3 Pumpkin - Notes on handling the Perl Patch Pumpkin And Porting Perl
7 There is no simple synopsis, yet.
11 This document attempts to begin to describe some of the considerations
12 involved in patching, porting, and maintaining perl.
14 This document is still under construction, and still subject to
15 significant changes. Still, I hope parts of it will be useful,
16 so I'm releasing it even though it's not done.
18 For the most part, it's a collection of anecdotal information that
19 already assumes some familiarity with the Perl sources. I really need
20 an introductory section that describes the organization of the sources
21 and all the various auxiliary files that are part of the distribution.
23 =head1 Where Do I Get Perl Sources and Related Material?
25 The Comprehensive Perl Archive Network (or CPAN) is the place to go.
26 There are many mirrors, but the easiest thing to use is probably
27 http://www.cpan.org/README.html , which automatically points you to a
28 mirror site "close" to you.
30 =head2 Perl5-porters mailing list
32 The mailing list perl5-porters@perl.org
33 is the main group working with the development of perl. If you're
34 interested in all the latest developments, you should definitely
35 subscribe. The list is high volume, but generally has a
36 fairly low noise level.
38 Subscribe by sending the message (in the body of your letter)
40 subscribe perl5-porters
42 to perl5-porters-request@perl.org .
44 Archives of the list are held at:
46 http://www.rosat.mpe-garching.mpg.de/mailing-lists/perl-porters/
48 =head1 How are Perl Releases Numbered?
50 Beginning with v5.6.0, even versions will stand for maintenance releases
51 and odd versions for development releases, i.e., v5.6.x for maintenance
52 releases, and v5.7.x for development releases. Before v5.6.0, subversions
53 _01 through _49 were reserved for bug-fix maintenance releases, and
54 subversions _50 through _99 for unstable development versions.
56 For example, in v5.6.1, the revision number is 5, the version is 6,
57 and 1 is the subversion.
59 For compatibility with the older numbering scheme the composite floating
60 point version number continues to be available as the magic variable $],
61 and amounts to C<$revision + $version/1000 + $subversion/100000>. This
62 can still be used in comparisons.
64 print "You've got an old perl\n" if $] < 5.005_03;
66 In addition, the version is also available as a string in $^V.
68 print "You've got a new perl\n" if $^V and $^V ge v5.6.0;
70 You can also require particular version (or later) with:
74 or using the new syntax available only from v5.6 onward:
78 At some point in the future, we may need to decide what to call the
79 next big revision. In the .package file used by metaconfig to
80 generate Configure, there are two variables that might be relevant:
81 $baserev=5 and $package=perl5.
83 Perl releases produced by the members of perl5-porters are usually
84 available on CPAN in the F<src/5.0/maint> and F<src/5.0/devel>
87 =head2 Maintenance and Development Subversions
89 The first rule of maintenance work is "First, do no harm."
91 Trial releases of bug-fix maintenance releases are announced on
92 perl5-porters. Trial releases use the new subversion number (to avoid
93 testers installing it over the previous release) and include a 'local
94 patch' entry in patchlevel.h. The distribution file contains the
95 string C<MAINT_TRIAL> to make clear that the file is not meant for
98 In general, the names of official distribution files for the public
99 always match the regular expression:
101 ^perl\d+\.(\d+)\.\d+(-MAINT_TRIAL_\d+)\.tar\.gz$
103 C<$1> in the pattern is always an even number for maintenance
104 versions, and odd for developer releases.
106 In the past it has been observed that pumkings tend to invent new
107 naming conventions on the fly. If you are a pumpking, before you
108 invent a new name for any of the three types of perl distributions,
109 please inform the guys from the CPAN who are doing indexing and
110 provide the trees of symlinks and the like. They will have to know
111 I<in advance> what you decide.
113 =head2 Why is it called the patch pumpkin?
115 Chip Salzenberg gets credit for that, with a nod to his cow orker,
116 David Croy. We had passed around various names (baton, token, hot
117 potato) but none caught on. Then, Chip asked:
121 Who has the patch pumpkin?
123 To explain: David Croy once told me once that at a previous job,
124 there was one tape drive and multiple systems that used it for backups.
125 But instead of some high-tech exclusion software, they used a low-tech
126 method to prevent multiple simultaneous backups: a stuffed pumpkin.
