This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
URL correction by Robert Spier
[perl5.git] / pod / perlhack.pod
CommitLineData
e8cd7eae
GS
1=head1 NAME
2
3perlhack - How to hack at the Perl internals
4
5=head1 DESCRIPTION
6
7This document attempts to explain how Perl development takes place,
8and ends with some suggestions for people wanting to become bona fide
9porters.
10
11The perl5-porters mailing list is where the Perl standard distribution
12is maintained and developed. The list can get anywhere from 10 to 150
13messages a day, depending on the heatedness of the debate. Most days
14there are two or three patches, extensions, features, or bugs being
15discussed at a time.
16
f8e3975a 17A searchable archive of the list is at either:
e8cd7eae
GS
18
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
20
f8e3975a
IP
21or
22
23 http://archive.develooper.com/perl5-porters@perl.org/
24
e8cd7eae
GS
25List subscribers (the porters themselves) come in several flavours.
26Some are quiet curious lurkers, who rarely pitch in and instead watch
27the ongoing development to ensure they're forewarned of new changes or
28features in Perl. Some are representatives of vendors, who are there
29to make sure that Perl continues to compile and work on their
30platforms. Some patch any reported bug that they know how to fix,
31some are actively patching their pet area (threads, Win32, the regexp
32engine), while others seem to do nothing but complain. In other
33words, it's your usual mix of technical people.
34
35Over this group of porters presides Larry Wall. He has the final word
f6c51b38 36in what does and does not change in the Perl language. Various
b432a672
AL
37releases of Perl are shepherded by a "pumpking", a porter
38responsible for gathering patches, deciding on a patch-by-patch,
f6c51b38 39feature-by-feature basis what will and will not go into the release.
caf100c0 40For instance, Gurusamy Sarathy was the pumpking for the 5.6 release of
961f29c6 41Perl, and Jarkko Hietaniemi was the pumpking for the 5.8 release, and
1a88dbf8 42Rafael Garcia-Suarez holds the pumpking crown for the 5.10 release.
e8cd7eae
GS
43
44In addition, various people are pumpkings for different things. For
961f29c6
MB
45instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the
46I<Configure> pumpkin up till the 5.8 release. For the 5.10 release
47H.Merijn Brand took over.
e8cd7eae
GS
48
49Larry sees Perl development along the lines of the US government:
50there's the Legislature (the porters), the Executive branch (the
51pumpkings), and the Supreme Court (Larry). The legislature can
52discuss and submit patches to the executive branch all they like, but
53the executive branch is free to veto them. Rarely, the Supreme Court
54will side with the executive branch over the legislature, or the
55legislature over the executive branch. Mostly, however, the
56legislature and the executive branch are supposed to get along and
57work out their differences without impeachment or court cases.
58
59You might sometimes see reference to Rule 1 and Rule 2. Larry's power
60as Supreme Court is expressed in The Rules:
61
62=over 4
63
64=item 1
65
66Larry is always by definition right about how Perl should behave.
67This means he has final veto power on the core functionality.
68
69=item 2
70
71Larry is allowed to change his mind about any matter at a later date,
72regardless of whether he previously invoked Rule 1.
73
74=back
75
76Got that? Larry is always right, even when he was wrong. It's rare
77to see either Rule exercised, but they are often alluded to.
78
79New features and extensions to the language are contentious, because
80the criteria used by the pumpkings, Larry, and other porters to decide
81which features should be implemented and incorporated are not codified
82in a few small design goals as with some other languages. Instead,
83the heuristics are flexible and often difficult to fathom. Here is
84one person's list, roughly in decreasing order of importance, of
85heuristics that new features have to be weighed against:
86
87=over 4
88
89=item Does concept match the general goals of Perl?
90
91These haven't been written anywhere in stone, but one approximation
92is:
93
94 1. Keep it fast, simple, and useful.
95 2. Keep features/concepts as orthogonal as possible.
96 3. No arbitrary limits (platforms, data sizes, cultures).
97 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
98 5. Either assimilate new technologies, or build bridges to them.
99
100=item Where is the implementation?
101
102All the talk in the world is useless without an implementation. In
103almost every case, the person or people who argue for a new feature
104will be expected to be the ones who implement it. Porters capable
105of coding new features have their own agendas, and are not available
106to implement your (possibly good) idea.
107
108=item Backwards compatibility
109
110It's a cardinal sin to break existing Perl programs. New warnings are
111contentious--some say that a program that emits warnings is not
112broken, while others say it is. Adding keywords has the potential to
113break programs, changing the meaning of existing token sequences or
114functions might break programs.
115
116=item Could it be a module instead?
117
118Perl 5 has extension mechanisms, modules and XS, specifically to avoid
119the need to keep changing the Perl interpreter. You can write modules
120that export functions, you can give those functions prototypes so they
121can be called like built-in functions, you can even write XS code to
122mess with the runtime data structures of the Perl interpreter if you
123want to implement really complicated things. If it can be done in a
124module instead of in the core, it's highly unlikely to be added.
125
126=item Is the feature generic enough?
127
128Is this something that only the submitter wants added to the language,
129or would it be broadly useful? Sometimes, instead of adding a feature
130with a tight focus, the porters might decide to wait until someone
131implements the more generalized feature. For instance, instead of
b432a672 132implementing a "delayed evaluation" feature, the porters are waiting
e8cd7eae
GS
133for a macro system that would permit delayed evaluation and much more.
134
135=item Does it potentially introduce new bugs?
136
137Radical rewrites of large chunks of the Perl interpreter have the
138potential to introduce new bugs. The smaller and more localized the
139change, the better.
140
141=item Does it preclude other desirable features?
142
143A patch is likely to be rejected if it closes off future avenues of
144development. For instance, a patch that placed a true and final
145interpretation on prototypes is likely to be rejected because there
146are still options for the future of prototypes that haven't been
147addressed.
148
149=item Is the implementation robust?
150
151Good patches (tight code, complete, correct) stand more chance of
152going in. Sloppy or incorrect patches might be placed on the back
153burner until the pumpking has time to fix, or might be discarded
154altogether without further notice.
155
156=item Is the implementation generic enough to be portable?
157
158The worst patches make use of a system-specific features. It's highly
159unlikely that nonportable additions to the Perl language will be
160accepted.
161
a936dd3c
NC
162=item Is the implementation tested?
163
164Patches which change behaviour (fixing bugs or introducing new features)
165must include regression tests to verify that everything works as expected.
166Without tests provided by the original author, how can anyone else changing
167perl in the future be sure that they haven't unwittingly broken the behaviour
168the patch implements? And without tests, how can the patch's author be
9d077eaa 169confident that his/her hard work put into the patch won't be accidentally
a936dd3c
NC
170thrown away by someone in the future?
171
e8cd7eae
GS
172=item Is there enough documentation?
173
174Patches without documentation are probably ill-thought out or
175incomplete. Nothing can be added without documentation, so submitting
176a patch for the appropriate manpages as well as the source code is
a936dd3c 177always a good idea.
e8cd7eae
GS
178
179=item Is there another way to do it?
180
b432a672
AL
181Larry said "Although the Perl Slogan is I<There's More Than One Way
182to Do It>, I hesitate to make 10 ways to do something". This is a
e8cd7eae
GS
183tricky heuristic to navigate, though--one man's essential addition is
184another man's pointless cruft.
185
186=item Does it create too much work?
187
188Work for the pumpking, work for Perl programmers, work for module
189authors, ... Perl is supposed to be easy.
190
f6c51b38
GS
191=item Patches speak louder than words
192
193Working code is always preferred to pie-in-the-sky ideas. A patch to
194add a feature stands a much higher chance of making it to the language
195than does a random feature request, no matter how fervently argued the
b432a672 196request might be. This ties into "Will it be useful?", as the fact
f6c51b38
GS
197that someone took the time to make the patch demonstrates a strong
198desire for the feature.
199
e8cd7eae
GS
200=back
201
b432a672
AL
202If you're on the list, you might hear the word "core" bandied
203around. It refers to the standard distribution. "Hacking on the
204core" means you're changing the C source code to the Perl
205interpreter. "A core module" is one that ships with Perl.
e8cd7eae 206
a1f349fd
MB
207=head2 Keeping in sync
208
e8cd7eae 209The source code to the Perl interpreter, in its different versions, is
f224927c
JH
210kept in a repository managed by a revision control system ( which is
211currently the Perforce program, see http://perforce.com/ ). The
e8cd7eae
GS
212pumpkings and a few others have access to the repository to check in
213changes. Periodically the pumpking for the development version of Perl
214will release a new version, so the rest of the porters can see what's
2be4c08b
GS
215changed. The current state of the main trunk of repository, and patches
216that describe the individual changes that have happened since the last
217public release are available at this location:
218
eb8afeef
SP
219 http://public.activestate.com/pub/apc/
220 ftp://public.activestate.com/pub/apc/
2be4c08b 221
0cfb3454
GS
222If you're looking for a particular change, or a change that affected
223a particular set of files, you may find the B<Perl Repository Browser>
224useful:
225
226 http://public.activestate.com/cgi-bin/perlbrowse
227
228You may also want to subscribe to the perl5-changes mailing list to
229receive a copy of each patch that gets submitted to the maintenance
230and development "branches" of the perl repository. See
231http://lists.perl.org/ for subscription information.
232
a1f349fd
MB
233If you are a member of the perl5-porters mailing list, it is a good
234thing to keep in touch with the most recent changes. If not only to
235verify if what you would have posted as a bug report isn't already
236solved in the most recent available perl development branch, also
237known as perl-current, bleading edge perl, bleedperl or bleadperl.
2be4c08b
GS
238
239Needless to say, the source code in perl-current is usually in a perpetual
240state of evolution. You should expect it to be very buggy. Do B<not> use
241it for any purpose other than testing and development.
e8cd7eae 242
3e148164
JH
243Keeping in sync with the most recent branch can be done in several ways,
244but the most convenient and reliable way is using B<rsync>, available at
245ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
246branch by FTP.)
a1f349fd
MB
247
248If you choose to keep in sync using rsync, there are two approaches
3e148164 249to doing so:
a1f349fd
MB
250
251=over 4
252
253=item rsync'ing the source tree
254
3e148164 255Presuming you are in the directory where your perl source resides
b432a672 256and you have rsync installed and available, you can "upgrade" to
a1f349fd
MB
257the bleadperl using:
258
eb8afeef 259 # rsync -avz rsync://public.activestate.com/perl-current/ .
a1f349fd
MB
260
261This takes care of updating every single item in the source tree to
262the latest applied patch level, creating files that are new (to your
263distribution) and setting date/time stamps of existing files to
264reflect the bleadperl status.
265
c6d0653e
LC
266Note that this will not delete any files that were in '.' before
267the rsync. Once you are sure that the rsync is running correctly,
268run it with the --delete and the --dry-run options like this:
269
eb8afeef 270 # rsync -avz --delete --dry-run rsync://public.activestate.com/perl-current/ .
c6d0653e
LC
271
272This will I<simulate> an rsync run that also deletes files not
273present in the bleadperl master copy. Observe the results from
274this run closely. If you are sure that the actual run would delete
275no files precious to you, you could remove the '--dry-run' option.
276
a1f349fd
MB
277You can than check what patch was the latest that was applied by
278looking in the file B<.patch>, which will show the number of the
279latest patch.
280
281If you have more than one machine to keep in sync, and not all of
282them have access to the WAN (so you are not able to rsync all the
283source trees to the real source), there are some ways to get around
284this problem.
285
286=over 4
287
288=item Using rsync over the LAN
289
290Set up a local rsync server which makes the rsynced source tree
3e148164 291available to the LAN and sync the other machines against this
a1f349fd
MB
292directory.
293
1577cd80 294From http://rsync.samba.org/README.html :
a1f349fd
MB
295
296 "Rsync uses rsh or ssh for communication. It does not need to be
297 setuid and requires no special privileges for installation. It
3958b146 298 does not require an inetd entry or a daemon. You must, however,
a1f349fd
MB
299 have a working rsh or ssh system. Using ssh is recommended for
300 its security features."
301
302=item Using pushing over the NFS
303
304Having the other systems mounted over the NFS, you can take an
3e148164
JH
305active pushing approach by checking the just updated tree against
306the other not-yet synced trees. An example would be
307
308 #!/usr/bin/perl -w
309
310 use strict;
311 use File::Copy;
312
313 my %MF = map {
314 m/(\S+)/;
315 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
316 } `cat MANIFEST`;
317
318 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
319
320 foreach my $host (keys %remote) {
321 unless (-d $remote{$host}) {
322 print STDERR "Cannot Xsync for host $host\n";
323 next;
324 }
325 foreach my $file (keys %MF) {
326 my $rfile = "$remote{$host}/$file";
327 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
328 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
329 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
330 printf "%4s %-34s %8d %9d %8d %9d\n",
331 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
332 unlink $rfile;
333 copy ($file, $rfile);
334 utime time, $MF{$file}[2], $rfile;
335 chmod $MF{$file}[0], $rfile;
336 }
337 }
338
339though this is not perfect. It could be improved with checking
a1f349fd
MB
340file checksums before updating. Not all NFS systems support
341reliable utime support (when used over the NFS).
342
343=back
344
345=item rsync'ing the patches
346
347The source tree is maintained by the pumpking who applies patches to
348the files in the tree. These patches are either created by the
349pumpking himself using C<diff -c> after updating the file manually or
350by applying patches sent in by posters on the perl5-porters list.
351These patches are also saved and rsync'able, so you can apply them
352yourself to the source files.
353
354Presuming you are in a directory where your patches reside, you can
3e148164 355get them in sync with
a1f349fd 356
eb8afeef 357 # rsync -avz rsync://public.activestate.com/perl-current-diffs/ .
a1f349fd
MB
358
359This makes sure the latest available patch is downloaded to your
360patch directory.
361
3e148164 362It's then up to you to apply these patches, using something like
a1f349fd 363
7d3c2c28 364 # last="`cat ../perl-current/.patch`.gz"
eb8afeef 365 # rsync -avz rsync://public.activestate.com/perl-current-diffs/ .
a1f349fd
MB
366 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
367 # cd ../perl-current
368 # patch -p1 -N <../perl-current-diffs/blead.patch
369
370or, since this is only a hint towards how it works, use CPAN-patchaperl
371from Andreas K├Ânig to have better control over the patching process.
372
373=back
374
f7e1e956 375=head2 Why rsync the source tree
a1f349fd
MB
376
377=over 4
378
10f58044 379=item It's easier to rsync the source tree
a1f349fd
MB
380
381Since you don't have to apply the patches yourself, you are sure all
382files in the source tree are in the right state.
383
a1f349fd
MB
384=item It's more reliable
385
0cfb3454
GS
386While both the rsync-able source and patch areas are automatically
387updated every few minutes, keep in mind that applying patches may
388sometimes mean careful hand-holding, especially if your version of
389the C<patch> program does not understand how to deal with new files,
390files with 8-bit characters, or files without trailing newlines.
a1f349fd
MB
391
392=back
393
f7e1e956 394=head2 Why rsync the patches
a1f349fd
MB
395
396=over 4
397
10f58044 398=item It's easier to rsync the patches
a1f349fd
MB
399
400If you have more than one machine that you want to keep in track with
3e148164 401bleadperl, it's easier to rsync the patches only once and then apply
a1f349fd
MB
402them to all the source trees on the different machines.
403
404In case you try to keep in pace on 5 different machines, for which
405only one of them has access to the WAN, rsync'ing all the source
3e148164 406trees should than be done 5 times over the NFS. Having
a1f349fd 407rsync'ed the patches only once, I can apply them to all the source
3e148164 408trees automatically. Need you say more ;-)
a1f349fd
MB
409
410=item It's a good reference
411
412If you do not only like to have the most recent development branch,
413but also like to B<fix> bugs, or extend features, you want to dive
414into the sources. If you are a seasoned perl core diver, you don't
415need no manuals, tips, roadmaps, perlguts.pod or other aids to find
416your way around. But if you are a starter, the patches may help you
417in finding where you should start and how to change the bits that
418bug you.
