fiddle with the stack directly: C<SP> gives you
1324 the first element in your portion of the stack, and C<TOP*> gives you
1325 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1326 negation of an integer:
1330 Just set the integer value of the top stack entry to its negation.
1332 Argument stack manipulation in the core is exactly the same as it is in
1333 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1334 description of the macros used in stack manipulation.
1338 I say "your portion of the stack" above because PP code doesn't
1339 necessarily get the whole stack to itself: if your function calls
1340 another function, you'll only want to expose the arguments aimed for the
1341 called function, and not (necessarily) let it get at your own data. The
1342 way we do this is to have a "virtual" bottom-of-stack, exposed to each
1343 function. The mark stack keeps bookmarks to locations in the argument
1344 stack usable by each function. For instance, when dealing with a tied
1345 variable, (internally, something with "P" magic) Perl has to call
1346 methods for accesses to the tied variables. However, we need to separate
1347 the arguments exposed to the method to the argument exposed to the
1348 original function - the store or fetch or whatever it may be. Here's how
1349 the tied C<push> is implemented; see C<av_push> in F<av.c>:
1353 3 PUSHs(SvTIED_obj((SV*)av, mg));
1357 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1361 The lines which concern the mark stack are the first, fifth and last
1362 lines: they save away, restore and remove the current position of the
1365 Let's examine the whole implementation, for practice:
1369 Push the current state of the stack pointer onto the mark stack. This is
1370 so that when we've finished adding items to the argument stack, Perl
1371 knows how many things we've added recently.
1374 3 PUSHs(SvTIED_obj((SV*)av, mg));
1377 We're going to add two more items onto the argument stack: when you have
1378 a tied array, the C<PUSH> subroutine receives the object and the value
1379 to be pushed, and that's exactly what we have here - the tied object,
1380 retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1384 Next we tell Perl to make the change to the global stack pointer: C<dSP>
1385 only gave us a local copy, not a reference to the global.
1388 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1391 C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1392 variables are tidied up, everything that has been localised gets
1393 its previous value returned, and so on. Think of them as the C<{> and
1394 C<}> of a Perl block.
1396 To actually do the magic method call, we have to call a subroutine in
1397 Perl space: C<call_method> takes care of that, and it's described in
1398 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1399 going to discard its return value.
1403 Finally, we remove the value we placed on the mark stack, since we
1404 don't need it any more.
1408 C doesn't have a concept of local scope, so perl provides one. We've
1409 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1410 stack implements the C equivalent of, for example:
1417 See L<perlguts/Localising Changes> for how to use the save stack.
1421 =head2 Millions of Macros
1423 One thing you'll notice about the Perl source is that it's full of
1424 macros. Some have called the pervasive use of macros the hardest thing
1425 to understand, others find it adds to clarity. Let's take an example,
1426 the code which implements the addition operator:
1430 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1433 6 SETn( left + right );
1438 Every line here (apart from the braces, of course) contains a macro. The
1439 first line sets up the function declaration as Perl expects for PP code;
1440 line 3 sets up variable declarations for the argument stack and the
1441 target, the return value of the operation. Finally, it tries to see if
1442 the addition operation is overloaded; if so, the appropriate subroutine
1445 Line 5 is another variable declaration - all variable declarations start
1446 with C<d> - which pops from the top of the argument stack two NVs (hence
1447 C<nn>) and puts them into the variables C<right> and C<left>, hence the
1448 C<rl>. These are the two operands to the addition operator. Next, we
1449 call C<SETn> to set the NV of the return value to the result of adding
1450 the two values. This done, we return - the C<RETURN> macro makes sure
1451 that our return value is properly handled, and we pass the next operator
1452 to run back to the main run loop.
1454 Most of these macros are explained in L<perlapi>, and some of the more
1455 important ones are explained in L<perlxs> as well. Pay special attention
1456 to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1457 the C<[pad]THX_?> macros.
1459 =head2 The .i Targets
1461 You can expand the macros in a F<foo.c> file by saying
1465 which will expand the macros using cpp. Don't be scared by the results.
1467 =head2 Poking at Perl
1469 To really poke around with Perl, you'll probably want to build Perl for
1470 debugging, like this:
1472 ./Configure -d -D optimize=-g
1475 C<-g> is a flag to the C compiler to have it produce debugging
1476 information which will allow us to step through a running program.
1477 F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1478 enables all the internal debugging code in Perl. There are a whole bunch
1479 of things you can debug with this: L<perlrun> lists them all, and the
1480 best way to find out about them is to play about with them. The most
1481 useful options are probably
1483 l Context (loop) stack processing
1485 o Method and overloading resolution
1486 c String/numeric conversions
1488 Some of the functionality of the debugging code can be achieved using XS
1491 -Dr => use re 'debug'
1492 -Dx => use O 'Debug'
1494 =head2 Using a source-level debugger
1496 If the debugging output of C<-D> doesn't help you, it's time to step
1497 through perl's execution with a source-level debugger.