127 No one was allowed to make backups unless they had the "backup pumpkin".
133 =head1 Philosophical Issues in Patching and Porting Perl
135 There are no absolute rules, but there are some general guidelines I
136 have tried to follow as I apply patches to the perl sources.
137 (This section is still under construction.)
139 =head2 Solve problems as generally as possible
141 Never implement a specific restricted solution to a problem when you
142 can solve the same problem in a more general, flexible way.
144 For example, for dynamic loading to work on some SVR4 systems, we had
145 to build a shared libperl.so library. In order to build "FAT" binaries
146 on NeXT 4.0 systems, we had to build a special libperl library. Rather
147 than continuing to build a contorted nest of special cases, I
148 generalized the process of building libperl so that NeXT and SVR4 users
149 could still get their work done, but others could build a shared
150 libperl if they wanted to as well.
152 Contain your changes carefully. Assume nothing about other operating
153 systems, not even closely related ones. Your changes must not affect
156 Spy shamelessly on how similar patching or porting issues have been
159 If feasible, try to keep filenames 8.3-compliant to humor those poor
160 souls that get joy from running Perl under such dire limitations.
161 There's a script, check83.pl, for keeping your nose 8.3-clean.
163 =head2 Seek consensus on major changes
165 If you are making big changes, don't do it in secret. Discuss the
166 ideas in advance on perl5-porters.
168 =head2 Keep the documentation up-to-date
170 If your changes may affect how users use perl, then check to be sure
171 that the documentation is in sync with your changes. Be sure to
172 check all the files F<pod/*.pod> and also the F<INSTALL> document.
174 Consider writing the appropriate documentation first and then
175 implementing your change to correspond to the documentation.
177 =head2 Avoid machine-specific #ifdef's
179 To the extent reasonable, try to avoid machine-specific #ifdef's in
180 the sources. Instead, use feature-specific #ifdef's. The reason is
181 that the machine-specific #ifdef's may not be valid across major
182 releases of the operating system. Further, the feature-specific tests
183 may help out folks on another platform who have the same problem.
185 =head2 Machine-specific files
191 If you have many machine-specific #defines or #includes, consider
192 creating an "osish.h" (os2ish.h, vmsish.h, and so on) and including
193 that in perl.h. If you have several machine-specific files (function
194 emulations, function stubs, build utility wrappers) you may create a
195 separate subdirectory (djgpp, win32) and put the files in there.
196 Remember to update C<MANIFEST> when you add files.
198 If your system supports dynamic loading but none of the existing
199 methods at F<ext/DynaLoader/dl_*.xs> work for you, you must write
200 a new one. Study the existing ones to see what kind of interface
205 There are two kinds of hints: hints for building Perl and hints for
206 extensions. The former live in the C<hints> subdirectory, the latter
207 in C<ext/*/hints> subdirectories.
209 The top level hints are Bourne-shell scripts that set, modify and
210 unset appropriate Configure variables, based on the Configure command
211 line options and possibly existing config.sh and Policy.sh files from
212 previous Configure runs.
214 The extension hints are written in Perl (by the time they are used
215 miniperl has been built) and control the building of their respective
216 extensions. They can be used to for example manipulate compilation
219 =item build and installation Makefiles, scripts, and so forth
221 Sometimes you will also need to tweak the Perl build and installation
222 procedure itself, like for example F<Makefile.SH> and F<installperl>.
223 Tread very carefully, even more than usual. Contain your changes
228 Many of the tests in C<t> subdirectory assume machine-specific things
229 like existence of certain functions, something about filesystem
230 semantics, certain external utilities and their error messages. Use
231 the C<$^O> and the C<Config> module (which contains the results of the
232 Configure run, in effect the C<config.sh> converted to Perl) to either
233 skip (preferably not) or customize (preferable) the tests for your
238 Certain standard modules may need updating if your operating system
239 sports for example a native filesystem naming. You may want to update
240 some or all of the modules File::Basename, File::Spec, File::Path, and
241 File::Copy to become aware of your native filesystem syntax and
244 Remember to have a $VERSION in the modules. You can use the
245 Porting/checkVERSION.pl script for checking this.
249 If your operating system comes from outside UNIX you almost certainly
250 will have differences in the available operating system functionality
251 (missing system calls, different semantics, whatever). Please
252 document these at F<pod/perlport.pod>. If your operating system is
253 the first B<not> to have a system call also update the list of
254 "portability-bewares" at the beginning of F<pod/perlfunc.pod>.