419
420The file B<Changes> is updated on occasions the pumpking sees as his
421own little sync points. On those occasions, he releases a tar-ball of
422the current source tree (i.e. perl@7582.tar.gz), which will be an
423excellent point to start with when choosing to use the 'rsync the
424patches' scheme. Starting with perl@7582, which means a set of source
425files on which the latest applied patch is number 7582, you apply all
f18956b7 426succeeding patches available from then on (7583, 7584, ...).
a1f349fd
MB
427
428You can use the patches later as a kind of search archive.
429
430=over 4
431
432=item Finding a start point
433
434If you want to fix/change the behaviour of function/feature Foo, just
435scan the patches for patches that mention Foo either in the subject,
3e148164 436the comments, or the body of the fix. A good chance the patch shows
a1f349fd
MB
437you the files that are affected by that patch which are very likely
438to be the starting point of your journey into the guts of perl.
439
440=item Finding how to fix a bug
441
442If you've found I<where> the function/feature Foo misbehaves, but you
443don't know how to fix it (but you do know the change you want to
444make), you can, again, peruse the patches for similar changes and
445look how others apply the fix.
446
447=item Finding the source of misbehaviour
448
449When you keep in sync with bleadperl, the pumpking would love to
3958b146 450I<see> that the community efforts really work. So after each of his
a1f349fd
MB
451sync points, you are to 'make test' to check if everything is still
452in working order. If it is, you do 'make ok', which will send an OK
902821cc 453report to I<perlbug@perl.org>. (If you do not have access to a mailer
3e148164 454from the system you just finished successfully 'make test', you can
a1f349fd
MB
455do 'make okfile', which creates the file C<perl.ok>, which you can
456than take to your favourite mailer and mail yourself).
457
3958b146 458But of course, as always, things will not always lead to a success
a1f349fd
MB
459path, and one or more test do not pass the 'make test'. Before
460sending in a bug report (using 'make nok' or 'make nokfile'), check
461the mailing list if someone else has reported the bug already and if
462so, confirm it by replying to that message. If not, you might want to
463trace the source of that misbehaviour B<before> sending in the bug,
464which will help all the other porters in finding the solution.
465
3e148164
JH
466Here the saved patches come in very handy. You can check the list of
467patches to see which patch changed what file and what change caused
468the misbehaviour. If you note that in the bug report, it saves the
469one trying to solve it, looking for that point.
a1f349fd
MB
470
471=back
472
473If searching the patches is too bothersome, you might consider using
474perl's bugtron to find more information about discussions and
475ramblings on posted bugs.
476
3e148164
JH
477If you want to get the best of both worlds, rsync both the source
478tree for convenience, reliability and ease and rsync the patches
479for reference.
480
52315700
RF
481=back
482
fcc89a64
DR
483=head2 Working with the source
484
485Because you cannot use the Perforce client, you cannot easily generate
486diffs against the repository, nor will merges occur when you update
487via rsync. If you edit a file locally and then rsync against the
488latest source, changes made in the remote copy will I<overwrite> your
489local versions!
490
491The best way to deal with this is to maintain a tree of symlinks to
492the rsync'd source. Then, when you want to edit a file, you remove
493the symlink, copy the real file into the other tree, and edit it. You
494can then diff your edited file against the original to generate a
495patch, and you can safely update the original tree.
496
497Perl's F<Configure> script can generate this tree of symlinks for you.
498The following example assumes that you have used rsync to pull a copy
499of the Perl source into the F<perl-rsync> directory. In the directory
500above that one, you can execute the following commands:
501
502 mkdir perl-dev
503 cd perl-dev
504 ../perl-rsync/Configure -Dmksymlinks -Dusedevel -D"optimize=-g"
505
506This will start the Perl configuration process. After a few prompts,
507you should see something like this:
508
509 Symbolic links are supported.
510
511 Checking how to test for symbolic links...
512 Your builtin 'test -h' may be broken.
513 Trying external '/usr/bin/test -h'.
514 You can test for symbolic links with '/usr/bin/test -h'.
515
516 Creating the symbolic links...
517 (First creating the subdirectories...)
518 (Then creating the symlinks...)
519
520The specifics may vary based on your operating system, of course.
521After you see this, you can abort the F<Configure> script, and you
522will see that the directory you are in has a tree of symlinks to the
523F<perl-rsync> directories and files.
524
525If you plan to do a lot of work with the Perl source, here are some
526Bourne shell script functions that can make your life easier:
527
528 function edit {
529 if [ -L $1 ]; then
530 mv $1 $1.orig
62aa7ed0
RGS
531 cp $1.orig $1
532 vi $1
fcc89a64 533 else
62aa7ed0
RGS
534 vi $1
535 fi
fcc89a64
DR
536 }
537
538 function unedit {
539 if [ -L $1.orig ]; then
540 rm $1
62aa7ed0
RGS
541 mv $1.orig $1
542 fi
fcc89a64
DR
543 }
544
545Replace "vi" with your favorite flavor of editor.
546
547Here is another function which will quickly generate a patch for the
548files which have been edited in your symlink tree:
549
550 mkpatchorig() {
551 local diffopts
62aa7ed0
RGS
552 for f in `find . -name '*.orig' | sed s,^\./,,`
553 do
554 case `echo $f | sed 's,.orig$,,;s,.*\.,,'` in
555 c) diffopts=-p ;;
fcc89a64
DR
556 pod) diffopts='-F^=' ;;
557 *) diffopts= ;;
62aa7ed0
RGS
558 esac
559 diff -du $diffopts $f `echo $f | sed 's,.orig$,,'`
560 done
fcc89a64
DR
561 }
562
563This function produces patches which include enough context to make
564your changes obvious. This makes it easier for the Perl pumpking(s)
565to review them when you send them to the perl5-porters list, and that
566means they're more likely to get applied.
567
568This function assumed a GNU diff, and may require some tweaking for
569other diff variants.
52315700 570
3fd28c4e 571=head2 Perlbug administration
52315700 572
902821cc
RGS
573There is a single remote administrative interface for modifying bug status,
574category, open issues etc. using the B<RT> bugtracker system, maintained
575by Robert Spier. Become an administrator, and close any bugs you can get
3fd28c4e 576your sticky mitts on:
52315700 577
39417508 578 http://bugs.perl.org/
52315700 579
3fd28c4e 580To email the bug system administrators:
52315700 581
3fd28c4e 582 "perlbug-admin" <perlbug-admin@perl.org>
52315700 583
a1f349fd
MB
584=head2 Submitting patches
585
f7e1e956
MS
586Always submit patches to I<perl5-porters@perl.org>. If you're
587patching a core module and there's an author listed, send the author a
588copy (see L<Patching a core module>). This lets other porters review
589your patch, which catches a surprising number of errors in patches.
590Either use the diff program (available in source code form from
f224927c 591ftp://ftp.gnu.org/pub/gnu/ , or use Johan Vromans' I<makepatch>
f7e1e956
MS
592(available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
593but context diffs are accepted. Do not send RCS-style diffs or diffs
594without context lines. More information is given in the
595I<Porting/patching.pod> file in the Perl source distribution. Please
824d470b
RGS
596patch against the latest B<development> version. (e.g., even if you're
597fixing a bug in the 5.8 track, patch against the latest B<development>
598version rsynced from rsync://public.activestate.com/perl-current/ )
599
600If changes are accepted, they are applied to the development branch. Then
601the 5.8 pumpking decides which of those patches is to be backported to the
602maint branch. Only patches that survive the heat of the development
f7e1e956
MS
603branch get applied to maintenance versions.
604
2c8694a7
JH
605Your patch should update the documentation and test suite. See
606L<Writing a test>. If you have added or removed files in the distribution,
607edit the MANIFEST file accordingly, sort the MANIFEST file using
608C<make manisort>, and include those changes as part of your patch.
e8cd7eae 609
824d470b
RGS
610Patching documentation also follows the same order: if accepted, a patch
611is first applied to B<development>, and if relevant then it's backported
612to B<maintenance>. (With an exception for some patches that document
613behaviour that only appears in the maintenance branch, but which has
614changed in the development version.)
615
e8cd7eae
GS
616To report a bug in Perl, use the program I<perlbug> which comes with
617Perl (if you can't get Perl to work, send mail to the address
f18956b7 618I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
e8cd7eae 619I<perlbug> feeds into the automated bug-tracking system, access to
902821cc 620which is provided through the web at http://rt.perl.org/rt3/ . It
e8cd7eae
GS
621often pays to check the archives of the perl5-porters mailing list to
622see whether the bug you're reporting has been reported before, and if
623so whether it was considered a bug. See above for the location of
624the searchable archives.
625
f224927c 626The CPAN testers ( http://testers.cpan.org/ ) are a group of
ba139f7d 627volunteers who test CPAN modules on a variety of platforms. Perl
d3e8af89
GS
628Smokers ( http://www.nntp.perl.org/group/perl.daily-build and
629http://www.nntp.perl.org/group/perl.daily-build.reports/ )
902821cc 630automatically test Perl source releases on platforms with various
d3e8af89
GS
631configurations. Both efforts welcome volunteers. In order to get
632involved in smoke testing of the perl itself visit
633L<http://search.cpan.org/dist/Test-Smoke>. In order to start smoke
634testing CPAN modules visit L<http://search.cpan.org/dist/CPAN-YACSmoke/>
635or L<http://search.cpan.org/dist/POE-Component-CPAN-YACSmoke/> or
636L<http://search.cpan.org/dist/CPAN-Reporter/>.
e8cd7eae 637
e8cd7eae
GS
638It's a good idea to read and lurk for a while before chipping in.
639That way you'll get to see the dynamic of the conversations, learn the
640personalities of the players, and hopefully be better prepared to make
641a useful contribution when do you speak up.
642
643If after all this you still think you want to join the perl5-porters
f6c51b38
GS
644mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
645unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
e8cd7eae 646
a422fd2d
SC
647To hack on the Perl guts, you'll need to read the following things:
648
649=over 3
650
651=item L<perlguts>
652
653This is of paramount importance, since it's the documentation of what
654goes where in the Perl source. Read it over a couple of times and it
655might start to make sense - don't worry if it doesn't yet, because the
656best way to study it is to read it in conjunction with poking at Perl
657source, and we'll do that later on.
658
659You might also want to look at Gisle Aas's illustrated perlguts -
660there's no guarantee that this will be absolutely up-to-date with the
661latest documentation in the Perl core, but the fundamentals will be
1577cd80 662right. ( http://gisle.aas.no/perl/illguts/ )
a422fd2d
SC
663
664=item L<perlxstut> and L<perlxs>
665
666A working knowledge of XSUB programming is incredibly useful for core
667hacking; XSUBs use techniques drawn from the PP code, the portion of the
668guts that actually executes a Perl program. It's a lot gentler to learn
669those techniques from simple examples and explanation than from the core
670itself.
671
672=item L<perlapi>
673
674The documentation for the Perl API explains what some of the internal
675functions do, as well as the many macros used in the source.
676
677=item F<Porting/pumpkin.pod>
678
679This is a collection of words of wisdom for a Perl porter; some of it is
680only useful to the pumpkin holder, but most of it applies to anyone
681wanting to go about Perl development.
682
683=item The perl5-porters FAQ
684
902821cc
RGS
685This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html .
686It contains hints on reading perl5-porters, information on how
687perl5-porters works and how Perl development in general works.
a422fd2d
SC
688
689=back
690
691=head2 Finding Your Way Around
692
693Perl maintenance can be split into a number of areas, and certain people
694(pumpkins) will have responsibility for each area. These areas sometimes
695correspond to files or directories in the source kit. Among the areas are:
696
697=over 3
698
699=item Core modules
700
701Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
702subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
703contains the core XS modules.
704
f7e1e956
MS
705=item Tests
706
707There are tests for nearly all the modules, built-ins and major bits
708of functionality. Test files all have a .t suffix. Module tests live
709in the F<lib/> and F<ext/> directories next to the module being
710tested. Others live in F<t/>. See L<Writing a test>
711
a422fd2d
SC
712=item Documentation
713
714Documentation maintenance includes looking after everything in the
715F<pod/> directory, (as well as contributing new documentation) and
716the documentation to the modules in core.
717
718=item Configure
719
720The configure process is the way we make Perl portable across the
721myriad of operating systems it supports. Responsibility for the
722configure, build and installation process, as well as the overall
723portability of the core code rests with the configure pumpkin - others
724help out with individual operating systems.
725
726The files involved are the operating system directories, (F<win32/>,
727F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
728and F<Makefile>, as well as the metaconfig files which generate
729F<Configure>. (metaconfig isn't included in the core distribution.)
730
731=item Interpreter
732
733And of course, there's the core of the Perl interpreter itself. Let's
734have a look at that in a little more detail.
735
736=back
737
738Before we leave looking at the layout, though, don't forget that
739F<MANIFEST> contains not only the file names in the Perl distribution,
740but short descriptions of what's in them, too. For an overview of the
741important files, try this:
742
743 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
744
745=head2 Elements of the interpreter
746
747The work of the interpreter has two main stages: compiling the code
748into the internal representation, or bytecode, and then executing it.
749L<perlguts/Compiled code> explains exactly how the compilation stage
750happens.
751
752Here is a short breakdown of perl's operation:
753
754=over 3
755
756=item Startup
757
758The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
759This is very high-level code, enough to fit on a single screen, and it
760resembles the code found in L<perlembed>; most of the real action takes
761place in F<perl.c>
762
763First, F<perlmain.c> allocates some memory and constructs a Perl
764interpreter:
765
766 1 PERL_SYS_INIT3(&argc,&argv,&env);
767 2
768 3 if (!PL_do_undump) {
769 4 my_perl = perl_alloc();
770 5 if (!my_perl)
771 6 exit(1);
772 7 perl_construct(my_perl);
773 8 PL_perl_destruct_level = 0;
774 9 }
775
776Line 1 is a macro, and its definition is dependent on your operating
777system. Line 3 references C<PL_do_undump>, a global variable - all
778global variables in Perl start with C<PL_>. This tells you whether the
779current running program was created with the C<-u> flag to perl and then
780F<undump>, which means it's going to be false in any sane context.
781
782Line 4 calls a function in F<perl.c> to allocate memory for a Perl
783interpreter. It's quite a simple function, and the guts of it looks like
784this:
785
786 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
787
788Here you see an example of Perl's system abstraction, which we'll see
789later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
790own C<malloc> as defined in F<malloc.c> if you selected that option at
791configure time.
792
793Next, in line 7, we construct the interpreter; this sets up all the
794special variables that Perl needs, the stacks, and so on.
795
796Now we pass Perl the command line options, and tell it to go:
797
798 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
799 if (!exitstatus) {
800 exitstatus = perl_run(my_perl);
801 }
802
803
804C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
805in F<perl.c>, which processes the command line options, sets up any
806statically linked XS modules, opens the program and calls C<yyparse> to
807parse it.
808
809=item Parsing
810
811The aim of this stage is to take the Perl source, and turn it into an op
812tree. We'll see what one of those looks like later. Strictly speaking,
813there's three things going on here.
814
815C<yyparse>, the parser, lives in F<perly.c>, although you're better off
816reading the original YACC input in F<perly.y>. (Yes, Virginia, there
817B<is> a YACC grammar for Perl!) The job of the parser is to take your
b432a672 818code and "understand" it, splitting it into sentences, deciding which
a422fd2d
SC
819operands go with which operators and so on.
820
821The parser is nobly assisted by the lexer, which chunks up your input
822into tokens, and decides what type of thing each token is: a variable
823name, an operator, a bareword, a subroutine, a core function, and so on.