1503 We'll use C<gdb> for our examples here; the principles will apply to any
1504 debugger, but check the manual of the one you're using.
1508 To fire up the debugger, type
1512 You'll want to do that in your Perl source tree so the debugger can read
1513 the source code. You should see the copyright message, followed by the
1518 C<help> will get you into the documentation, but here are the most
1525 Run the program with the given arguments.
1527 =item break function_name
1529 =item break source.c:xxx
1531 Tells the debugger that we'll want to pause execution when we reach
1532 either the named function (but see L<perlguts/Internal Functions>!) or the given
1533 line in the named source file.
1537 Steps through the program a line at a time.
1541 Steps through the program a line at a time, without descending into
1546 Run until the next breakpoint.
1550 Run until the end of the current function, then stop again.
1554 Just pressing Enter will do the most recent operation again - it's a
1555 blessing when stepping through miles of source code.
1559 Execute the given C code and print its results. B<WARNING>: Perl makes
1560 heavy use of macros, and F<gdb> does not necessarily support macros
1561 (see later L</"gdb macro support">). You'll have to substitute them
1562 yourself, or to invoke cpp on the source code files
1563 (see L</"The .i Targets">)
1564 So, for instance, you can't say
1566 print SvPV_nolen(sv)
1570 print Perl_sv_2pv_nolen(sv)
1574 You may find it helpful to have a "macro dictionary", which you can
1575 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1576 recursively apply those macros for you.
1578 =head2 gdb macro support
1580 Recent versions of F<gdb> have fairly good macro support, but
1581 in order to use it you'll need to compile perl with macro definitions
1582 included in the debugging information. Using F<gcc> version 3.1, this
1583 means configuring with C<-Doptimize=-g3>. Other compilers might use a
1584 different switch (if they support debugging macros at all).
1586 =head2 Dumping Perl Data Structures
1588 One way to get around this macro hell is to use the dumping functions in
1589 F<dump.c>; these work a little like an internal
1590 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1591 that you can't get at from Perl. Let's take an example. We'll use the
1592 C<$a = $b + $c> we used before, but give it a bit of context:
1593 C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1595 What about C<pp_add>, the function we examined earlier to implement the
1598 (gdb) break Perl_pp_add
1599 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1601 Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
1602 With the breakpoint in place, we can run our program:
1604 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1606 Lots of junk will go past as gdb reads in the relevant source files and
1607 libraries, and then:
1609 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
1610 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1615 We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1616 arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1619 #define dPOPTOPnnrl_ul NV right = POPn; \
1620 SV *leftsv = TOPs; \
1621 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1623 C<POPn> takes the SV from the top of the stack and obtains its NV either
1624 directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1625 C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1626 C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1627 C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1629 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1630 convert it. If we step again, we'll find ourselves there:
1632 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1636 We can now use C<Perl_sv_dump> to investigate the SV:
1638 SV = PV(0xa057cc0) at 0xa0675d0
1641 PV = 0xa06a510 "6XXXX"\0
1646 We know we're going to get C<6> from this, so let's finish the
1650 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1651 0x462669 in Perl_pp_add () at pp_hot.c:311
1654 We can also dump out this op: the current op is always stored in
1655 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1656 similar output to L<B::Debug|B::Debug>.
1659 13 TYPE = add ===> 14
1661 FLAGS = (SCALAR,KIDS)
1663 TYPE = null ===> (12)
1665 FLAGS = (SCALAR,KIDS)
1667 11 TYPE = gvsv ===> 12
1673 # finish this later #
1677 All right, we've now had a look at how to navigate the Perl sources and
1678 some things you'll need to know when fiddling with them. Let's now get
1679 on and create a simple patch. Here's something Larry suggested: if a
1680 C<U> is the first active format during a C<pack>, (for example,
1681 C<pack "U3C8", @stuff>) then the resulting string should be treated as
1684 How do we prepare to fix this up? First we locate the code in question -
1685 the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1686 files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1687 altering this file, let's copy it to F<pp.c~>.
1689 [Well, it was in F<pp.c> when this tutorial was written. It has now been
1690 split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1692 Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1693 loop over the pattern, taking each format character in turn into
1694 C<datum_type>. Then for each possible format character, we swallow up
1695 the other arguments in the pattern (a field width, an asterisk, and so
1696 on) and convert the next chunk input into the specified format, adding
1697 it onto the output SV C<cat>.