256 A file called F<README.youros> at the top level that explains things
257 like how to install perl at this platform, where to get any possibly
258 required additional software, and for example what test suite errors
259 to expect, is nice too. Such files are in the process of being written
260 in pod format and will eventually be renamed F<INSTALL.youros>.
262 You may also want to write a separate F<.pod> file for your operating
263 system to tell about existing mailing lists, os-specific modules,
264 documentation, whatever. Please name these along the lines of
265 F<perl>I<youros>.pod. [unfinished: where to put this file (the pod/
266 subdirectory, of course: but more importantly, which/what index files
271 =head2 Allow for lots of testing
273 We should never release a main version without testing it as a
276 =head2 Test popular applications and modules.
278 We should never release a main version without testing whether or not
279 it breaks various popular modules and applications. A partial list of
280 such things would include majordomo, metaconfig, apache, Tk, CGI,
281 libnet, and libwww, to name just a few. Of course it's quite possible
282 that some of those things will be just plain broken and need to be fixed,
283 but, in general, we ought to try to avoid breaking widely-installed
286 =head2 Automated generation of derivative files
288 The F<embed.h>, F<keywords.h>, F<opcode.h>, and F<perltoc.pod> files
289 are all automatically generated by perl scripts. In general, don't
290 patch these directly; patch the data files instead.
292 F<Configure> and F<config_h.SH> are also automatically generated by
293 B<metaconfig>. In general, you should patch the metaconfig units
294 instead of patching these files directly. However, very minor changes
295 to F<Configure> may be made in between major sync-ups with the
296 metaconfig units, which tends to be complicated operations. But be
297 careful, this can quickly spiral out of control. Running metaconfig
300 Also F<Makefile> is automatically produced from F<Makefile.SH>.
301 In general, look out for all F<*.SH> files.
303 Finally, the sample files in the F<Porting/> subdirectory are
304 generated automatically by the script F<U/mksample> included
305 with the metaconfig units. See L<"run metaconfig"> below for
306 information on obtaining the metaconfig units.
308 =head1 How to Make a Distribution
310 There really ought to be a 'make dist' target, but there isn't.
311 The 'dist' suite of tools also contains a number of tools that I haven't
312 learned how to use yet. Some of them may make this all a bit easier.
314 Here are the steps I go through to prepare a patch & distribution.
316 Lots of it could doubtless be automated but isn't. The Porting/makerel
317 (make release) perl script does now help automate some parts of it.
319 =head2 Announce your intentions
321 First, you should volunteer out loud to take the patch pumpkin. It's
322 generally counter-productive to have multiple people working in secret
325 At the same time, announce what you plan to do with the patch pumpkin,
326 to allow folks a chance to object or suggest alternatives, or do it for
327 you. Naturally, the patch pumpkin holder ought to incorporate various
328 bug fixes and documentation improvements that are posted while he or
329 she has the pumpkin, but there might also be larger issues at stake.
331 One of the precepts of the subversion idea is that we shouldn't give
332 the patch pumpkin to anyone unless we have some idea what he or she
333 is going to do with it.
335 =head2 refresh pod/perltoc.pod
337 Presumably, you have done a full C<make> in your working source
338 directory. Before you C<make spotless> (if you do), and if you have
339 changed any documentation in any module or pod file, change to the
340 F<pod> directory and run C<make toc>.
342 =head2 run installhtml to check the validity of the pod files
344 =head2 update patchlevel.h
346 Don't be shy about using the subversion number, even for a relatively
347 modest patch. We've never even come close to using all 99 subversions,
348 and it's better to have a distinctive number for your patch. If you
349 need feedback on your patch, go ahead and issue it and promise to
350 incorporate that feedback quickly (e.g. within 1 week) and send out a
353 If you update the subversion number, you may need to change the version
354 number near the top of the F<Changes> file.
356 =head2 run metaconfig
358 If you need to make changes to Configure or config_h.SH, it may be best to
359 change the appropriate metaconfig units instead, and regenerate Configure.
363 will regenerate Configure and config_h.SH. Much more information
364 on obtaining and running metaconfig is in the F<U/README> file
365 that comes with Perl's metaconfig units. Perl's metaconfig units
366 should be available on CPAN. A set of units that will work with
367 perl5.005 is in the file F<mc_units-5.005_00-01.tar.gz> under
368 http://www.cpan.org/authors/id/ANDYD/ . The mc_units tar file
369 should be unpacked in your main perl source directory. Note: those
370 units were for use with 5.005. There may have been changes since then.