824The main point of entry to the lexer is C<yylex>, and that and its
825associated routines can be found in F<toke.c>. Perl isn't much like
826other computer languages; it's highly context sensitive at times, it can
827be tricky to work out what sort of token something is, or where a token
828ends. As such, there's a lot of interplay between the tokeniser and the
829parser, which can get pretty frightening if you're not used to it.
830
831As the parser understands a Perl program, it builds up a tree of
832operations for the interpreter to perform during execution. The routines
833which construct and link together the various operations are to be found
834in F<op.c>, and will be examined later.
835
836=item Optimization
837
838Now the parsing stage is complete, and the finished tree represents
839the operations that the Perl interpreter needs to perform to execute our
840program. Next, Perl does a dry run over the tree looking for
841optimisations: constant expressions such as C<3 + 4> will be computed
842now, and the optimizer will also see if any multiple operations can be
843replaced with a single one. For instance, to fetch the variable C<$foo>,
844instead of grabbing the glob C<*foo> and looking at the scalar
845component, the optimizer fiddles the op tree to use a function which
846directly looks up the scalar in question. The main optimizer is C<peep>
847in F<op.c>, and many ops have their own optimizing functions.
848
849=item Running
850
851Now we're finally ready to go: we have compiled Perl byte code, and all
852that's left to do is run it. The actual execution is done by the
853C<runops_standard> function in F<run.c>; more specifically, it's done by
854these three innocent looking lines:
855
856 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
857 PERL_ASYNC_CHECK();
858 }
859
860You may be more comfortable with the Perl version of that:
861
862 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
863
864Well, maybe not. Anyway, each op contains a function pointer, which
865stipulates the function which will actually carry out the operation.
866This function will return the next op in the sequence - this allows for
867things like C<if> which choose the next op dynamically at run time.
868The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
869execution if required.
870
871The actual functions called are known as PP code, and they're spread
b432a672 872between four files: F<pp_hot.c> contains the "hot" code, which is most
a422fd2d
SC
873often used and highly optimized, F<pp_sys.c> contains all the
874system-specific functions, F<pp_ctl.c> contains the functions which
875implement control structures (C<if>, C<while> and the like) and F<pp.c>
876contains everything else. These are, if you like, the C code for Perl's
877built-in functions and operators.
878
dfc98234
DM
879Note that each C<pp_> function is expected to return a pointer to the next
880op. Calls to perl subs (and eval blocks) are handled within the same
881runops loop, and do not consume extra space on the C stack. For example,
882C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
883struct onto the context stack which contain the address of the op
884following the sub call or eval. They then return the first op of that sub
885or eval block, and so execution continues of that sub or block. Later, a
886C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
887retrieves the return op from it, and returns it.
888
889=item Exception handing
890
28a5cf3b 891Perl's exception handing (i.e. C<die> etc) is built on top of the low-level
dfc98234 892C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
28a5cf3b 893way to capture the current PC and SP registers and later restore them; i.e.
dfc98234
DM
894a C<longjmp()> continues at the point in code where a previous C<setjmp()>
895was done, with anything further up on the C stack being lost. This is why
896code should always save values using C<SAVE_FOO> rather than in auto
897variables.
898
899The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
900C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
901C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
902C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
903
904At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
905C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
906loop or whatever, and handle possible exception returns. For a 2 return,
907final cleanup is performed, such as popping stacks and calling C<CHECK> or
908C<END> blocks. Amongst other things, this is how scope cleanup still
909occurs during an C<exit>.
910
911If a C<die> can find a C<CxEVAL> block on the context stack, then the
912stack is popped to that level and the return op in that block is assigned
913to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
914passes control back to the guard. In the case of C<perl_run> and
915C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
916loop. The is the normal way that C<die> or C<croak> is handled within an
917C<eval>.
918
919Sometimes ops are executed within an inner runops loop, such as tie, sort
920or overload code. In this case, something like
921
922 sub FETCH { eval { die } }
923
924would cause a longjmp right back to the guard in C<perl_run>, popping both
925runops loops, which is clearly incorrect. One way to avoid this is for the
926tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
927runops loop, but for efficiency reasons, perl in fact just sets a flag,
928using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
929C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
930which does a C<JMPENV_PUSH> and starts a new runops level to execute the
931code, rather than doing it on the current loop.
932
933As a further optimisation, on exit from the eval block in the C<FETCH>,
934execution of the code following the block is still carried on in the inner
935loop. When an exception is raised, C<docatch> compares the C<JMPENV>
936level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
937re-throws the exception. In this way any inner loops get popped.
938
939Here's an example.
940
941 1: eval { tie @a, 'A' };
942 2: sub A::TIEARRAY {
943 3: eval { die };
944 4: die;
945 5: }
946
947To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
948enters a runops loop. This loop executes the eval and tie ops on line 1,
949with the eval pushing a C<CxEVAL> onto the context stack.
950
951The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
952to execute the body of C<TIEARRAY>. When it executes the entertry op on
953line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
954does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
955the die op. At this point the C call stack looks like this:
956
957 Perl_pp_die
958 Perl_runops # third loop
959 S_docatch_body
960 S_docatch
961 Perl_pp_entertry
962 Perl_runops # second loop
963 S_call_body
964 Perl_call_sv
965 Perl_pp_tie
966 Perl_runops # first loop
967 S_run_body
968 perl_run
969 main
970
971and the context and data stacks, as shown by C<-Dstv>, look like:
972
973 STACK 0: MAIN
974 CX 0: BLOCK =>
975 CX 1: EVAL => AV() PV("A"\0)
976 retop=leave
977 STACK 1: MAGIC
978 CX 0: SUB =>
979 retop=(null)
980 CX 1: EVAL => *
981 retop=nextstate
982
983The die pops the first C<CxEVAL> off the context stack, sets
984C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
985the top C<docatch>. This then starts another third-level runops level,
986which executes the nextstate, pushmark and die ops on line 4. At the point
987that the second C<pp_die> is called, the C call stack looks exactly like
988that above, even though we are no longer within an inner eval; this is
989because of the optimization mentioned earlier. However, the context stack
990now looks like this, ie with the top CxEVAL popped:
991
992 STACK 0: MAIN
993 CX 0: BLOCK =>
994 CX 1: EVAL => AV() PV("A"\0)
995 retop=leave
996 STACK 1: MAGIC
997 CX 0: SUB =>
998 retop=(null)
999
1000The die on line 4 pops the context stack back down to the CxEVAL, leaving
1001it as:
1002
1003 STACK 0: MAIN
1004 CX 0: BLOCK =>
1005
1006As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
1007C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
1008
1009 S_docatch
1010 Perl_pp_entertry
1011 Perl_runops # second loop
1012 S_call_body
1013 Perl_call_sv
1014 Perl_pp_tie
1015 Perl_runops # first loop
1016 S_run_body
1017 perl_run
1018 main
1019
1020In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
1021differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
1022and the C stack unwinds to:
1023
1024 perl_run
1025 main
1026
1027Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
1028and execution continues.
1029
a422fd2d
SC
1030=back
1031
1032=head2 Internal Variable Types
1033
1034You should by now have had a look at L<perlguts>, which tells you about
1035Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
1036that now.
1037
1038These variables are used not only to represent Perl-space variables, but
1039also any constants in the code, as well as some structures completely
1040internal to Perl. The symbol table, for instance, is an ordinary Perl
1041hash. Your code is represented by an SV as it's read into the parser;
1042any program files you call are opened via ordinary Perl filehandles, and
1043so on.
1044
1045The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
1046Perl program. Let's see, for instance, how Perl treats the constant
1047C<"hello">.
1048
1049 % perl -MDevel::Peek -e 'Dump("hello")'
1050 1 SV = PV(0xa041450) at 0xa04ecbc
1051 2 REFCNT = 1
1052 3 FLAGS = (POK,READONLY,pPOK)
1053 4 PV = 0xa0484e0 "hello"\0
1054 5 CUR = 5
1055 6 LEN = 6
1056
1057Reading C<Devel::Peek> output takes a bit of practise, so let's go
1058through it line by line.
1059
1060Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
1061memory. SVs themselves are very simple structures, but they contain a
1062pointer to a more complex structure. In this case, it's a PV, a
1063structure which holds a string value, at location C<0xa041450>. Line 2
1064is the reference count; there are no other references to this data, so
1065it's 1.
1066
1067Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
1068read-only SV (because it's a constant) and the data is a PV internally.
1069Next we've got the contents of the string, starting at location
1070C<0xa0484e0>.
1071
1072Line 5 gives us the current length of the string - note that this does
1073B<not> include the null terminator. Line 6 is not the length of the
1074string, but the length of the currently allocated buffer; as the string
1075grows, Perl automatically extends the available storage via a routine
1076called C<SvGROW>.
1077
1078You can get at any of these quantities from C very easily; just add
1079C<Sv> to the name of the field shown in the snippet, and you've got a
1080macro which will return the value: C<SvCUR(sv)> returns the current
1081length of the string, C<SvREFCOUNT(sv)> returns the reference count,
1082C<SvPV(sv, len)> returns the string itself with its length, and so on.
1083More macros to manipulate these properties can be found in L<perlguts>.
1084
1085Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
1086
1087 1 void
1088 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
1089 3 {
1090 4 STRLEN tlen;
1091 5 char *junk;
1092
1093 6 junk = SvPV_force(sv, tlen);
1094 7 SvGROW(sv, tlen + len + 1);
1095 8 if (ptr == junk)
1096 9 ptr = SvPVX(sv);
1097 10 Move(ptr,SvPVX(sv)+tlen,len,char);
1098 11 SvCUR(sv) += len;
1099 12 *SvEND(sv) = '\0';
1100 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
1101 14 SvTAINT(sv);
1102 15 }
1103
1104This is a function which adds a string, C<ptr>, of length C<len> onto
1105the end of the PV stored in C<sv>. The first thing we do in line 6 is
1106make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
1107macro to force a PV. As a side effect, C<tlen> gets set to the current
1108value of the PV, and the PV itself is returned to C<junk>.
1109
b1866b2d 1110In line 7, we make sure that the SV will have enough room to accommodate
a422fd2d
SC
1111the old string, the new string and the null terminator. If C<LEN> isn't
1112big enough, C<SvGROW> will reallocate space for us.
1113
1114Now, if C<junk> is the same as the string we're trying to add, we can
1115grab the string directly from the SV; C<SvPVX> is the address of the PV
1116in the SV.
1117
1118Line 10 does the actual catenation: the C<Move> macro moves a chunk of
1119memory around: we move the string C<ptr> to the end of the PV - that's
1120the start of the PV plus its current length. We're moving C<len> bytes
1121of type C<char>. After doing so, we need to tell Perl we've extended the
1122string, by altering C<CUR> to reflect the new length. C<SvEND> is a
1123macro which gives us the end of the string, so that needs to be a
1124C<"\0">.
1125
1126Line 13 manipulates the flags; since we've changed the PV, any IV or NV
1127values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
1e54db1a 1128want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
a422fd2d
SC
1129version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
1130and turns on POK. The final C<SvTAINT> is a macro which launders tainted
1131data if taint mode is turned on.
1132
1133AVs and HVs are more complicated, but SVs are by far the most common
1134variable type being thrown around. Having seen something of how we
1135manipulate these, let's go on and look at how the op tree is
1136constructed.
1137
1138=head2 Op Trees
1139
1140First, what is the op tree, anyway? The op tree is the parsed
1141representation of your program, as we saw in our section on parsing, and
1142it's the sequence of operations that Perl goes through to execute your
1143program, as we saw in L</Running>.
1144
1145An op is a fundamental operation that Perl can perform: all the built-in
1146functions and operators are ops, and there are a series of ops which
1147deal with concepts the interpreter needs internally - entering and
1148leaving a block, ending a statement, fetching a variable, and so on.
1149
1150The op tree is connected in two ways: you can imagine that there are two
1151"routes" through it, two orders in which you can traverse the tree.
1152First, parse order reflects how the parser understood the code, and
1153secondly, execution order tells perl what order to perform the
1154operations in.
1155
1156The easiest way to examine the op tree is to stop Perl after it has
1157finished parsing, and get it to dump out the tree. This is exactly what
7d7d5695
RGS
1158the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
1159and L<B::Debug|B::Debug> do.
a422fd2d
SC
1160
1161Let's have a look at how Perl sees C<$a = $b + $c>:
1162
1163 % perl -MO=Terse -e '$a=$b+$c'
1164 1 LISTOP (0x8179888) leave
1165 2 OP (0x81798b0) enter
1166 3 COP (0x8179850) nextstate
1167 4 BINOP (0x8179828) sassign
1168 5 BINOP (0x8179800) add [1]
1169 6 UNOP (0x81796e0) null [15]
1170 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
1171 8 UNOP (0x81797e0) null [15]
1172 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
1173 10 UNOP (0x816b4f0) null [15]
1174 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
1175
1176Let's start in the middle, at line 4. This is a BINOP, a binary
1177operator, which is at location C<0x8179828>. The specific operator in
1178question is C<sassign> - scalar assignment - and you can find the code
1179which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
1180binary operator, it has two children: the add operator, providing the
1181result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
1182line 10.
1183
1184Line 10 is the null op: this does exactly nothing. What is that doing
1185there? If you see the null op, it's a sign that something has been
1186optimized away after parsing. As we mentioned in L</Optimization>,
1187the optimization stage sometimes converts two operations into one, for
1188example when fetching a scalar variable. When this happens, instead of
1189rewriting the op tree and cleaning up the dangling pointers, it's easier
1190just to replace the redundant operation with the null op. Originally,
1191the tree would have looked like this:
1192
1193 10 SVOP (0x816b4f0) rv2sv [15]
1194 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
1195
1196That is, fetch the C<a> entry from the main symbol table, and then look
1197at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
1198happens to do both these things.
1199
1200The right hand side, starting at line 5 is similar to what we've just
1201seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
1202two C<gvsv>s.
1203
1204Now, what's this about?
1205
1206 1 LISTOP (0x8179888) leave
1207 2 OP (0x81798b0) enter
1208 3 COP (0x8179850) nextstate
1209
1210C<enter> and C<leave> are scoping ops, and their job is to perform any
1211housekeeping every time you enter and leave a block: lexical variables
1212are tidied up, unreferenced variables are destroyed, and so on. Every
1213program will have those first three lines: C<leave> is a list, and its
1214children are all the statements in the block. Statements are delimited
1215by C<nextstate>, so a block is a collection of C<nextstate> ops, with
1216the ops to be performed for each statement being the children of
1217C<nextstate>. C<enter> is a single op which functions as a marker.
1218
1219That's how Perl parsed the program, from top to bottom:
1220
1221 Program
1222 |
1223 Statement
1224 |
1225 =
1226 / \
1227 / \
1228 $a +
1229 / \
1230 $b $c
1231
1232However, it's impossible to B<perform> the operations in this order:
1233you have to find the values of C<$b> and C<$c> before you add them
1234together, for instance. So, the other thread that runs through the op
1235tree is the execution order: each op has a field C<op_next> which points
1236to the next op to be run, so following these pointers tells us how perl
1237executes the code. We can traverse the tree in this order using
1238the C<exec> option to C<B::Terse>:
1239
1240 % perl -MO=Terse,exec -e '$a=$b+$c'
1241 1 OP (0x8179928) enter
1242 2 COP (0x81798c8) nextstate
1243 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
1244 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
1245 5 BINOP (0x8179878) add [1]
1246 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
1247 7 BINOP (0x81798a0) sassign
1248 8 LISTOP (0x8179900) leave
1249
1250This probably makes more sense for a human: enter a block, start a
1251statement. Get the values of C<$b> and C<$c>, and add them together.
1252Find C<$a>, and assign one to the other. Then leave.