1699 How do we know if the C<U> is the first format in the C<pat>? Well, if
1700 we have a pointer to the start of C<pat> then, if we see a C<U> we can
1701 test whether we're still at the start of the string. So, here's where
1705 register char *pat = SvPVx(*++MARK, fromlen);
1706 register char *patend = pat + fromlen;
1711 We'll have another string pointer in there:
1714 register char *pat = SvPVx(*++MARK, fromlen);
1715 register char *patend = pat + fromlen;
1721 And just before we start the loop, we'll set C<patcopy> to be the start
1726 sv_setpvn(cat, "", 0);
1728 while (pat < patend) {
1730 Now if we see a C<U> which was at the start of the string, we turn on
1731 the C<UTF8> flag for the output SV, C<cat>:
1733 + if (datumtype == 'U' && pat==patcopy+1)
1735 if (datumtype == '#') {
1736 while (pat < patend && *pat != '\n')
1739 Remember that it has to be C<patcopy+1> because the first character of
1740 the string is the C<U> which has been swallowed into C<datumtype!>
1742 Oops, we forgot one thing: what if there are spaces at the start of the
1743 pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1744 character, even though it's not the first thing in the pattern. In this
1745 case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1747 if (isSPACE(datumtype))
1752 if (isSPACE(datumtype)) {
1757 OK. That's the C part done. Now we must do two additional things before
1758 this patch is ready to go: we've changed the behaviour of Perl, and so
1759 we must document that change. We must also provide some more regression
1760 tests to make sure our patch works and doesn't create a bug somewhere
1761 else along the line.
1763 The regression tests for each operator live in F<t/op/>, and so we
1764 make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1765 tests to the end. First, we'll test that the C<U> does indeed create
1768 t/op/pack.t has a sensible ok() function, but if it didn't we could
1769 use the one from t/test.pl.
1771 require './test.pl';
1772 plan( tests => 159 );
1776 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1777 print "ok $test\n"; $test++;
1779 we can write the more sensible (see L<Test::More> for a full
1780 explanation of is() and other testing functions).
1782 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
1783 "U* produces unicode" );
1785 Now we'll test that we got that space-at-the-beginning business right:
1787 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
1788 " with spaces at the beginning" );
1790 And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1791 the first active format:
1793 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
1794 "U* not first isn't unicode" );
1796 Mustn't forget to change the number of tests which appears at the top,
1797 or else the automated tester will get confused. This will either look
1804 plan( tests => 156 );
1806 We now compile up Perl, and run it through the test suite. Our new
1809 Finally, the documentation. The job is never done until the paperwork is
1810 over, so let's describe the change we've just made. The relevant place
1811 is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1812 this text in the description of C<pack>:
1816 If the pattern begins with a C<U>, the resulting string will be treated
1817 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1818 with an initial C<U0>, and the bytes that follow will be interpreted as
1819 Unicode characters. If you don't want this to happen, you can begin your
1820 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
1821 string, and then follow this with a C<U*> somewhere in your pattern.
1823 All done. Now let's create the patch. F<Porting/patching.pod> tells us
1824 that if we're making major changes, we should copy the entire directory
1825 to somewhere safe before we begin fiddling, and then do
1827 diff -ruN old new > patch
1829 However, we know which files we've changed, and we can simply do this:
1831 diff -u pp.c~ pp.c > patch
1832 diff -u t/op/pack.t~ t/op/pack.t >> patch
1833 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1835 We end up with a patch looking a little like this:
1837 --- pp.c~ Fri Jun 02 04:34:10 2000
1838 +++ pp.c Fri Jun 16 11:37:25 2000
1839 @@ -4375,6 +4375,7 @@
1842 register char *pat = SvPVx(*++MARK, fromlen);
1844 register char *patend = pat + fromlen;
1847 @@ -4405,6 +4406,7 @@
1850 And finally, we submit it, with our rationale, to perl5-porters. Job
1853 =head2 Patching a core module
1855 This works just like patching anything else, with an extra
1856 consideration. Many core modules also live on CPAN. If this is so,
1857 patch the CPAN version instead of the core and send the patch off to
1858 the module maintainer (with a copy to p5p). This will help the module
1859 maintainer keep the CPAN version in sync with the core version without
1860 constantly scanning p5p.
1862 =head2 Adding a new function to the core
1864 If, as part of a patch to fix a bug, or just because you have an
1865 especially good idea, you decide to add a new function to the core,
1866 discuss your ideas on p5p well before you start work. It may be that
1867 someone else has already attempted to do what you are considering and
1868 can give lots of good advice or even provide you with bits of code
1869 that they already started (but never finished).
1871 You have to follow all of the advice given above for patching. It is
1872 extremely important to test any addition thoroughly and add new tests
1873 to explore all boundary conditions that your new function is expected
1874 to handle. If your new function is used only by one module (e.g. toke),
1875 then it should probably be named S_your_function (for static); on the
1876 other hand, if you expect it to accessible from other functions in
1877 Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1880 The location of any new code is also an important consideration. Don't
1881 just create a new top level .c file and put your code there; you would
1882 have to make changes to Configure (so the Makefile is created properly),
1883 as well as possibly lots of include files. This is strictly pumpking
1886 It is better to add your function to one of the existing top level
1887 source code files, but your choice is complicated by the nature of
1888 the Perl distribution. Only the files that are marked as compiled
1889 static are located in the perl executable. Everything else is located
1890 in the shared library (or DLL if you are running under WIN32). So,
1891 for example, if a function was only used by functions located in
1892 toke.c, then your code can go in toke.c. If, however, you want to call
1893 the function from universal.c, then you should put your code in another
1894 location, for example util.c.
1896 In addition to writing your c-code, you will need to create an
1897 appropriate entry in embed.pl describing your function, then run
1898 'make regen_headers' to create the entries in the numerous header
1899 files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1900 for information on the various options that you can set in embed.pl.
1901 You will forget to do this a few (or many) times and you will get
1902 warnings during the compilation phase. Make sure that you mention
1903 this when you post your patch to P5P; the pumpking needs to know this.
1905 When you write your new code, please be conscious of existing code
1906 conventions used in the perl source files. See L<perlstyle> for
1907 details. Although most of the guidelines discussed seem to focus on
1908 Perl code, rather than c, they all apply (except when they don't ;).
1909 See also I<Porting/patching.pod> file in the Perl source distribution
1910 for lots of details about both formatting and submitting patches of
1913 Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1914 Test on as many platforms as you can find. Test as many perl
1915 Configure options as you can (e.g. MULTIPLICITY). If you have
1916 profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
1917 below for how to use them to further test your code. Remember that
1918 most of the people on P5P are doing this on their own time and
1919 don't have the time to debug your code.
1921 =head2 Writing a test
1923 Every module and built-in function has an associated test file (or
1924 should...). If you add or change functionality, you have to write a
1925 test. If you fix a bug, you have to write a test so that bug never
1926 comes back. If you alter the docs, it would be nice to test what the
1927 new documentation says.
1929 In short, if you submit a patch you probably also have to patch the
1932 For modules, the test file is right next to the module itself.
1933 F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1934 so there are some snags (and it would be wonderful for you to brush
1935 them out), but it basically works that way. Everything else lives in
1942 Testing of the absolute basic functionality of Perl. Things like
1943 C<if>, basic file reads and writes, simple regexes, etc. These are
1944 run first in the test suite and if any of them fail, something is
1949 These test the basic control structures, C<if/else>, C<while>,
1954 Tests basic issues of how Perl parses and compiles itself.
1958 Tests for built-in IO functions, including command line arguments.
1962 The old home for the module tests, you shouldn't put anything new in
1963 here. There are still some bits and pieces hanging around in here
1964 that need to be moved. Perhaps you could move them? Thanks!
1968 Tests for perl's built in functions that don't fit into any of the
1973 Tests for POD directives. There are still some tests for the Pod
1974 modules hanging around in here that need to be moved out into F<lib/>.
1978 Testing features of how perl actually runs, including exit codes and
1979 handling of PERL* environment variables.
1983 Tests for the core support of Unicode.
1987 Windows-specific tests.
1991 A test suite for the s2p converter.
1995 The core uses the same testing style as the rest of Perl, a simple
1996 "ok/not ok" run through Test::Harness, but there are a few special
1999 There are three ways to write a test in the core. Test::More,
2000 t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
2001 decision of which to use depends on what part of the test suite you're
2002 working on. This is a measure to prevent a high-level failure (such
2003 as Config.pm breaking) from causing basic functionality tests to fail.
2009 Since we don't know if require works, or even subroutines, use ad hoc
2010 tests for these two. Step carefully to avoid using the feature being
2013 =item t/cmd t/run t/io t/op
2015 Now that basic require() and subroutines are tested, you can use the
2016 t/test.pl library which emulates the important features of Test::More
2017 while using a minimum of core features.
2019 You can also conditionally use certain libraries like Config, but be
2020 sure to skip the test gracefully if it's not there.
2024 Now that the core of Perl is tested, Test::More can be used. You can
2025 also use the full suite of core modules in the tests.
2029 When you say "make test" Perl uses the F<t/TEST> program to run the
2030 test suite (except under Win32 where it uses F<t/harness> instead.)
2031 All tests are run from the F<t/> directory, B<not> the directory
2032 which contains the test. This causes some problems with the tests
2033 in F<lib/>, so here's some opportunity for some patching.
2035 You must be triply conscious of cross-platform concerns. This usually
2036 boils down to using File::Spec and avoiding things like C<fork()> and
2037 C<system()> unless absolutely necessary.
2039 =head2 Special Make Test Targets
2041 There are various special make targets that can be used to test Perl
2042 slightly differently than the standard "test" target. Not all them
2043 are expected to give a 100% success rate. Many of them have several
2044 aliases, and many of them are not available on certain operating
2051 Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
2053 (Not available on Win32)
2057 Run all the tests through B::Deparse. Not all tests will succeed.
2059 (Not available on Win32)
2061 =item test.taintwarn
2063 Run all tests with the B<-t> command-line switch. Not all tests
2064 are expected to succeed (until they're specifically fixed, of course).