1253
1254The way Perl builds up these op trees in the parsing process can be
1255unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1256piece we need to construct the tree for C<$a = $b + $c>
1257
1258 1 term : term ASSIGNOP term
1259 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1260 3 | term ADDOP term
1261 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1262
1263If you're not used to reading BNF grammars, this is how it works: You're
1264fed certain things by the tokeniser, which generally end up in upper
1265case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1266code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
b432a672 1267"terminal symbols", because you can't get any simpler than them.
a422fd2d
SC
1268
1269The grammar, lines one and three of the snippet above, tells you how to
b432a672 1270build up more complex forms. These complex forms, "non-terminal symbols"
a422fd2d
SC
1271are generally placed in lower case. C<term> here is a non-terminal
1272symbol, representing a single expression.
1273
1274The grammar gives you the following rule: you can make the thing on the
1275left of the colon if you see all the things on the right in sequence.
1276This is called a "reduction", and the aim of parsing is to completely
1277reduce the input. There are several different ways you can perform a
1278reduction, separated by vertical bars: so, C<term> followed by C<=>
1279followed by C<term> makes a C<term>, and C<term> followed by C<+>
1280followed by C<term> can also make a C<term>.
1281
1282So, if you see two terms with an C<=> or C<+>, between them, you can
1283turn them into a single expression. When you do this, you execute the
1284code in the block on the next line: if you see C<=>, you'll do the code
1285in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1286which contributes to the op tree.
1287
1288 | term ADDOP term
1289 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1290
1291What this does is creates a new binary op, and feeds it a number of
1292variables. The variables refer to the tokens: C<$1> is the first token in
1293the input, C<$2> the second, and so on - think regular expression
1294backreferences. C<$$> is the op returned from this reduction. So, we
1295call C<newBINOP> to create a new binary operator. The first parameter to
1296C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1297operator, so we want the type to be C<ADDOP>. We could specify this
1298directly, but it's right there as the second token in the input, so we
b432a672
AL
1299use C<$2>. The second parameter is the op's flags: 0 means "nothing
1300special". Then the things to add: the left and right hand side of our
a422fd2d
SC
1301expression, in scalar context.
1302
1303=head2 Stacks
1304
1305When perl executes something like C<addop>, how does it pass on its
1306results to the next op? The answer is, through the use of stacks. Perl
1307has a number of stacks to store things it's currently working on, and
1308we'll look at the three most important ones here.
1309
1310=over 3
1311
1312=item Argument stack
1313
1314Arguments are passed to PP code and returned from PP code using the
1315argument stack, C<ST>. The typical way to handle arguments is to pop
1316them off the stack, deal with them how you wish, and then push the result
1317back onto the stack. This is how, for instance, the cosine operator
1318works:
1319
1320 NV value;
1321 value = POPn;
1322 value = Perl_cos(value);
1323 XPUSHn(value);
1324
1325We'll see a more tricky example of this when we consider Perl's macros
1326below. C<POPn> gives you the NV (floating point value) of the top SV on
1327the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1328the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1329should be extended if necessary - it can't be necessary here, because we
1330know there's room for one more item on the stack, since we've just
1331removed one! The C<XPUSH*> macros at least guarantee safety.
1332
1333Alternatively, you can fiddle with the stack directly: C<SP> gives you
1334the first element in your portion of the stack, and C<TOP*> gives you
1335the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1336negation of an integer:
1337
1338 SETi(-TOPi);
1339
1340Just set the integer value of the top stack entry to its negation.
1341
1342Argument stack manipulation in the core is exactly the same as it is in
1343XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1344description of the macros used in stack manipulation.
1345
1346=item Mark stack
1347
b432a672 1348I say "your portion of the stack" above because PP code doesn't
a422fd2d
SC
1349necessarily get the whole stack to itself: if your function calls
1350another function, you'll only want to expose the arguments aimed for the
1351called function, and not (necessarily) let it get at your own data. The
b432a672 1352way we do this is to have a "virtual" bottom-of-stack, exposed to each
a422fd2d
SC
1353function. The mark stack keeps bookmarks to locations in the argument
1354stack usable by each function. For instance, when dealing with a tied
b432a672 1355variable, (internally, something with "P" magic) Perl has to call
a422fd2d
SC
1356methods for accesses to the tied variables. However, we need to separate
1357the arguments exposed to the method to the argument exposed to the
ed233832
DM
1358original function - the store or fetch or whatever it may be. Here's
1359roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
a422fd2d
SC
1360
1361 1 PUSHMARK(SP);
1362 2 EXTEND(SP,2);
1363 3 PUSHs(SvTIED_obj((SV*)av, mg));
1364 4 PUSHs(val);
1365 5 PUTBACK;
1366 6 ENTER;
1367 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1368 8 LEAVE;
13a2d996 1369
a422fd2d
SC
1370Let's examine the whole implementation, for practice:
1371
1372 1 PUSHMARK(SP);
1373
1374Push the current state of the stack pointer onto the mark stack. This is
1375so that when we've finished adding items to the argument stack, Perl
1376knows how many things we've added recently.
1377
1378 2 EXTEND(SP,2);
1379 3 PUSHs(SvTIED_obj((SV*)av, mg));
1380 4 PUSHs(val);
1381
1382We're going to add two more items onto the argument stack: when you have
1383a tied array, the C<PUSH> subroutine receives the object and the value
1384to be pushed, and that's exactly what we have here - the tied object,
1385retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1386
1387 5 PUTBACK;
1388
e89a6d4e
JD
1389Next we tell Perl to update the global stack pointer from our internal
1390variable: C<dSP> only gave us a local copy, not a reference to the global.
a422fd2d
SC
1391
1392 6 ENTER;
1393 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1394 8 LEAVE;
1395
1396C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1397variables are tidied up, everything that has been localised gets
1398its previous value returned, and so on. Think of them as the C<{> and
1399C<}> of a Perl block.
1400
1401To actually do the magic method call, we have to call a subroutine in
1402Perl space: C<call_method> takes care of that, and it's described in
1403L<perlcall>. We call the C<PUSH> method in scalar context, and we're
e89a6d4e
JD
1404going to discard its return value. The call_method() function
1405removes the top element of the mark stack, so there is nothing for
1406the caller to clean up.
a422fd2d 1407
a422fd2d
SC
1408=item Save stack
1409
1410C doesn't have a concept of local scope, so perl provides one. We've
1411seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1412stack implements the C equivalent of, for example:
1413
1414 {
1415 local $foo = 42;
1416 ...
1417 }
1418
1419See L<perlguts/Localising Changes> for how to use the save stack.
1420
1421=back
1422
1423=head2 Millions of Macros
1424
1425One thing you'll notice about the Perl source is that it's full of
1426macros. Some have called the pervasive use of macros the hardest thing
1427to understand, others find it adds to clarity. Let's take an example,
1428the code which implements the addition operator:
1429
1430 1 PP(pp_add)
1431 2 {
39644a26 1432 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1433 4 {
1434 5 dPOPTOPnnrl_ul;
1435 6 SETn( left + right );
1436 7 RETURN;
1437 8 }
1438 9 }
1439
1440Every line here (apart from the braces, of course) contains a macro. The
1441first line sets up the function declaration as Perl expects for PP code;
1442line 3 sets up variable declarations for the argument stack and the
1443target, the return value of the operation. Finally, it tries to see if
1444the addition operation is overloaded; if so, the appropriate subroutine
1445is called.
1446
1447Line 5 is another variable declaration - all variable declarations start
1448with C<d> - which pops from the top of the argument stack two NVs (hence
1449C<nn>) and puts them into the variables C<right> and C<left>, hence the
1450C<rl>. These are the two operands to the addition operator. Next, we
1451call C<SETn> to set the NV of the return value to the result of adding
1452the two values. This done, we return - the C<RETURN> macro makes sure
1453that our return value is properly handled, and we pass the next operator
1454to run back to the main run loop.
1455
1456Most of these macros are explained in L<perlapi>, and some of the more
1457important ones are explained in L<perlxs> as well. Pay special attention
1458to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1459the C<[pad]THX_?> macros.
1460
52d59bef
JH
1461=head2 The .i Targets
1462
1463You can expand the macros in a F<foo.c> file by saying
1464
1465 make foo.i
1466
1467which will expand the macros using cpp. Don't be scared by the results.
1468
955fec6b
JH
1469=head1 SOURCE CODE STATIC ANALYSIS
1470
1471Various tools exist for analysing C source code B<statically>, as
1472opposed to B<dynamically>, that is, without executing the code.
1473It is possible to detect resource leaks, undefined behaviour, type
1474mismatches, portability problems, code paths that would cause illegal
1475memory accesses, and other similar problems by just parsing the C code
1476and looking at the resulting graph, what does it tell about the
1477execution and data flows. As a matter of fact, this is exactly
1478how C compilers know to give warnings about dubious code.
1479
1480=head2 lint, splint
1481
1482The good old C code quality inspector, C<lint>, is available in
1483several platforms, but please be aware that there are several
1484different implementations of it by different vendors, which means that
1485the flags are not identical across different platforms.
1486
1487There is a lint variant called C<splint> (Secure Programming Lint)
1488available from http://www.splint.org/ that should compile on any
1489Unix-like platform.
1490
1491There are C<lint> and <splint> targets in Makefile, but you may have
1492to diddle with the flags (see above).
1493
1494=head2 Coverity
1495
1496Coverity (http://www.coverity.com/) is a product similar to lint and
1497as a testbed for their product they periodically check several open
1498source projects, and they give out accounts to open source developers
1499to the defect databases.
1500
1501=head2 cpd (cut-and-paste detector)
1502
1503The cpd tool detects cut-and-paste coding. If one instance of the
1504cut-and-pasted code changes, all the other spots should probably be
1505changed, too. Therefore such code should probably be turned into a
1506subroutine or a macro.
1507
1508cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1509(http://pmd.sourceforge.net/). pmd was originally written for static
1510analysis of Java code, but later the cpd part of it was extended to
1511parse also C and C++.
1512
a52aaefa
RGS
1513Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1514pmd-X.Y.jar from it, and then run that on source code thusly:
955fec6b
JH
1515
1516 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1517
1518You may run into memory limits, in which case you should use the -Xmx option:
1519
1520 java -Xmx512M ...
1521
1522=head2 gcc warnings
1523
1524Though much can be written about the inconsistency and coverage
1525problems of gcc warnings (like C<-Wall> not meaning "all the
1526warnings", or some common portability problems not being covered by
1527C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1528collection of warnings, and so forth), gcc is still a useful tool in
1529keeping our coding nose clean.
1530
1531The C<-Wall> is by default on.
1532
a8e98a71
JH
1533The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1534always, but unfortunately they are not safe on all platforms, they can
1535for example cause fatal conflicts with the system headers (Solaris
1536being a prime example). If Configure C<-Dgccansipedantic> is used,
1537the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1538where they are known to be safe.
955fec6b
JH
1539
1540Starting from Perl 5.9.4 the following extra flags are added:
1541
1542=over 4
1543
1544=item *
1545
1546C<-Wendif-labels>
1547
1548=item *
1549
1550C<-Wextra>
1551
1552=item *
1553
1554C<-Wdeclaration-after-statement>
1555
1556=back
1557
1558The following flags would be nice to have but they would first need
1559their own Stygian stablemaster:
1560
1561=over 4
1562
1563=item *
1564
1565C<-Wpointer-arith>
1566
1567=item *
1568
1569C<-Wshadow>
1570
1571=item *
1572
1573C<-Wstrict-prototypes>
1574
955fec6b
JH
1575=back
1576
1577The C<-Wtraditional> is another example of the annoying tendency of
1578gcc to bundle a lot of warnings under one switch -- it would be
1579impossible to deploy in practice because it would complain a lot -- but
1580it does contain some warnings that would be beneficial to have available
1581on their own, such as the warning about string constants inside macros
1582containing the macro arguments: this behaved differently pre-ANSI
1583than it does in ANSI, and some C compilers are still in transition,
1584AIX being an example.
1585
1586=head2 Warnings of other C compilers
1587
1588Other C compilers (yes, there B<are> other C compilers than gcc) often
1589have their "strict ANSI" or "strict ANSI with some portability extensions"
1590modes on, like for example the Sun Workshop has its C<-Xa> mode on
1591(though implicitly), or the DEC (these days, HP...) has its C<-std1>
1592mode on.
1593
1594=head2 DEBUGGING
1595
1596You can compile a special debugging version of Perl, which allows you
1597to use the C<-D> option of Perl to tell more about what Perl is doing.
1598But sometimes there is no alternative than to dive in with a debugger,
1599either to see the stack trace of a core dump (very useful in a bug
1600report), or trying to figure out what went wrong before the core dump
1601happened, or how did we end up having wrong or unexpected results.
1602
a422fd2d
SC
1603=head2 Poking at Perl
1604
1605To really poke around with Perl, you'll probably want to build Perl for
1606debugging, like this:
1607
1608 ./Configure -d -D optimize=-g
1609 make
1610
1611C<-g> is a flag to the C compiler to have it produce debugging
955fec6b
JH
1612information which will allow us to step through a running program,
1613and to see in which C function we are at (without the debugging
1614information we might see only the numerical addresses of the functions,
1615which is not very helpful).
1616
a422fd2d
SC
1617F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1618enables all the internal debugging code in Perl. There are a whole bunch
1619of things you can debug with this: L<perlrun> lists them all, and the
1620best way to find out about them is to play about with them. The most
1621useful options are probably
1622
1623 l Context (loop) stack processing
1624 t Trace execution
1625 o Method and overloading resolution
1626 c String/numeric conversions
1627
1628Some of the functionality of the debugging code can be achieved using XS
1629modules.
13a2d996 1630
a422fd2d
SC
1631 -Dr => use re 'debug'
1632 -Dx => use O 'Debug'
1633
1634=head2 Using a source-level debugger
1635
1636If the debugging output of C<-D> doesn't help you, it's time to step
1637through perl's execution with a source-level debugger.
1638
1639=over 3
1640
1641=item *
1642
955fec6b
JH
1643We'll use C<gdb> for our examples here; the principles will apply to
1644any debugger (many vendors call their debugger C<dbx>), but check the
1645manual of the one you're using.
a422fd2d
SC
1646
1647=back
1648
1649To fire up the debugger, type
1650
1651 gdb ./perl
1652
955fec6b
JH
1653Or if you have a core dump:
1654
1655 gdb ./perl core
1656
a422fd2d
SC
1657You'll want to do that in your Perl source tree so the debugger can read
1658the source code. You should see the copyright message, followed by the
1659prompt.
1660
1661 (gdb)
1662
1663C<help> will get you into the documentation, but here are the most
1664useful commands:
1665
1666=over 3
1667
1668=item run [args]
1669
1670Run the program with the given arguments.
1671
1672=item break function_name
1673
1674=item break source.c:xxx
1675
1676Tells the debugger that we'll want to pause execution when we reach
cea6626f 1677either the named function (but see L<perlguts/Internal Functions>!) or the given
a422fd2d
SC
1678line in the named source file.
1679
1680=item step
1681
1682Steps through the program a line at a time.
1683
1684=item next
1685
1686Steps through the program a line at a time, without descending into
1687functions.
1688
1689=item continue
1690
1691Run until the next breakpoint.
1692
1693=item finish
1694
1695Run until the end of the current function, then stop again.
1696
13a2d996 1697=item 'enter'
a422fd2d
SC
1698
1699Just pressing Enter will do the most recent operation again - it's a
1700blessing when stepping through miles of source code.
1701
1702=item print
1703
1704Execute the given C code and print its results. B<WARNING>: Perl makes
52d59bef
JH
1705heavy use of macros, and F<gdb> does not necessarily support macros
1706(see later L</"gdb macro support">). You'll have to substitute them
1707yourself, or to invoke cpp on the source code files
1708(see L</"The .i Targets">)
1709So, for instance, you can't say
a422fd2d
SC
1710
1711 print SvPV_nolen(sv)
1712
1713but you have to say
1714
1715 print Perl_sv_2pv_nolen(sv)
1716
ffc145e8
RK
1717=back
1718
a422fd2d
SC
1719You may find it helpful to have a "macro dictionary", which you can
1720produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
07aa3531 1721recursively apply those macros for you.