2066 (Not available on Win32)
2070 Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
2071 F<t/op>, and F<t/uni> tests.
2073 =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
2075 (Only in Linux) Run all the tests using the memory leak + naughty
2076 memory access tool "valgrind". The log files will be named
2077 F<testname.valgrind>.
2079 =item test.third check.third utest.third ucheck.third
2081 (Only in Tru64) Run all the tests using the memory leak + naughty
2082 memory access tool "Third Degree". The log files will be named
2083 F<perl3.log.testname>.
2085 =item test.torture torturetest
2087 Run all the usual tests and some extra tests. As of Perl 5.8.0 the
2088 only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
2090 You can also run the torture test with F<t/harness> by giving
2091 C<-torture> argument to F<t/harness>.
2093 =item utest ucheck test.utf8 check.utf8
2095 Run all the tests with -Mutf8. Not all tests will succeed.
2097 (Not available on Win32)
2099 =item minitest.utf16 test.utf16
2101 Runs the tests with UTF-16 encoded scripts, encoded with different
2102 versions of this encoding.
2104 C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
2105 C<-utf16> arguments to F<t/TEST>.
2107 (Not available on Win32)
2111 Run the test suite with the F<t/harness> controlling program, instead of
2112 F<t/TEST>. F<t/harness> is more sophisticated, and uses the
2113 L<Test::Harness> module, thus using this test target supposes that perl
2114 mostly works. The main advantage for our purposes is that it prints a
2115 detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
2116 doesn't redirect stderr to stdout.
2118 Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
2119 there is no special "test_harness" target.
2121 Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
2122 environment variables to control the behaviour of F<t/harness>. This means
2125 nmake test TEST_FILES="op/*.t"
2126 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
2128 =item test-notty test_notty
2130 Sets PERL_SKIP_TTY_TEST to true before running normal test.
2134 =head2 Running tests by hand
2136 You can run part of the test suite by hand by using one the following
2137 commands from the F<t/> directory :
2139 ./perl -I../lib TEST list-of-.t-files
2143 ./perl -I../lib harness list-of-.t-files
2145 (if you don't specify test scripts, the whole test suite will be run.)
2147 =head3 Using t/harness for testing
2149 If you use C<harness> for testing you have several command line options
2150 available to you. The arguments are as follows, and are in the order
2151 that they must appear if used together.
2153 harness -v -torture -re=pattern LIST OF FILES TO TEST
2154 harness -v -torture -re LIST OF PATTERNS TO MATCH
2156 If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
2157 the manifest. The file list may include shell wildcards which will be
2164 Run the tests under verbose mode so you can see what tests were run,
2169 Run the torture tests as well as the normal set.
2173 Filter the file list so that all the test files run match PATTERN.
2174 Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2175 in that it allows the file list to be provided as well.
2177 =item -re LIST OF PATTERNS
2179 Filter the file list so that all the test files run match
2180 /(LIST|OF|PATTERNS)/. Note that with this form the patterns
2181 are joined by '|' and you cannot supply a list of files, instead
2182 the test files are obtained from the MANIFEST.
2186 You can run an individual test by a command similar to
2188 ./perl -I../lib patho/to/foo.t
2190 except that the harnesses set up some environment variables that may
2191 affect the execution of the test :
2197 indicates that we're running this test part of the perl core test suite.
2198 This is useful for modules that have a dual life on CPAN.
2200 =item PERL_DESTRUCT_LEVEL=2
2202 is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2206 (used only by F<t/TEST>) if set, overrides the path to the perl executable
2207 that should be used to run the tests (the default being F<./perl>).
2209 =item PERL_SKIP_TTY_TEST
2211 if set, tells to skip the tests that need a terminal. It's actually set
2212 automatically by the Makefile, but can also be forced artificially by
2213 running 'make test_notty'.
2217 =head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2219 Sometimes it helps to use external tools while debugging and
2220 testing Perl. This section tries to guide you through using
2221 some common testing and debugging tools with Perl. This is
2222 meant as a guide to interfacing these tools with Perl, not
2223 as any kind of guide to the use of the tools themselves.
2225 B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2226 Third Degree greatly slows down the execution: seconds become minutes,
2227 minutes become hours. For example as of Perl 5.8.1, the
2228 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2229 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2230 than six hours, even on a snappy computer-- the said test must be
2231 doing something that is quite unfriendly for memory debuggers. If you
2232 don't feel like waiting, that you can simply kill away the perl
2235 B<NOTE 2>: To minimize the number of memory leak false alarms (see
2236 L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2237 environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2238 and harness scripts do that automatically. But if you are running
2239 some of the tests manually-- for csh-like shells:
2241 setenv PERL_DESTRUCT_LEVEL 2
2243 and for Bourne-type shells:
2245 PERL_DESTRUCT_LEVEL=2
2246 export PERL_DESTRUCT_LEVEL
2248 or in UNIXy environments you can also use the C<env> command:
2250 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
2252 B<NOTE 3>: There are known memory leaks when there are compile-time
2253 errors within eval or require, seeing C<S_doeval> in the call stack
2254 is a good sign of these. Fixing these leaks is non-trivial,
2255 unfortunately, but they must be fixed eventually.