52d59bef
JH
1722
1723=head2 gdb macro support
a422fd2d 1724
52d59bef 1725Recent versions of F<gdb> have fairly good macro support, but
ea031e66
RGS
1726in order to use it you'll need to compile perl with macro definitions
1727included in the debugging information. Using F<gcc> version 3.1, this
1728means configuring with C<-Doptimize=-g3>. Other compilers might use a
1729different switch (if they support debugging macros at all).
1730
a422fd2d
SC
1731=head2 Dumping Perl Data Structures
1732
1733One way to get around this macro hell is to use the dumping functions in
1734F<dump.c>; these work a little like an internal
1735L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1736that you can't get at from Perl. Let's take an example. We'll use the
07aa3531 1737C<$a = $b + $c> we used before, but give it a bit of context:
a422fd2d
SC
1738C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1739
1740What about C<pp_add>, the function we examined earlier to implement the
1741C<+> operator:
1742
1743 (gdb) break Perl_pp_add
1744 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1745
cea6626f 1746Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
a422fd2d
SC
1747With the breakpoint in place, we can run our program:
1748
1749 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1750
1751Lots of junk will go past as gdb reads in the relevant source files and
1752libraries, and then:
1753
1754 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
39644a26 1755 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
a422fd2d
SC
1756 (gdb) step
1757 311 dPOPTOPnnrl_ul;
1758 (gdb)
1759
1760We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1761arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1762slightly expand it:
1763
1764 #define dPOPTOPnnrl_ul NV right = POPn; \
1765 SV *leftsv = TOPs; \
1766 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1767
1768C<POPn> takes the SV from the top of the stack and obtains its NV either
1769directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1770C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1771C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
07aa3531 1772C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
a422fd2d
SC
1773
1774Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1775convert it. If we step again, we'll find ourselves there:
1776
1777 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1778 1669 if (!sv)
1779 (gdb)
1780
1781We can now use C<Perl_sv_dump> to investigate the SV:
1782
1783 SV = PV(0xa057cc0) at 0xa0675d0
1784 REFCNT = 1
1785 FLAGS = (POK,pPOK)
1786 PV = 0xa06a510 "6XXXX"\0
1787 CUR = 5
1788 LEN = 6
1789 $1 = void
1790
1791We know we're going to get C<6> from this, so let's finish the
1792subroutine:
1793
1794 (gdb) finish
1795 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1796 0x462669 in Perl_pp_add () at pp_hot.c:311
1797 311 dPOPTOPnnrl_ul;
1798
1799We can also dump out this op: the current op is always stored in
1800C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1801similar output to L<B::Debug|B::Debug>.
1802
1803 {
1804 13 TYPE = add ===> 14
1805 TARG = 1
1806 FLAGS = (SCALAR,KIDS)
1807 {
1808 TYPE = null ===> (12)
1809 (was rv2sv)
1810 FLAGS = (SCALAR,KIDS)
1811 {
1812 11 TYPE = gvsv ===> 12
1813 FLAGS = (SCALAR)
1814 GV = main::b
1815 }
1816 }
1817
10f58044 1818# finish this later #
a422fd2d
SC
1819
1820=head2 Patching
1821
1822All right, we've now had a look at how to navigate the Perl sources and
1823some things you'll need to know when fiddling with them. Let's now get
1824on and create a simple patch. Here's something Larry suggested: if a
07aa3531 1825C<U> is the first active format during a C<pack>, (for example,
a422fd2d 1826C<pack "U3C8", @stuff>) then the resulting string should be treated as
1e54db1a 1827UTF-8 encoded.
a422fd2d
SC
1828
1829How do we prepare to fix this up? First we locate the code in question -
1830the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1831files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1832altering this file, let's copy it to F<pp.c~>.
1833
a6ec74c1
JH
1834[Well, it was in F<pp.c> when this tutorial was written. It has now been
1835split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1836
a422fd2d
SC
1837Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1838loop over the pattern, taking each format character in turn into
1839C<datum_type>. Then for each possible format character, we swallow up
1840the other arguments in the pattern (a field width, an asterisk, and so
1841on) and convert the next chunk input into the specified format, adding
1842it onto the output SV C<cat>.
1843
1844How do we know if the C<U> is the first format in the C<pat>? Well, if
1845we have a pointer to the start of C<pat> then, if we see a C<U> we can
1846test whether we're still at the start of the string. So, here's where
1847C<pat> is set up:
1848
1849 STRLEN fromlen;
1850 register char *pat = SvPVx(*++MARK, fromlen);
1851 register char *patend = pat + fromlen;
1852 register I32 len;
1853 I32 datumtype;
1854 SV *fromstr;
1855
1856We'll have another string pointer in there:
1857
1858 STRLEN fromlen;
1859 register char *pat = SvPVx(*++MARK, fromlen);
1860 register char *patend = pat + fromlen;
1861 + char *patcopy;
1862 register I32 len;
1863 I32 datumtype;
1864 SV *fromstr;
1865
1866And just before we start the loop, we'll set C<patcopy> to be the start
1867of C<pat>:
1868
1869 items = SP - MARK;
1870 MARK++;
1871 sv_setpvn(cat, "", 0);
1872 + patcopy = pat;
1873 while (pat < patend) {
1874
1875Now if we see a C<U> which was at the start of the string, we turn on
1e54db1a 1876the C<UTF8> flag for the output SV, C<cat>:
a422fd2d
SC
1877
1878 + if (datumtype == 'U' && pat==patcopy+1)
1879 + SvUTF8_on(cat);
1880 if (datumtype == '#') {
1881 while (pat < patend && *pat != '\n')
1882 pat++;
1883
1884Remember that it has to be C<patcopy+1> because the first character of
1885the string is the C<U> which has been swallowed into C<datumtype!>
1886
1887Oops, we forgot one thing: what if there are spaces at the start of the
1888pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1889character, even though it's not the first thing in the pattern. In this
1890case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1891
1892 if (isSPACE(datumtype))
1893 continue;
1894
1895needs to become
1896
1897 if (isSPACE(datumtype)) {
1898 patcopy++;
1899 continue;
1900 }
1901
1902OK. That's the C part done. Now we must do two additional things before
1903this patch is ready to go: we've changed the behaviour of Perl, and so
1904we must document that change. We must also provide some more regression
1905tests to make sure our patch works and doesn't create a bug somewhere
1906else along the line.
1907
b23b8711
MS
1908The regression tests for each operator live in F<t/op/>, and so we
1909make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1910tests to the end. First, we'll test that the C<U> does indeed create
07aa3531 1911Unicode strings.
b23b8711
MS
1912
1913t/op/pack.t has a sensible ok() function, but if it didn't we could
35c336e6 1914use the one from t/test.pl.
b23b8711 1915
35c336e6
MS
1916 require './test.pl';
1917 plan( tests => 159 );
b23b8711
MS
1918
1919so instead of this:
a422fd2d
SC
1920
1921 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1922 print "ok $test\n"; $test++;
1923
35c336e6
MS
1924we can write the more sensible (see L<Test::More> for a full
1925explanation of is() and other testing functions).
b23b8711 1926
07aa3531 1927 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
38a44b82 1928 "U* produces Unicode" );
b23b8711 1929
a422fd2d
SC
1930Now we'll test that we got that space-at-the-beginning business right:
1931
35c336e6 1932 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
812f5127 1933 " with spaces at the beginning" );
a422fd2d
SC
1934
1935And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1936the first active format:
1937
35c336e6 1938 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
38a44b82 1939 "U* not first isn't Unicode" );
a422fd2d 1940
35c336e6
MS
1941Mustn't forget to change the number of tests which appears at the top,
1942or else the automated tester will get confused. This will either look
1943like this:
a422fd2d 1944
35c336e6
MS
1945 print "1..156\n";
1946
1947or this:
1948
1949 plan( tests => 156 );
a422fd2d
SC
1950
1951We now compile up Perl, and run it through the test suite. Our new
1952tests pass, hooray!
1953
1954Finally, the documentation. The job is never done until the paperwork is
1955over, so let's describe the change we've just made. The relevant place
1956is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1957this text in the description of C<pack>:
1958
1959 =item *
1960
1961 If the pattern begins with a C<U>, the resulting string will be treated
1e54db1a
JH
1962 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1963 with an initial C<U0>, and the bytes that follow will be interpreted as
1964 Unicode characters. If you don't want this to happen, you can begin your
1965 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
a422fd2d
SC
1966 string, and then follow this with a C<U*> somewhere in your pattern.
1967
1968All done. Now let's create the patch. F<Porting/patching.pod> tells us
1969that if we're making major changes, we should copy the entire directory
1970to somewhere safe before we begin fiddling, and then do
13a2d996 1971
a422fd2d
SC
1972 diff -ruN old new > patch
1973
1974However, we know which files we've changed, and we can simply do this:
1975
1976 diff -u pp.c~ pp.c > patch
1977 diff -u t/op/pack.t~ t/op/pack.t >> patch
1978 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1979
1980We end up with a patch looking a little like this:
1981
1982 --- pp.c~ Fri Jun 02 04:34:10 2000
1983 +++ pp.c Fri Jun 16 11:37:25 2000
1984 @@ -4375,6 +4375,7 @@
1985 register I32 items;
1986 STRLEN fromlen;
1987 register char *pat = SvPVx(*++MARK, fromlen);
1988 + char *patcopy;
1989 register char *patend = pat + fromlen;
1990 register I32 len;
1991 I32 datumtype;
1992 @@ -4405,6 +4406,7 @@
1993 ...
1994
1995And finally, we submit it, with our rationale, to perl5-porters. Job
1996done!
1997
f7e1e956
MS
1998=head2 Patching a core module
1999
2000This works just like patching anything else, with an extra
2001consideration. Many core modules also live on CPAN. If this is so,
2002patch the CPAN version instead of the core and send the patch off to
2003the module maintainer (with a copy to p5p). This will help the module
2004maintainer keep the CPAN version in sync with the core version without
2005constantly scanning p5p.
2006
db300100
RGS
2007The list of maintainers of core modules is usefully documented in
2008F<Porting/Maintainers.pl>.
2009
acbe17fc
JP
2010=head2 Adding a new function to the core
2011
2012If, as part of a patch to fix a bug, or just because you have an
2013especially good idea, you decide to add a new function to the core,
2014discuss your ideas on p5p well before you start work. It may be that
2015someone else has already attempted to do what you are considering and
2016can give lots of good advice or even provide you with bits of code
2017that they already started (but never finished).
2018
2019You have to follow all of the advice given above for patching. It is
2020extremely important to test any addition thoroughly and add new tests
2021to explore all boundary conditions that your new function is expected
2022to handle. If your new function is used only by one module (e.g. toke),
2023then it should probably be named S_your_function (for static); on the
210b36aa 2024other hand, if you expect it to accessible from other functions in
acbe17fc
JP
2025Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
2026for more details.
2027
2028The location of any new code is also an important consideration. Don't
2029just create a new top level .c file and put your code there; you would
2030have to make changes to Configure (so the Makefile is created properly),
2031as well as possibly lots of include files. This is strictly pumpking
2032business.
2033
2034It is better to add your function to one of the existing top level
2035source code files, but your choice is complicated by the nature of
2036the Perl distribution. Only the files that are marked as compiled
2037static are located in the perl executable. Everything else is located
2038in the shared library (or DLL if you are running under WIN32). So,
2039for example, if a function was only used by functions located in
2040toke.c, then your code can go in toke.c. If, however, you want to call
2041the function from universal.c, then you should put your code in another
2042location, for example util.c.
2043
2044In addition to writing your c-code, you will need to create an
2045appropriate entry in embed.pl describing your function, then run
2046'make regen_headers' to create the entries in the numerous header
2047files that perl needs to compile correctly. See L<perlguts/Internal Functions>
2048for information on the various options that you can set in embed.pl.
2049You will forget to do this a few (or many) times and you will get
2050warnings during the compilation phase. Make sure that you mention
2051this when you post your patch to P5P; the pumpking needs to know this.
2052
2053When you write your new code, please be conscious of existing code
884bad00 2054conventions used in the perl source files. See L<perlstyle> for
acbe17fc
JP
2055details. Although most of the guidelines discussed seem to focus on
2056Perl code, rather than c, they all apply (except when they don't ;).
2057See also I<Porting/patching.pod> file in the Perl source distribution
2058for lots of details about both formatting and submitting patches of
2059your changes.
2060
2061Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
2062Test on as many platforms as you can find. Test as many perl
2063Configure options as you can (e.g. MULTIPLICITY). If you have
2064profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
210b36aa 2065below for how to use them to further test your code. Remember that
acbe17fc
JP
2066most of the people on P5P are doing this on their own time and
2067don't have the time to debug your code.
f7e1e956
MS
2068
2069=head2 Writing a test
2070
2071Every module and built-in function has an associated test file (or
2072should...). If you add or change functionality, you have to write a
2073test. If you fix a bug, you have to write a test so that bug never
2074comes back. If you alter the docs, it would be nice to test what the
2075new documentation says.
2076
2077In short, if you submit a patch you probably also have to patch the
2078tests.
2079
2080For modules, the test file is right next to the module itself.
2081F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
2082so there are some snags (and it would be wonderful for you to brush
2083them out), but it basically works that way. Everything else lives in
2084F<t/>.
2085
2086=over 3
2087
2088=item F<t/base/>
2089
2090Testing of the absolute basic functionality of Perl. Things like
2091C<if>, basic file reads and writes, simple regexes, etc. These are
2092run first in the test suite and if any of them fail, something is
2093I<really> broken.
2094
2095=item F<t/cmd/>
2096
2097These test the basic control structures, C<if/else>, C<while>,
35c336e6 2098subroutines, etc.
f7e1e956
MS
2099
2100=item F<t/comp/>
2101
2102Tests basic issues of how Perl parses and compiles itself.
2103
2104=item F<t/io/>
2105
2106Tests for built-in IO functions, including command line arguments.
2107
2108=item F<t/lib/>
2109
2110The old home for the module tests, you shouldn't put anything new in
2111here. There are still some bits and pieces hanging around in here
2112that need to be moved. Perhaps you could move them? Thanks!
2113
2114=item F<t/op/>
2115
2116Tests for perl's built in functions that don't fit into any of the
2117other directories.
2118
2119=item F<t/pod/>
2120
2121Tests for POD directives. There are still some tests for the Pod
2122modules hanging around in here that need to be moved out into F<lib/>.
2123
2124=item F<t/run/>
2125
2126Testing features of how perl actually runs, including exit codes and
2127handling of PERL* environment variables.
2128
244d9cb7
RGS
2129=item F<t/uni/>
2130
2131Tests for the core support of Unicode.
2132
2133=item F<t/win32/>
2134
2135Windows-specific tests.
2136
2137=item F<t/x2p>
2138
2139A test suite for the s2p converter.
2140
f7e1e956
MS
2141=back
2142
2143The core uses the same testing style as the rest of Perl, a simple
2144"ok/not ok" run through Test::Harness, but there are a few special
2145considerations.
2146
35c336e6
MS
2147There are three ways to write a test in the core. Test::More,
2148t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
2149decision of which to use depends on what part of the test suite you're
2150working on. This is a measure to prevent a high-level failure (such
2151as Config.pm breaking) from causing basic functionality tests to fail.
2152
07aa3531 2153=over 4
35c336e6
MS
2154
2155=item t/base t/comp
2156
2157Since we don't know if require works, or even subroutines, use ad hoc
2158tests for these two. Step carefully to avoid using the feature being
2159tested.
2160
2161=item t/cmd t/run t/io t/op
2162
2163Now that basic require() and subroutines are tested, you can use the
2164t/test.pl library which emulates the important features of Test::More
2165while using a minimum of core features.
2166
2167You can also conditionally use certain libraries like Config, but be
2168sure to skip the test gracefully if it's not there.
2169
2170=item t/lib ext lib
2171
2172Now that the core of Perl is tested, Test::More can be used. You can
2173also use the full suite of core modules in the tests.
2174
2175=back
f7e1e956
MS
2176
2177When you say "make test" Perl uses the F<t/TEST> program to run the
07aa3531
JC
2178test suite (except under Win32 where it uses F<t/harness> instead.)
2179All tests are run from the F<t/> directory, B<not> the directory
2180which contains the test. This causes some problems with the tests
7205a85d 2181in F<lib/>, so here's some opportunity for some patching.
f7e1e956
MS
2182
2183You must be triply conscious of cross-platform concerns. This usually
2184boils down to using File::Spec and avoiding things like C<fork()> and
2185C<system()> unless absolutely necessary.
2186
e018f8be
JH
2187=head2 Special Make Test Targets
2188
2189There are various special make targets that can be used to test Perl
2190slightly differently than the standard "test" target. Not all them
2191are expected to give a 100% success rate. Many of them have several
7205a85d
YO
2192aliases, and many of them are not available on certain operating
2193systems.
e018f8be
JH
2194
2195=over 4
2196
2197=item coretest
2198
7d7d5695 2199Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
e018f8be 2200
7205a85d
YO
2201(Not available on Win32)
2202
e018f8be
JH
2203=item test.deparse
2204
b26492ee
RGS
2205Run all the tests through B::Deparse. Not all tests will succeed.
2206
7205a85d
YO
2207(Not available on Win32)
2208
b26492ee
RGS
2209=item test.taintwarn
2210
2211Run all tests with the B<-t> command-line switch. Not all tests
2212are expected to succeed (until they're specifically fixed, of course).
e018f8be 2213
7205a85d
YO
2214(Not available on Win32)
2215
e018f8be
JH
2216=item minitest
2217
2218Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
2219F<t/op>, and F<t/uni> tests.
2220
7a834142
JH
2221=item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
2222
2223(Only in Linux) Run all the tests using the memory leak + naughty
2224memory access tool "valgrind". The log files will be named
2225F<testname.valgrind>.
2226
e018f8be
JH
2227=item test.third check.third utest.third ucheck.third
2228
2229(Only in Tru64) Run all the tests using the memory leak + naughty
2230memory access tool "Third Degree". The log files will be named
60a57c1c 2231F<perl.3log.testname>.
e018f8be
JH
2232
2233=item test.torture torturetest
2234
2235Run all the usual tests and some extra tests. As of Perl 5.8.0 the
244d9cb7 2236only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
e018f8be
JH
2237
2238You can also run the torture test with F<t/harness> by giving
2239C<-torture> argument to F<t/harness>.
2240
2241=item utest ucheck test.utf8 check.utf8
2242
2243Run all the tests with -Mutf8. Not all tests will succeed.
2244
7205a85d
YO
2245(Not available on Win32)
2246
cc0710ff
RGS
2247=item minitest.utf16 test.utf16
2248
2249Runs the tests with UTF-16 encoded scripts, encoded with different
2250versions of this encoding.
2251
2252C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
2253C<-utf16> arguments to F<t/TEST>.
2254
7205a85d
YO
2255(Not available on Win32)
2256
244d9cb7
RGS
2257=item test_harness
2258
2259Run the test suite with the F<t/harness> controlling program, instead of
2260F<t/TEST>. F<t/harness> is more sophisticated, and uses the
2261L<Test::Harness> module, thus using this test target supposes that perl
2262mostly works. The main advantage for our purposes is that it prints a
00bf5cd9
RGS
2263detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
2264doesn't redirect stderr to stdout.
244d9cb7 2265
7205a85d
YO
2266Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
2267there is no special "test_harness" target.
2268
2269Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
2270environment variables to control the behaviour of F<t/harness>. This means
2271you can say
2272
2273 nmake test TEST_FILES="op/*.t"
2274 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
2275
2276=item test-notty test_notty
2277
2278Sets PERL_SKIP_TTY_TEST to true before running normal test.
2279
244d9cb7
RGS
2280=back
2281
2282=head2 Running tests by hand
2283
2284You can run part of the test suite by hand by using one the following
2285commands from the F<t/> directory :
2286
2287 ./perl -I../lib TEST list-of-.t-files
2288
2289or
2290
2291 ./perl -I../lib harness list-of-.t-files
2292
2293(if you don't specify test scripts, the whole test suite will be run.)
2294
7205a85d
YO
2295=head3 Using t/harness for testing
2296
2297If you use C<harness> for testing you have several command line options
2298available to you. The arguments are as follows, and are in the order
2299that they must appear if used together.
2300
2301 harness -v -torture -re=pattern LIST OF FILES TO TEST
2302 harness -v -torture -re LIST OF PATTERNS TO MATCH
2303
2304If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
07aa3531 2305the manifest. The file list may include shell wildcards which will be
7205a85d
YO
2306expanded out.
2307
2308=over 4
2309
2310=item -v
2311
07aa3531 2312Run the tests under verbose mode so you can see what tests were run,
7205a85d
YO
2313and debug outbut.
2314
2315=item -torture
2316
2317Run the torture tests as well as the normal set.
2318
2319=item -re=PATTERN
2320
2321Filter the file list so that all the test files run match PATTERN.
2322Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2323in that it allows the file list to be provided as well.
2324
2325=item -re LIST OF PATTERNS
2326
07aa3531 2327Filter the file list so that all the test files run match
7205a85d
YO
2328/(LIST|OF|PATTERNS)/. Note that with this form the patterns
2329are joined by '|' and you cannot supply a list of files, instead
2330the test files are obtained from the MANIFEST.
2331
2332=back
2333
244d9cb7
RGS
2334You can run an individual test by a command similar to
2335
2336 ./perl -I../lib patho/to/foo.t
2337
2338except that the harnesses set up some environment variables that may
2339affect the execution of the test :
2340
07aa3531 2341=over 4
244d9cb7
RGS
2342
2343=item PERL_CORE=1
2344
2345indicates that we're running this test part of the perl core test suite.
2346This is useful for modules that have a dual life on CPAN.
2347
2348=item PERL_DESTRUCT_LEVEL=2
2349
2350is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2351
2352=item PERL
2353
2354(used only by F<t/TEST>) if set, overrides the path to the perl executable
2355that should be used to run the tests (the default being F<./perl>).
2356
2357=item PERL_SKIP_TTY_TEST
2358
2359if set, tells to skip the tests that need a terminal. It's actually set
2360automatically by the Makefile, but can also be forced artificially by
2361running 'make test_notty'.
2362
e018f8be 2363=back
f7e1e956 2364
d7889f52
JH
2365=head2 Common problems when patching Perl source code
2366
2367Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2368some cases we have to take pre-ANSI requirements into consideration.
2369You don't care about some particular platform having broken Perl?
2370I hear there is still a strong demand for J2EE programmers.
2371
2372=head2 Perl environment problems
2373
2374=over 4
2375
2376=item *
2377
2378Not compiling with threading
2379
2380Compiling with threading (-Duseithreads) completely rewrites
2381the function prototypes of Perl. You better try your changes
0bec6c03 2382with that. Related to this is the difference between "Perl_-less"
d7889f52
JH
2383and "Perl_-ly" APIs, for example:
2384
2385 Perl_sv_setiv(aTHX_ ...);
2386 sv_setiv(...);
2387
ee9468a2
RGS
2388The first one explicitly passes in the context, which is needed for e.g.
2389threaded builds. The second one does that implicitly; do not get them
def4ed7d
JH
2390mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2391(or a dVAR) as the first thing in the function.
d7889f52
JH
2392
2393See L<perlguts/"How multiple interpreters and concurrency are supported">
2394for further discussion about context.
2395
2396=item *
2397
2398Not compiling with -DDEBUGGING
2399
2400The DEBUGGING define exposes more code to the compiler,
0bec6c03 2401therefore more ways for things to go wrong. You should try it.
d7889f52
JH
2402
2403=item *
2404
ee9468a2
RGS
2405Introducing (non-read-only) globals
2406
2407Do not introduce any modifiable globals, truly global or file static.
bc028b6b
JH
2408They are bad form and complicate multithreading and other forms of
2409concurrency. The right way is to introduce them as new interpreter
2410variables, see F<intrpvar.h> (at the very end for binary compatibility).
ee9468a2
RGS
2411
2412Introducing read-only (const) globals is okay, as long as you verify
2413with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2414BSD-style output) that the data you added really is read-only.
2415(If it is, it shouldn't show up in the output of that command.)
2416
2417If you want to have static strings, make them constant:
2418
2419 static const char etc[] = "...";
2420
bc028b6b 2421If you want to have arrays of constant strings, note carefully
ee9468a2
RGS
2422the right combination of C<const>s:
2423
2424 static const char * const yippee[] =
2425 {"hi", "ho", "silver"};
2426
bc028b6b
JH
2427There is a way to completely hide any modifiable globals (they are all
2428moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2429It is not normally used, but can be used for testing, read more
def4ed7d 2430about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
bc028b6b 2431
ee9468a2
RGS
2432=item *
2433
d7889f52
JH
2434Not exporting your new function
2435
2436Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2437function that is part of the public API (the shared Perl library)
2438to be explicitly marked as exported. See the discussion about
2439F<embed.pl> in L<perlguts>.
2440
2441=item *
2442
2443Exporting your new function
2444
2445The new shiny result of either genuine new functionality or your
2446arduous refactoring is now ready and correctly exported. So what
def4ed7d 2447could possibly go wrong?
d7889f52
JH
2448
2449Maybe simply that your function did not need to be exported in the
2450first place. Perl has a long and not so glorious history of exporting
2451functions that it should not have.
2452
2453If the function is used only inside one source code file, make it
2454static. See the discussion about F<embed.pl> in L<perlguts>.
2455
2456If the function is used across several files, but intended only for
2457Perl's internal use (and this should be the common case), do not
2458export it to the public API. See the discussion about F<embed.pl>
2459in L<perlguts>.
2460
2461=back
2462
5b38b9cd 2463=head2 Portability problems
d7889f52
JH
2464
2465The following are common causes of compilation and/or execution
2466failures, not common to Perl as such. The C FAQ is good bedtime
0bec6c03
JH
2467reading. Please test your changes with as many C compilers and
2468platforms as possible -- we will, anyway, and it's nice to save
2469oneself from public embarrassment.
2470
9aaf14db
RGS
2471If using gcc, you can add the C<-std=c89> option which will hopefully
2472catch most of these unportabilities. (However it might also catch
2473incompatibilities in your system's header files.)
d1307786 2474
a8e98a71
JH
2475Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2476C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2477
def4ed7d
JH
2478If using the C<gcc -Wall> note that not all the possible warnings
2479(like C<-Wunitialized>) are given unless you also compile with C<-O>.
2480
2481Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2482code files (the ones at the top level of the source code distribution,
2483but not e.g. the extensions under ext/) are automatically compiled
2484with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2485and a selection of C<-W> flags (see cflags.SH).
27565cb6 2486
0bec6c03 2487Also study L<perlport> carefully to avoid any bad assumptions
def4ed7d 2488about the operating system, filesystems, and so forth.
0bec6c03 2489
606fd33d 2490You may once in a while try a "make microperl" to see whether we
63796a85 2491can still compile Perl with just the bare minimum of interfaces.
606fd33d 2492(See README.micro.)
ee9468a2 2493
0bec6c03 2494Do not assume an operating system indicates a certain compiler.
d7889f52
JH
2495
2496=over 4
2497
2498=item *
2499
2500Casting pointers to integers or casting integers to pointers
2501
2502 void castaway(U8* p)
2503 {
2504 IV i = p;
2505
2506or
2507
2508 void castaway(U8* p)
2509 {
2510 IV i = (IV)p;
2511
ee9468a2 2512Both are bad, and broken, and unportable. Use the PTR2IV()
d7889f52
JH
2513macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2514INT2PTR(), and NUM2PTR().)
2515
2516=item *
2517
0bec6c03
JH
2518Casting between data function pointers and data pointers
2519
d7889f52
JH
2520Technically speaking casting between function pointers and data
2521pointers is unportable and undefined, but practically speaking
2522it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
0bec6c03 2523macros. Sometimes you can also play games with unions.
d7889f52
JH
2524
2525=item *
2526
2527Assuming sizeof(int) == sizeof(long)
2528
2529There are platforms where longs are 64 bits, and platforms where ints
2530are 64 bits, and while we are out to shock you, even platforms where
2531shorts are 64 bits. This is all legal according to the C standard.
2532(In other words, "long long" is not a portable way to specify 64 bits,
2533and "long long" is not even guaranteed to be any wider than "long".)
63796a85
JH
2534
2535Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2536Avoid things like I32 because they are B<not> guaranteed to be
2537I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2538guaranteed to be B<int> or B<long>. If you really explicitly need
253964-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
d7889f52
JH
2540
2541=item *
2542
2543Assuming one can dereference any type of pointer for any type of data
2544
2545 char *p = ...;
def4ed7d 2546 long pony = *p; /* BAD */
d7889f52
JH
2547
2548Many platforms, quite rightly so, will give you a core dump instead
2549of a pony if the p happens not be correctly aligned.
2550
2551=item *
2552
2553Lvalue casts
2554
def4ed7d 2555 (int)*p = ...; /* BAD */
d7889f52
JH
2556
2557Simply not portable. Get your lvalue to be of the right type,
27565cb6
JH
2558or maybe use temporary variables, or dirty tricks with unions.
2559
2560=item *
2561
606fd33d
JH
2562Assume B<anything> about structs (especially the ones you
2563don't control, like the ones coming from the system headers)
27565cb6
JH
2564
2565=over 8
2566
2567=item *
2568
2569That a certain field exists in a struct
2570
2571=item *
2572
902821cc 2573That no other fields exist besides the ones you know of
27565cb6
JH
2574
2575=item *
2576
606fd33d 2577That a field is of certain signedness, sizeof, or type
27565cb6
JH
2578
2579=item *
2580
2581That the fields are in a certain order
2582
606fd33d
JH
2583=over 8
2584
27565cb6
JH
2585=item *
2586
606fd33d
JH
2587While C guarantees the ordering specified in the struct definition,
2588between different platforms the definitions might differ
2589
2590=back
27565cb6
JH
2591
2592=item *
2593
606fd33d
JH
2594That the sizeof(struct) or the alignments are the same everywhere
2595
2596=over 8
27565cb6
JH
2597
2598=item *
2599
606fd33d
JH
2600There might be padding bytes between the fields to align the fields -
2601the bytes can be anything
2602
2603=item *
2604
2605Structs are required to be aligned to the maximum alignment required
2606by the fields - which for native types is for usually equivalent to
2607sizeof() of the field
2608
2609=back
27565cb6
JH
2610
2611=back
d7889f52
JH
2612
2613=item *
2614
0bec6c03
JH
2615Mixing #define and #ifdef
2616
2617 #define BURGLE(x) ... \
def4ed7d 2618 #ifdef BURGLE_OLD_STYLE /* BAD */
0bec6c03
JH
2619 ... do it the old way ... \
2620 #else
2621 ... do it the new way ... \
2622 #endif
2623
ee9468a2
RGS
2624You cannot portably "stack" cpp directives. For example in the above
2625you need two separate BURGLE() #defines, one for each #ifdef branch.
2626
2627=item *
2628
2629Adding stuff after #endif or #else
2630
2631 #ifdef SNOSH
2632 ...
def4ed7d 2633 #else !SNOSH /* BAD */
ee9468a2 2634 ...
def4ed7d 2635 #endif SNOSH /* BAD */
ee9468a2 2636
def4ed7d
JH
2637The #endif and #else cannot portably have anything non-comment after
2638them. If you want to document what is going (which is a good idea
2639especially if the branches are long), use (C) comments:
ee9468a2
RGS
2640
2641 #ifdef SNOSH
2642 ...
2643 #else /* !SNOSH */
2644 ...
2645 #endif /* SNOSH */
2646
2647The gcc option C<-Wendif-labels> warns about the bad variant
2648(by default on starting from Perl 5.9.4).