2257 =head2 Rational Software's Purify
2259 Purify is a commercial tool that is helpful in identifying
2260 memory overruns, wild pointers, memory leaks and other such
2261 badness. Perl must be compiled in a specific way for
2262 optimal testing with Purify. Purify is available under
2263 Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2265 =head2 Purify on Unix
2267 On Unix, Purify creates a new Perl binary. To get the most
2268 benefit out of Purify, you should create the perl to Purify
2271 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2272 -Uusemymalloc -Dusemultiplicity
2274 where these arguments mean:
2278 =item -Accflags=-DPURIFY
2280 Disables Perl's arena memory allocation functions, as well as
2281 forcing use of memory allocation functions derived from the
2284 =item -Doptimize='-g'
2286 Adds debugging information so that you see the exact source
2287 statements where the problem occurs. Without this flag, all
2288 you will see is the source filename of where the error occurred.
2292 Disable Perl's malloc so that Purify can more closely monitor
2293 allocations and leaks. Using Perl's malloc will make Purify
2294 report most leaks in the "potential" leaks category.
2296 =item -Dusemultiplicity
2298 Enabling the multiplicity option allows perl to clean up
2299 thoroughly when the interpreter shuts down, which reduces the
2300 number of bogus leak reports from Purify.
2304 Once you've compiled a perl suitable for Purify'ing, then you
2309 which creates a binary named 'pureperl' that has been Purify'ed.
2310 This binary is used in place of the standard 'perl' binary
2311 when you want to debug Perl memory problems.
2313 As an example, to show any memory leaks produced during the
2314 standard Perl testset you would create and run the Purify'ed
2319 ../pureperl -I../lib harness
2321 which would run Perl on test.pl and report any memory problems.
2323 Purify outputs messages in "Viewer" windows by default. If
2324 you don't have a windowing environment or if you simply
2325 want the Purify output to unobtrusively go to a log file
2326 instead of to the interactive window, use these following
2327 options to output to the log file "perl.log":
2329 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2330 -log-file=perl.log -append-logfile=yes"
2332 If you plan to use the "Viewer" windows, then you only need this option:
2334 setenv PURIFYOPTIONS "-chain-length=25"
2336 In Bourne-type shells:
2339 export PURIFYOPTIONS
2341 or if you have the "env" utility:
2343 env PURIFYOPTIONS="..." ../pureperl ...
2347 Purify on Windows NT instruments the Perl binary 'perl.exe'
2348 on the fly. There are several options in the makefile you
2349 should change to get the most use out of Purify:
2355 You should add -DPURIFY to the DEFINES line so the DEFINES
2356 line looks something like:
2358 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
2360 to disable Perl's arena memory allocation functions, as
2361 well as to force use of memory allocation functions derived
2362 from the system malloc.
2364 =item USE_MULTI = define
2366 Enabling the multiplicity option allows perl to clean up
2367 thoroughly when the interpreter shuts down, which reduces the
2368 number of bogus leak reports from Purify.
2370 =item #PERL_MALLOC = define
2372 Disable Perl's malloc so that Purify can more closely monitor
2373 allocations and leaks. Using Perl's malloc will make Purify
2374 report most leaks in the "potential" leaks category.
2378 Adds debugging information so that you see the exact source
2379 statements where the problem occurs. Without this flag, all
2380 you will see is the source filename of where the error occurred.
2384 As an example, to show any memory leaks produced during the
2385 standard Perl testset you would create and run Purify as:
2390 purify ../perl -I../lib harness
2392 which would instrument Perl in memory, run Perl on test.pl,
2393 then finally report any memory problems.
2397 The excellent valgrind tool can be used to find out both memory leaks
2398 and illegal memory accesses. As of August 2003 it unfortunately works
2399 only on x86 (ELF) Linux. The special "test.valgrind" target can be used
2400 to run the tests under valgrind. Found errors and memory leaks are
2401 logged in files named F<test.valgrind>.
2403 As system libraries (most notably glibc) are also triggering errors,
2404 valgrind allows to suppress such errors using suppression files. The
2405 default suppression file that comes with valgrind already catches a lot
2406 of them. Some additional suppressions are defined in F<t/perl.supp>.
2408 To get valgrind and for more information see
2410 http://developer.kde.org/~sewardj/
2412 =head2 Compaq's/Digital's/HP's Third Degree
2414 Third Degree is a tool for memory leak detection and memory access checks.
2415 It is one of the many tools in the ATOM toolkit. The toolkit is only
2416 available on Tru64 (formerly known as Digital UNIX formerly known as
2419 When building Perl, you must first run Configure with -Doptimize=-g
2420 and -Uusemymalloc flags, after that you can use the make targets
2421 "perl.third" and "test.third". (What is required is that Perl must be
2422 compiled using the C<-g> flag, you may need to re-Configure.)
2424 The short story is that with "atom" you can instrument the Perl
2425 executable to create a new executable called F<perl.third>. When the
2426 instrumented executable is run, it creates a log of dubious memory
2427 traffic in file called F<perl.3log>. See the manual pages of atom and
2428 third for more information. The most extensive Third Degree
2429 documentation is available in the Compaq "Tru64 UNIX Programmer's
2430 Guide", chapter "Debugging Programs with Third Degree".
2432 The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
2433 subdirectory. There is a problem with these files: Third Degree is so
2434 effective that it finds problems also in the system libraries.