0bec6c03
JH
2649
2650=item *
2651
27565cb6
JH
2652Having a comma after the last element of an enum list
2653
2654 enum color {
2655 CERULEAN,
2656 CHARTREUSE,
def4ed7d 2657 CINNABAR, /* BAD */
27565cb6
JH
2658 };
2659
2660is not portable. Leave out the last comma.
2661
2662Also note that whether enums are implicitly morphable to ints
2663varies between compilers, you might need to (int).
2664
2665=item *
2666
d7889f52
JH
2667Using //-comments
2668
def4ed7d 2669 // This function bamfoodles the zorklator. /* BAD */
d7889f52
JH
2670
2671That is C99 or C++. Perl is C89. Using the //-comments is silently
0bec6c03
JH
2672allowed by many C compilers but cranking up the ANSI C89 strictness
2673(which we like to do) causes the compilation to fail.
d7889f52
JH
2674
2675=item *
2676
2677Mixing declarations and code
2678
2679 void zorklator()
2680 {
2681 int n = 3;
def4ed7d 2682 set_zorkmids(n); /* BAD */
d7889f52
JH
2683 int q = 4;
2684
0bec6c03
JH
2685That is C99 or C++. Some C compilers allow that, but you shouldn't.
2686
63796a85
JH
2687The gcc option C<-Wdeclaration-after-statements> scans for such problems
2688(by default on starting from Perl 5.9.4).
2689
0bec6c03
JH
2690=item *
2691
2692Introducing variables inside for()
2693
def4ed7d 2694 for(int i = ...; ...; ...) { /* BAD */
0bec6c03
JH
2695
2696That is C99 or C++. While it would indeed be awfully nice to have that
2697also in C89, to limit the scope of the loop variable, alas, we cannot.
d7889f52
JH
2698
2699=item *
2700
2701Mixing signed char pointers with unsigned char pointers
2702
2703 int foo(char *s) { ... }
2704 ...
2705 unsigned char *t = ...; /* Or U8* t = ... */
def4ed7d 2706 foo(t); /* BAD */
d7889f52
JH
2707
2708While this is legal practice, it is certainly dubious, and downright
2709fatal in at least one platform: for example VMS cc considers this a
def4ed7d
JH
2710fatal error. One cause for people often making this mistake is that a
2711"naked char" and therefore dereferencing a "naked char pointer" have
2712an undefined signedness: it depends on the compiler and the flags of
2713the compiler and the underlying platform whether the result is signed
2714or unsigned. For this very same reason using a 'char' as an array
2715index is bad.
d7889f52
JH
2716
2717=item *
2718
2719Macros that have string constants and their arguments as substrings of
2720the string constants
2721
def4ed7d 2722 #define FOO(n) printf("number = %d\n", n) /* BAD */
d7889f52
JH
2723 FOO(10);
2724
2725Pre-ANSI semantics for that was equivalent to
2726
2727 printf("10umber = %d\10");
2728
0bec6c03
JH
2729which is probably not what you were expecting. Unfortunately at least
2730one reasonably common and modern C compiler does "real backward
63796a85 2731compatibility" here, in AIX that is what still happens even though the
0bec6c03
JH
2732rest of the AIX compiler is very happily C89.
2733
2734=item *
2735
ee9468a2
RGS
2736Using printf formats for non-basic C types
2737
2738 IV i = ...;
def4ed7d 2739 printf("i = %d\n", i); /* BAD */
ee9468a2
RGS
2740
2741While this might by accident work in some platform (where IV happens
2742to be an C<int>), in general it cannot. IV might be something larger.
2743Even worse the situation is with more specific types (defined by Perl's
2744configuration step in F<config.h>):
2745
2746 Uid_t who = ...;
def4ed7d 2747 printf("who = %d\n", who); /* BAD */
ee9468a2
RGS
2748
2749The problem here is that Uid_t might be not only not C<int>-wide
2750but it might also be unsigned, in which case large uids would be
2751printed as negative values.
2752
2753There is no simple solution to this because of printf()'s limited
2754intelligence, but for many types the right format is available as
2755with either 'f' or '_f' suffix, for example:
2756
2757 IVdf /* IV in decimal */
2758 UVxf /* UV is hexadecimal */
2759
2760 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2761
2762 Uid_t_f /* Uid_t in decimal */
2763
2764 printf("who = %"Uid_t_f"\n", who);
2765
63796a85
JH
2766Or you can try casting to a "wide enough" type:
2767
2768 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2769
2770Also remember that the C<%p> format really does require a void pointer:
2771
2772 U8* p = ...;
2773 printf("p = %p\n", (void*)p);
2774
ee9468a2
RGS
2775The gcc option C<-Wformat> scans for such problems.
2776
2777=item *
2778
0bec6c03
JH
2779Blindly using variadic macros
2780
63796a85
JH
2781gcc has had them for a while with its own syntax, and C99 brought
2782them with a standardized syntax. Don't use the former, and use
2783the latter only if the HAS_C99_VARIADIC_MACROS is defined.
0bec6c03
JH
2784
2785=item *
2786
2787Blindly passing va_list
2788
2789Not all platforms support passing va_list to further varargs (stdarg)
2790functions. The right thing to do is to copy the va_list using the
2791Perl_va_copy() if the NEED_VA_COPY is defined.
d7889f52 2792
ee9468a2
RGS
2793=item *
2794
e5afc1ae 2795Using gcc statement expressions
63796a85 2796
def4ed7d 2797 val = ({...;...;...}); /* BAD */
63796a85 2798
def4ed7d
JH
2799While a nice extension, it's not portable. The Perl code does
2800admittedly use them if available to gain some extra speed
2801(essentially as a funky form of inlining), but you shouldn't.
63796a85
JH
2802
2803=item *
2804
2805Binding together several statements
2806
2807Use the macros STMT_START and STMT_END.
2808
2809 STMT_START {
2810 ...
2811 } STMT_END
2812
2813=item *
2814
ee9468a2
RGS
2815Testing for operating systems or versions when should be testing for features
2816
def4ed7d 2817 #ifdef __FOONIX__ /* BAD */
ee9468a2
RGS
2818 foo = quux();
2819 #endif
2820
2821Unless you know with 100% certainty that quux() is only ever available
2822for the "Foonix" operating system B<and> that is available B<and>
2823correctly working for B<all> past, present, B<and> future versions of
2824"Foonix", the above is very wrong. This is more correct (though still
2825not perfect, because the below is a compile-time check):
2826
2827 #ifdef HAS_QUUX
2828 foo = quux();
2829 #endif
2830
def4ed7d 2831How does the HAS_QUUX become defined where it needs to be? Well, if
ee9468a2
RGS
2832Foonix happens to be UNIXy enought to be able to run the Configure
2833script, and Configure has been taught about detecting and testing
2834quux(), the HAS_QUUX will be correctly defined. In other platforms,
2835the corresponding configuration step will hopefully do the same.
2836
2837In a pinch, if you cannot wait for Configure to be educated,
2838or if you have a good hunch of where quux() might be available,
2839you can temporarily try the following:
2840
2841 #if (defined(__FOONIX__) || defined(__BARNIX__))
2842 # define HAS_QUUX
2843 #endif
2844
2845 ...
2846
2847 #ifdef HAS_QUUX
2848 foo = quux();
2849 #endif
2850
2851But in any case, try to keep the features and operating systems separate.
2852
d7889f52
JH
2853=back
2854
ad7244db
JH
2855=head2 Problematic System Interfaces
2856
2857=over 4
2858
2859=item *
2860
2861malloc(0), realloc(0), calloc(0, 0) are nonportable. To be portable
2862allocate at least one byte. (In general you should rarely need to
2863work at this low level, but instead use the various malloc wrappers.)
2864
2865=item *
2866
2867snprintf() - the return type is unportable. Use my_snprintf() instead.
2868
2869=back
2870
d7889f52
JH
2871=head2 Security problems
2872
2873Last but not least, here are various tips for safer coding.
2874
2875=over 4
2876
2877=item *
2878
2879Do not use gets()
2880
2881Or we will publicly ridicule you. Seriously.
2882
2883=item *
2884
d1307786 2885Do not use strcpy() or strcat() or strncpy() or strncat()
d7889f52 2886
d1307786
JH
2887Use my_strlcpy() and my_strlcat() instead: they either use the native
2888implementation, or Perl's own implementation (borrowed from the public
2889domain implementation of INN).
d7889f52
JH
2890
2891=item *
2892
2893Do not use sprintf() or vsprintf()
2894
0bec6c03 2895If you really want just plain byte strings, use my_snprintf()
64d9b66b 2896and my_vsnprintf() instead, which will try to use snprintf() and
0bec6c03
JH
2897vsnprintf() if those safer APIs are available. If you want something
2898fancier than a plain byte string, use SVs and Perl_sv_catpvf().
d7889f52
JH
2899
2900=back
2901
902b9dbf
MF
2902=head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2903
2904Sometimes it helps to use external tools while debugging and
2905testing Perl. This section tries to guide you through using
2906some common testing and debugging tools with Perl. This is
2907meant as a guide to interfacing these tools with Perl, not
2908as any kind of guide to the use of the tools themselves.
2909
a958818a
JH
2910B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2911Third Degree greatly slows down the execution: seconds become minutes,
2912minutes become hours. For example as of Perl 5.8.1, the
2913ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2914e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2915than six hours, even on a snappy computer-- the said test must be
2916doing something that is quite unfriendly for memory debuggers. If you
2917don't feel like waiting, that you can simply kill away the perl
2918process.
2919
2920B<NOTE 2>: To minimize the number of memory leak false alarms (see
2921L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2922environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2923and harness scripts do that automatically. But if you are running
2924some of the tests manually-- for csh-like shells:
2925
2926 setenv PERL_DESTRUCT_LEVEL 2
2927
2928and for Bourne-type shells:
2929
2930 PERL_DESTRUCT_LEVEL=2
2931 export PERL_DESTRUCT_LEVEL
2932
2933or in UNIXy environments you can also use the C<env> command:
2934
2935 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
a1b65709 2936
37c0adeb
JH
2937B<NOTE 3>: There are known memory leaks when there are compile-time
2938errors within eval or require, seeing C<S_doeval> in the call stack
2939is a good sign of these. Fixing these leaks is non-trivial,
2940unfortunately, but they must be fixed eventually.
2941
f50e5b73
MH
2942B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2943unless Perl is built with the Configure option
2944C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2945
902b9dbf
MF
2946=head2 Rational Software's Purify
2947
2948Purify is a commercial tool that is helpful in identifying
2949memory overruns, wild pointers, memory leaks and other such
2950badness. Perl must be compiled in a specific way for
2951optimal testing with Purify. Purify is available under
2952Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2953
902b9dbf
MF
2954=head2 Purify on Unix
2955
2956On Unix, Purify creates a new Perl binary. To get the most
2957benefit out of Purify, you should create the perl to Purify
2958using:
2959
2960 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2961 -Uusemymalloc -Dusemultiplicity
2962
2963where these arguments mean:
2964
2965=over 4
2966
2967=item -Accflags=-DPURIFY
2968
2969Disables Perl's arena memory allocation functions, as well as
2970forcing use of memory allocation functions derived from the
2971system malloc.
2972
2973=item -Doptimize='-g'
2974
2975Adds debugging information so that you see the exact source
2976statements where the problem occurs. Without this flag, all
2977you will see is the source filename of where the error occurred.
2978
2979=item -Uusemymalloc
2980
2981Disable Perl's malloc so that Purify can more closely monitor
2982allocations and leaks. Using Perl's malloc will make Purify
2983report most leaks in the "potential" leaks category.
2984
2985=item -Dusemultiplicity
2986
2987Enabling the multiplicity option allows perl to clean up
2988thoroughly when the interpreter shuts down, which reduces the
2989number of bogus leak reports from Purify.
2990
2991=back
2992
2993Once you've compiled a perl suitable for Purify'ing, then you
2994can just:
2995
07aa3531 2996 make pureperl
902b9dbf
MF
2997
2998which creates a binary named 'pureperl' that has been Purify'ed.
2999This binary is used in place of the standard 'perl' binary
3000when you want to debug Perl memory problems.
3001
3002As an example, to show any memory leaks produced during the
3003standard Perl testset you would create and run the Purify'ed
3004perl as:
3005
3006 make pureperl
3007 cd t
07aa3531 3008 ../pureperl -I../lib harness
902b9dbf
MF
3009
3010which would run Perl on test.pl and report any memory problems.
3011
3012Purify outputs messages in "Viewer" windows by default. If
3013you don't have a windowing environment or if you simply
3014want the Purify output to unobtrusively go to a log file
3015instead of to the interactive window, use these following
3016options to output to the log file "perl.log":
3017
3018 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
3019 -log-file=perl.log -append-logfile=yes"
3020
3021If you plan to use the "Viewer" windows, then you only need this option:
3022
3023 setenv PURIFYOPTIONS "-chain-length=25"
3024
c406981e
JH
3025In Bourne-type shells:
3026
98631ff8
JL
3027 PURIFYOPTIONS="..."
3028 export PURIFYOPTIONS
c406981e
JH
3029
3030or if you have the "env" utility:
3031
98631ff8 3032 env PURIFYOPTIONS="..." ../pureperl ...
c406981e 3033
902b9dbf
MF
3034=head2 Purify on NT
3035
3036Purify on Windows NT instruments the Perl binary 'perl.exe'
3037on the fly. There are several options in the makefile you
3038should change to get the most use out of Purify:
3039
3040=over 4
3041
3042=item DEFINES
3043
3044You should add -DPURIFY to the DEFINES line so the DEFINES
3045line looks something like:
3046
07aa3531 3047 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
902b9dbf
MF
3048
3049to disable Perl's arena memory allocation functions, as
3050well as to force use of memory allocation functions derived
3051from the system malloc.
3052
3053=item USE_MULTI = define
3054
3055Enabling the multiplicity option allows perl to clean up
3056thoroughly when the interpreter shuts down, which reduces the
3057number of bogus leak reports from Purify.
3058
3059=item #PERL_MALLOC = define
3060
3061Disable Perl's malloc so that Purify can more closely monitor
3062allocations and leaks. Using Perl's malloc will make Purify
3063report most leaks in the "potential" leaks category.
3064
3065=item CFG = Debug
3066
3067Adds debugging information so that you see the exact source
3068statements where the problem occurs. Without this flag, all
3069you will see is the source filename of where the error occurred.
3070
3071=back
3072
3073As an example, to show any memory leaks produced during the
3074standard Perl testset you would create and run Purify as:
3075
3076 cd win32
3077 make
3078 cd ../t
07aa3531 3079 purify ../perl -I../lib harness
902b9dbf
MF
3080
3081which would instrument Perl in memory, run Perl on test.pl,
3082then finally report any memory problems.
3083
7a834142
JH
3084=head2 valgrind
3085
3086The excellent valgrind tool can be used to find out both memory leaks
3087and illegal memory accesses. As of August 2003 it unfortunately works
3088only on x86 (ELF) Linux. The special "test.valgrind" target can be used
d44161bf 3089to run the tests under valgrind. Found errors and memory leaks are
07aa3531
JC
3090logged in files named F<testfile.valgrind>.
3091
3092Valgrind also provides a cachegrind tool, invoked on perl as:
3093
038c294a 3094 VG_OPTS=--tool=cachegrind make test.valgrind
d44161bf
MHM
3095
3096As system libraries (most notably glibc) are also triggering errors,
3097valgrind allows to suppress such errors using suppression files. The
3098default suppression file that comes with valgrind already catches a lot
3099of them. Some additional suppressions are defined in F<t/perl.supp>.
7a834142
JH
3100
3101To get valgrind and for more information see
3102
3103 http://developer.kde.org/~sewardj/
3104
f134cc4e 3105=head2 Compaq's/Digital's/HP's Third Degree
09187cb1
JH
3106
3107Third Degree is a tool for memory leak detection and memory access checks.
3108It is one of the many tools in the ATOM toolkit. The toolkit is only
3109available on Tru64 (formerly known as Digital UNIX formerly known as
3110DEC OSF/1).
3111
3112When building Perl, you must first run Configure with -Doptimize=-g
3113and -Uusemymalloc flags, after that you can use the make targets
51a35ef1
JH
3114"perl.third" and "test.third". (What is required is that Perl must be
3115compiled using the C<-g> flag, you may need to re-Configure.)
09187cb1 3116
64cea5fd 3117The short story is that with "atom" you can instrument the Perl
83f0ef60 3118executable to create a new executable called F<perl.third>. When the
4ae3d70a 3119instrumented executable is run, it creates a log of dubious memory
83f0ef60 3120traffic in file called F<perl.3log>. See the manual pages of atom and
4ae3d70a
JH
3121third for more information. The most extensive Third Degree
3122documentation is available in the Compaq "Tru64 UNIX Programmer's
3123Guide", chapter "Debugging Programs with Third Degree".
64cea5fd 3124
9c54ecba 3125The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
64cea5fd
JH
3126subdirectory. There is a problem with these files: Third Degree is so
3127effective that it finds problems also in the system libraries.
9c54ecba
JH
3128Therefore you should used the Porting/thirdclean script to cleanup
3129the F<*.3log> files.
64cea5fd
JH
3130
3131There are also leaks that for given certain definition of a leak,
3132aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
3133
3134=head2 PERL_DESTRUCT_LEVEL
3135
a958818a
JH
3136If you want to run any of the tests yourself manually using e.g.
3137valgrind, or the pureperl or perl.third executables, please note that
3138by default perl B<does not> explicitly cleanup all the memory it has
3139allocated (such as global memory arenas) but instead lets the exit()
3140of the whole program "take care" of such allocations, also known as
3141"global destruction of objects".
64cea5fd
JH
3142
3143There is a way to tell perl to do complete cleanup: set the
3144environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
3145The t/TEST wrapper does set this to 2, and this is what you
3146need to do too, if you don't want to see the "global leaks":
1f56d61a 3147For example, for "third-degreed" Perl:
64cea5fd 3148
1f56d61a 3149 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
09187cb1 3150
414f2397
RGS
3151(Note: the mod_perl apache module uses also this environment variable
3152for its own purposes and extended its semantics. Refer to the mod_perl
287a822c
RGS
3153documentation for more information. Also, spawned threads do the
3154equivalent of setting this variable to the value 1.)
5a6c59ef
DM
3155
3156If, at the end of a run you get the message I<N scalars leaked>, you can
fd0854ff
DM
3157recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
3158of all those leaked SVs to be dumped along with details as to where each
3159SV was originally allocated. This information is also displayed by
3160Devel::Peek. Note that the extra details recorded with each SV increases
3161memory usage, so it shouldn't be used in production environments. It also
3162converts C<new_SV()> from a macro into a real function, so you can use
3163your favourite debugger to discover where those pesky SVs were allocated.
414f2397 3164
46c6c7e2
JH
3165=head2 PERL_MEM_LOG
3166
9f653bb5 3167If compiled with C<-DPERL_MEM_LOG>, all Newx() and Renew() allocations
46c6c7e2
JH
3168and Safefree() in the Perl core go through logging functions, which is
3169handy for breakpoint setting. If also compiled with C<-DPERL_MEM_LOG_STDERR>,
e352bcff
JH
3170the allocations and frees are logged to STDERR (or more precisely, to the
3171file descriptor 2) in these logging functions, with the calling source code
3172file and line number (and C function name, if supported by the C compiler).
3173
3174This logging is somewhat similar to C<-Dm> but independent of C<-DDEBUGGING>,
3175and at a higher level (the C<-Dm> is directly at the point of C<malloc()>,
a015a77e 3176while the C<PERL_MEM_LOG> is at the level of C<New()>).
46c6c7e2 3177
51a35ef1
JH
3178=head2 Profiling
3179
3180Depending on your platform there are various of profiling Perl.
3181
3182There are two commonly used techniques of profiling executables:
10f58044 3183I<statistical time-sampling> and I<basic-block counting>.
51a35ef1
JH
3184
3185The first method takes periodically samples of the CPU program
3186counter, and since the program counter can be correlated with the code
3187generated for functions, we get a statistical view of in which
3188functions the program is spending its time. The caveats are that very
3189small/fast functions have lower probability of showing up in the
3190profile, and that periodically interrupting the program (this is
3191usually done rather frequently, in the scale of milliseconds) imposes
3192an additional overhead that may skew the results. The first problem
3193can be alleviated by running the code for longer (in general this is a
3194good idea for profiling), the second problem is usually kept in guard
3195by the profiling tools themselves.
3196
10f58044 3197The second method divides up the generated code into I<basic blocks>.
51a35ef1
JH
3198Basic blocks are sections of code that are entered only in the
3199beginning and exited only at the end. For example, a conditional jump
3200starts a basic block. Basic block profiling usually works by
10f58044 3201I<instrumenting> the code by adding I<enter basic block #nnnn>
51a35ef1
JH
3202book-keeping code to the generated code. During the execution of the
3203code the basic block counters are then updated appropriately. The
3204caveat is that the added extra code can skew the results: again, the
3205profiling tools usually try to factor their own effects out of the
3206results.
3207
83f0ef60
JH
3208=head2 Gprof Profiling
3209
51a35ef1
JH
3210gprof is a profiling tool available in many UNIX platforms,
3211it uses F<statistical time-sampling>.
83f0ef60
JH
3212
3213You can build a profiled version of perl called "perl.gprof" by
51a35ef1
JH
3214invoking the make target "perl.gprof" (What is required is that Perl
3215must be compiled using the C<-pg> flag, you may need to re-Configure).
3216Running the profiled version of Perl will create an output file called
3217F<gmon.out> is created which contains the profiling data collected
3218during the execution.
83f0ef60
JH
3219
3220The gprof tool can then display the collected data in various ways.
3221Usually gprof understands the following options:
3222
3223=over 4
3224
3225=item -a
3226
3227Suppress statically defined functions from the profile.
3228
3229=item -b
3230
3231Suppress the verbose descriptions in the profile.
3232
3233=item -e routine
3234
3235Exclude the given routine and its descendants from the profile.
3236
3237=item -f routine
3238
3239Display only the given routine and its descendants in the profile.
3240
3241=item -s
3242
3243Generate a summary file called F<gmon.sum> which then may be given
3244to subsequent gprof runs to accumulate data over several runs.
3245
3246=item -z
3247
3248Display routines that have zero usage.
3249
3250=back
3251
3252For more detailed explanation of the available commands and output
3253formats, see your own local documentation of gprof.
3254
038c294a 3255quick hint:
07aa3531
JC
3256
3257 $ sh Configure -des -Dusedevel -Doptimize='-g' -Accflags='-pg' -Aldflags='-pg' && make
3258 $ ./perl someprog # creates gmon.out in current directory
3259 $ gprof perl > out
3260 $ view out
3261
51a35ef1
JH
3262=head2 GCC gcov Profiling
3263
10f58044 3264Starting from GCC 3.0 I<basic block profiling> is officially available
51a35ef1
JH
3265for the GNU CC.
3266
3267You can build a profiled version of perl called F<perl.gcov> by
3268invoking the make target "perl.gcov" (what is required that Perl must
3269be compiled using gcc with the flags C<-fprofile-arcs
3270-ftest-coverage>, you may need to re-Configure).
3271
3272Running the profiled version of Perl will cause profile output to be
3273generated. For each source file an accompanying ".da" file will be
3274created.
3275
3276To display the results you use the "gcov" utility (which should
3277be installed if you have gcc 3.0 or newer installed). F<gcov> is
3278run on source code files, like this
3279
3280 gcov sv.c
3281
3282which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3283contain the source code annotated with relative frequencies of
3284execution indicated by "#" markers.
3285
3286Useful options of F<gcov> include C<-b> which will summarise the
3287basic block, branch, and function call coverage, and C<-c> which
3288instead of relative frequencies will use the actual counts. For
3289more information on the use of F<gcov> and basic block profiling
3290with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3291
3292 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3293
3294and its section titled "8. gcov: a Test Coverage Program"
3295
3296 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3297
07aa3531
JC
3298quick hint:
3299
3300 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3301 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3302 $ rm -f regexec.c.gcov regexec.gcda
3303 $ ./perl.gcov
3304 $ gcov regexec.c
3305 $ view regexec.c.gcov
3306
4ae3d70a
JH
3307=head2 Pixie Profiling
3308
51a35ef1
JH
3309Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3310UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
10f58044 3311I<basic-block counting>.
4ae3d70a 3312
83f0ef60 3313You can build a profiled version of perl called F<perl.pixie> by
51a35ef1
JH
3314invoking the make target "perl.pixie" (what is required is that Perl
3315must be compiled using the C<-g> flag, you may need to re-Configure).
3316
3317In Tru64 a file called F<perl.Addrs> will also be silently created,
3318this file contains the addresses of the basic blocks. Running the
3319profiled version of Perl will create a new file called "perl.Counts"
3320which contains the counts for the basic block for that particular
3321program execution.
4ae3d70a 3322
51a35ef1 3323To display the results you use the F<prof> utility. The exact
4ae3d70a
JH
3324incantation depends on your operating system, "prof perl.Counts" in
3325IRIX, and "prof -pixie -all -L. perl" in Tru64.
3326
6c41479b
JH
3327In IRIX the following prof options are available:
3328
3329=over 4
3330
3331=item -h
3332
3333Reports the most heavily used lines in descending order of use.
6e36760b 3334Useful for finding the hotspot lines.
6c41479b
JH
3335
3336=item -l
3337
3338Groups lines by procedure, with procedures sorted in descending order of use.
3339Within a procedure, lines are listed in source order.
6e36760b 3340Useful for finding the hotspots of procedures.
6c41479b
JH
3341
3342=back
3343
3344In Tru64 the following options are available:
3345
3346=over 4
3347
3958b146 3348=item -p[rocedures]
6c41479b 3349
3958b146 3350Procedures sorted in descending order by the number of cycles executed
6e36760b 3351in each procedure. Useful for finding the hotspot procedures.
6c41479b
JH
3352(This is the default option.)
3353
24000d2f 3354=item -h[eavy]
6c41479b 3355
6e36760b
JH
3356Lines sorted in descending order by the number of cycles executed in
3357each line. Useful for finding the hotspot lines.
6c41479b 3358
24000d2f 3359=item -i[nvocations]
6c41479b 3360
6e36760b
JH
3361The called procedures are sorted in descending order by number of calls
3362made to the procedures. Useful for finding the most used procedures.
6c41479b 3363
24000d2f 3364=item -l[ines]
6c41479b
JH
3365
3366Grouped by procedure, sorted by cycles executed per procedure.
6e36760b 3367Useful for finding the hotspots of procedures.
6c41479b
JH
3368
3369=item -testcoverage
3370
3371The compiler emitted code for these lines, but the code was unexecuted.
3372
24000d2f 3373=item -z[ero]
6c41479b
JH
3374
3375Unexecuted procedures.
3376
aa500c9e 3377=back
6c41479b
JH
3378
3379For further information, see your system's manual pages for pixie and prof.
4ae3d70a 3380
b8ddf6b3
SB
3381=head2 Miscellaneous tricks
3382
3383=over 4
3384
3385=item *
3386
cc177e1a 3387Those debugging perl with the DDD frontend over gdb may find the
b8ddf6b3
SB
3388following useful:
3389
3390You can extend the data conversion shortcuts menu, so for example you
3391can display an SV's IV value with one click, without doing any typing.
3392To do that simply edit ~/.ddd/init file and add after:
3393
3394 ! Display shortcuts.
3395 Ddd*gdbDisplayShortcuts: \
3396 /t () // Convert to Bin\n\
3397 /d () // Convert to Dec\n\
3398 /x () // Convert to Hex\n\
3399 /o () // Convert to Oct(\n\
3400
3401the following two lines:
3402
3403 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3404 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3405
3406so now you can do ivx and pvx lookups or you can plug there the
3407sv_peek "conversion":
3408
3409 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3410
3411(The my_perl is for threaded builds.)
3412Just remember that every line, but the last one, should end with \n\
3413
3414Alternatively edit the init file interactively via:
34153rd mouse button -> New Display -> Edit Menu
3416
3417Note: you can define up to 20 conversion shortcuts in the gdb
3418section.
3419
9965345d
JH
3420=item *
3421
7e337ee0
JH
3422If you see in a debugger a memory area mysteriously full of 0xABABABAB
3423or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3424see L<perlclib>.
9965345d 3425
f1fac472
NC
3426=item *
3427
3428Under ithreads the optree is read only. If you want to enforce this, to check
3429for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3430enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3431that allocates op memory via C<mmap>, and sets it read-only at run time.
3432Any write access to an op results in a C<SIGBUS> and abort.
3433
3434This code is intended for development only, and may not be portable even to
3435all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3436all ops read only. Specifically it
3437
3438=over
3439
3440=item 1
3441
3442Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3443later via C<require> or C<eval> will be re-write
3444
3445=item 2
3446
3447Turns an entire slab of ops read-write if the refcount of any op in the slab
3448needs to be decreased.
3449
3450=item 3
3451
3452Turns an entire slab of ops read-write if any op from the slab is freed.
3453
b8ddf6b3
SB
3454=back
3455
f1fac472
NC
3456It's not possible to turn the slabs to read-only after an action requiring
3457read-write access, as either can happen during op tree building time, so
3458there may still be legitimate write access.
3459
3460However, as an 80% solution it is still effective, as currently it catches
3461a write access during the generation of F<Config.pm>, which means that we
3462can't yet build F<perl> with this enabled.
3463
3464=back
3465
3466
955fec6b 3467=head1 CONCLUSION
a422fd2d 3468
955fec6b
JH
3469We've had a brief look around the Perl source, how to maintain quality
3470of the source code, an overview of the stages F<perl> goes through
3471when it's running your code, how to use debuggers to poke at the Perl
3472guts, and finally how to analyse the execution of Perl. We took a very
3473simple problem and demonstrated how to solve it fully - with
3474documentation, regression tests, and finally a patch for submission to
3475p5p. Finally, we talked about how to use external tools to debug and
3476test Perl.
a422fd2d
SC
3477
3478I'd now suggest you read over those references again, and then, as soon
3479as possible, get your hands dirty. The best way to learn is by doing,
07aa3531 3480so:
a422fd2d
SC
3481
3482=over 3
3483
3484=item *
3485
3486Subscribe to perl5-porters, follow the patches and try and understand
3487them; don't be afraid to ask if there's a portion you're not clear on -
3488who knows, you may unearth a bug in the patch...
3489
3490=item *
3491
3492Keep up to date with the bleeding edge Perl distributions and get
3493familiar with the changes. Try and get an idea of what areas people are
3494working on and the changes they're making.
3495
3496=item *
3497
3e148164 3498Do read the README associated with your operating system, e.g. README.aix
a1f349fd
MB
3499on the IBM AIX OS. Don't hesitate to supply patches to that README if
3500you find anything missing or changed over a new OS release.
3501
3502=item *
3503
a422fd2d
SC
3504Find an area of Perl that seems interesting to you, and see if you can
3505work out how it works. Scan through the source, and step over it in the
3506debugger. Play, poke, investigate, fiddle! You'll probably get to
3507understand not just your chosen area but a much wider range of F<perl>'s
3508activity as well, and probably sooner than you'd think.
3509
3510=back
3511
3512=over 3
3513
3514=item I<The Road goes ever on and on, down from the door where it began.>
3515
3516=back
3517
64d9b66b 3518If you can do these things, you've started on the long road to Perl porting.
a422fd2d
SC
3519Thanks for wanting to help make Perl better - and happy hacking!
3520
e8cd7eae
GS
3521=head1 AUTHOR
3522
3523This document was written by Nathan Torkington, and is maintained by
3524the perl5-porters mailing list.