2435 Therefore you should used the Porting/thirdclean script to cleanup
2436 the F<*.3log> files.
2438 There are also leaks that for given certain definition of a leak,
2439 aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2441 =head2 PERL_DESTRUCT_LEVEL
2443 If you want to run any of the tests yourself manually using e.g.
2444 valgrind, or the pureperl or perl.third executables, please note that
2445 by default perl B<does not> explicitly cleanup all the memory it has
2446 allocated (such as global memory arenas) but instead lets the exit()
2447 of the whole program "take care" of such allocations, also known as
2448 "global destruction of objects".
2450 There is a way to tell perl to do complete cleanup: set the
2451 environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2452 The t/TEST wrapper does set this to 2, and this is what you
2453 need to do too, if you don't want to see the "global leaks":
2454 For example, for "third-degreed" Perl:
2456 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
2458 (Note: the mod_perl apache module uses also this environment variable
2459 for its own purposes and extended its semantics. Refer to the mod_perl
2460 documentation for more information. Also, spawned threads do the
2461 equivalent of setting this variable to the value 1.)
2463 If, at the end of a run you get the message I<N scalars leaked>, you can
2464 recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2465 of all those leaked SVs to be dumped along with details as to where each
2466 SV was originally allocated. This information is also displayed by
2467 Devel::Peek. Note that the extra details recorded with each SV increases
2468 memory usage, so it shouldn't be used in production environments. It also
2469 converts C<new_SV()> from a macro into a real function, so you can use
2470 your favourite debugger to discover where those pesky SVs were allocated.
2474 Depending on your platform there are various of profiling Perl.
2476 There are two commonly used techniques of profiling executables:
2477 I<statistical time-sampling> and I<basic-block counting>.
2479 The first method takes periodically samples of the CPU program
2480 counter, and since the program counter can be correlated with the code
2481 generated for functions, we get a statistical view of in which
2482 functions the program is spending its time. The caveats are that very
2483 small/fast functions have lower probability of showing up in the
2484 profile, and that periodically interrupting the program (this is
2485 usually done rather frequently, in the scale of milliseconds) imposes
2486 an additional overhead that may skew the results. The first problem
2487 can be alleviated by running the code for longer (in general this is a
2488 good idea for profiling), the second problem is usually kept in guard
2489 by the profiling tools themselves.
2491 The second method divides up the generated code into I<basic blocks>.
2492 Basic blocks are sections of code that are entered only in the
2493 beginning and exited only at the end. For example, a conditional jump
2494 starts a basic block. Basic block profiling usually works by
2495 I<instrumenting> the code by adding I<enter basic block #nnnn>
2496 book-keeping code to the generated code. During the execution of the
2497 code the basic block counters are then updated appropriately. The
2498 caveat is that the added extra code can skew the results: again, the
2499 profiling tools usually try to factor their own effects out of the
2502 =head2 Gprof Profiling
2504 gprof is a profiling tool available in many UNIX platforms,
2505 it uses F<statistical time-sampling>.
2507 You can build a profiled version of perl called "perl.gprof" by
2508 invoking the make target "perl.gprof" (What is required is that Perl
2509 must be compiled using the C<-pg> flag, you may need to re-Configure).
2510 Running the profiled version of Perl will create an output file called
2511 F<gmon.out> is created which contains the profiling data collected
2512 during the execution.
2514 The gprof tool can then display the collected data in various ways.
2515 Usually gprof understands the following options:
2521 Suppress statically defined functions from the profile.
2525 Suppress the verbose descriptions in the profile.
2529 Exclude the given routine and its descendants from the profile.
2533 Display only the given routine and its descendants in the profile.
2537 Generate a summary file called F<gmon.sum> which then may be given
2538 to subsequent gprof runs to accumulate data over several runs.
2542 Display routines that have zero usage.
2546 For more detailed explanation of the available commands and output
2547 formats, see your own local documentation of gprof.
2549 =head2 GCC gcov Profiling
2551 Starting from GCC 3.0 I<basic block profiling> is officially available
2554 You can build a profiled version of perl called F<perl.gcov> by
2555 invoking the make target "perl.gcov" (what is required that Perl must
2556 be compiled using gcc with the flags C<-fprofile-arcs
2557 -ftest-coverage>, you may need to re-Configure).
2559 Running the profiled version of Perl will cause profile output to be
2560 generated. For each source file an accompanying ".da" file will be
2563 To display the results you use the "gcov" utility (which should
2564 be installed if you have gcc 3.0 or newer installed). F<gcov> is
2565 run on source code files, like this
2569 which will cause F<sv.c.gcov> to be created. The F<.gcov> files
2570 contain the source code annotated with relative frequencies of
2571 execution indicated by "#" markers.
2573 Useful options of F<gcov> include C<-b> which will summarise the
2574 basic block, branch, and function call coverage, and C<-c> which
2575 instead of relative frequencies will use the actual counts. For
2576 more information on the use of F<gcov> and basic block profiling
2577 with gcc, see the latest GNU CC manual, as of GCC 3.0 see
2579 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
2581 and its section titled "8. gcov: a Test Coverage Program"
2583 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
2585 =head2 Pixie Profiling
2587 Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
2588 UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
2589 I<basic-block counting>.
2591 You can build a profiled version of perl called F<perl.pixie> by
2592 invoking the make target "perl.pixie" (what is required is that Perl
2593 must be compiled using the C<-g> flag, you may need to re-Configure).
2595 In Tru64 a file called F<perl.Addrs> will also be silently created,
2596 this file contains the addresses of the basic blocks. Running the
2597 profiled version of Perl will create a new file called "perl.Counts"
2598 which contains the counts for the basic block for that particular
2601 To display the results you use the F<prof> utility. The exact
2602 incantation depends on your operating system, "prof perl.Counts" in
2603 IRIX, and "prof -pixie -all -L. perl" in Tru64.
2605 In IRIX the following prof options are available:
2611 Reports the most heavily used lines in descending order of use.
2612 Useful for finding the hotspot lines.
2616 Groups lines by procedure, with procedures sorted in descending order of use.
2617 Within a procedure, lines are listed in source order.
2618 Useful for finding the hotspots of procedures.
2622 In Tru64 the following options are available:
2628 Procedures sorted in descending order by the number of cycles executed
2629 in each procedure. Useful for finding the hotspot procedures.
2630 (This is the default option.)
2634 Lines sorted in descending order by the number of cycles executed in
2635 each line. Useful for finding the hotspot lines.
2637 =item -i[nvocations]
2639 The called procedures are sorted in descending order by number of calls
2640 made to the procedures. Useful for finding the most used procedures.
2644 Grouped by procedure, sorted by cycles executed per procedure.
2645 Useful for finding the hotspots of procedures.
2649 The compiler emitted code for these lines, but the code was unexecuted.
2653 Unexecuted procedures.
2657 For further information, see your system's manual pages for pixie and prof.
2659 =head2 Miscellaneous tricks
2665 Those debugging perl with the DDD frontend over gdb may find the
2668 You can extend the data conversion shortcuts menu, so for example you
2669 can display an SV's IV value with one click, without doing any typing.
2670 To do that simply edit ~/.ddd/init file and add after:
2672 ! Display shortcuts.
2673 Ddd*gdbDisplayShortcuts: \
2674 /t () // Convert to Bin\n\
2675 /d () // Convert to Dec\n\
2676 /x () // Convert to Hex\n\
2677 /o () // Convert to Oct(\n\
2679 the following two lines:
2681 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
2682 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
2684 so now you can do ivx and pvx lookups or you can plug there the
2685 sv_peek "conversion":
2687 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
2689 (The my_perl is for threaded builds.)
2690 Just remember that every line, but the last one, should end with \n\
2692 Alternatively edit the init file interactively via:
2693 3rd mouse button -> New Display -> Edit Menu
2695 Note: you can define up to 20 conversion shortcuts in the gdb
2700 If you see in a debugger a memory area mysteriously full of 0xabababab,
2701 you may be seeing the effect of the Poison() macro, see L<perlclib>.
2707 We've had a brief look around the Perl source, an overview of the stages
2708 F<perl> goes through when it's running your code, and how to use a
2709 debugger to poke at the Perl guts. We took a very simple problem and
2710 demonstrated how to solve it fully - with documentation, regression
2711 tests, and finally a patch for submission to p5p. Finally, we talked
2712 about how to use external tools to debug and test Perl.
2714 I'd now suggest you read over those references again, and then, as soon
2715 as possible, get your hands dirty. The best way to learn is by doing,
2722 Subscribe to perl5-porters, follow the patches and try and understand
2723 them; don't be afraid to ask if there's a portion you're not clear on -
2724 who knows, you may unearth a bug in the patch...
2728 Keep up to date with the bleeding edge Perl distributions and get
2729 familiar with the changes. Try and get an idea of what areas people are
2730 working on and the changes they're making.
2734 Do read the README associated with your operating system, e.g. README.aix
2735 on the IBM AIX OS. Don't hesitate to supply patches to that README if
2736 you find anything missing or changed over a new OS release.
2740 Find an area of Perl that seems interesting to you, and see if you can
2741 work out how it works. Scan through the source, and step over it in the
2742 debugger. Play, poke, investigate, fiddle! You'll probably get to
2743 understand not just your chosen area but a much wider range of F<perl>'s
2744 activity as well, and probably sooner than you'd think.
2750 =item I<The Road goes ever on and on, down from the door where it began.>
2754 If you can do these things, you've started on the long road to Perl porting.
2755 Thanks for wanting to help make Perl better - and happy hacking!
2759 This document was written by Nathan Torkington, and is maintained by
2760 the perl5-porters mailing list.