371 Check for later versions or contact perl5-porters@perl.org to obtain a
372 pointer to the current version.
374 Alternatively, do consider if the F<*ish.h> files might be a better
375 place for your changes.
379 Make sure the MANIFEST is up-to-date. You can use dist's B<manicheck>
380 program for this. You can also use
382 perl -w -MExtUtils::Manifest=fullcheck -e fullcheck
384 Both commands will also list extra files in the directory that are not
387 The MANIFEST is normally sorted.
389 If you are using metaconfig to regenerate Configure, then you should note
390 that metaconfig actually uses MANIFEST.new, so you want to be sure
391 MANIFEST.new is up-to-date too. I haven't found the MANIFEST/MANIFEST.new
392 distinction particularly useful, but that's probably because I still haven't
393 learned how to use the full suite of tools in the dist distribution.
395 =head2 Check permissions
397 All the tests in the t/ directory ought to be executable. The
398 main makefile used to do a 'chmod t/*/*.t', but that resulted in
399 a self-modifying distribution--something some users would strongly
400 prefer to avoid. The F<t/TEST> script will check for this
401 and do the chmod if needed, but the tests still ought to be
404 In all, the following files should probably be executable:
419 vms/ext/Stdio/test.pl
423 Other things ought to be readable, at least :-).
425 Probably, the permissions for the files could be encoded in MANIFEST
426 somehow, but I'm reluctant to change MANIFEST itself because that
427 could break old scripts that use MANIFEST.
429 I seem to recall that some SVR3 systems kept some sort of file that listed
430 permissions for system files; something like that might be appropriate.
434 This will build a config.sh and config.h. You can skip this if you haven't
435 changed Configure or config_h.SH at all. I use the following command
437 sh Configure -Dprefix=/opt/perl -Doptimize=-O -Dusethreads \
439 -Dcf_email='yourname@yourhost.yourplace.com' \
440 -Dperladmin='yourname@yourhost.yourplace.com' \
441 -Dmydomain='.yourplace.com' \
442 -Dmyhostname='yourhost' \
445 =head2 Update Porting/config.sh and Porting/config_H
448 This section needs revision. We're currently working on easing
449 the task of keeping the vms, win32, and plan9 config.sh info
450 up-to-date. The plan is to use keep up-to-date 'canned' config.sh
451 files in the appropriate subdirectories and then generate 'canned'
452 config.h files for vms, win32, etc. from the generic config.sh file.
453 This is to ease maintenance. When Configure gets updated, the parts
454 sometimes get scrambled around, and the changes in config_H can
455 sometimes be very hard to follow. config.sh, on the other hand, can
456 safely be sorted, so it's easy to track (typically very small) changes
457 to config.sh and then propoagate them to a canned 'config.h' by any
458 number of means, including a perl script in win32/ or carrying
459 config.sh and config_h.SH to a Unix system and running sh
460 config_h.SH.) Vms uses configure.com to generate its own config.sh
461 and config.h. If you want to add a new variable to config.sh check
462 with vms folk how to add it to configure.com too.
465 The Porting/config.sh and Porting/config_H files are provided to
466 help those folks who can't run Configure. It is important to keep
467 them up-to-date. If you have changed config_h.SH, those changes must
468 be reflected in config_H as well. (The name config_H was chosen to
469 distinguish the file from config.h even on case-insensitive file systems.)
470 Simply edit the existing config_H file; keep the first few explanatory
471 lines and then copy your new config.h below.
473 It may also be necessary to update win32/config.?c, and
474 plan9/config.plan9, though you should be quite careful in doing so if
475 you are not familiar with those systems. You might want to issue your
476 patch with a promise to quickly issue a follow-up that handles those
479 =head2 make run_byacc
481 If you have byacc-1.8.2 (available from CPAN), and if there have been
482 changes to F<perly.y>, you can regenerate the F<perly.c> file. The
483 run_byacc makefile target does this by running byacc and then applying
484 some patches so that byacc dynamically allocates space, rather than
485 having fixed limits. This patch is handled by the F<perly.fixer>
486 script. Depending on the nature of the changes to F<perly.y>, you may
487 or may not have to hand-edit the patch to apply correctly. If you do,
488 you should include the edited patch in the new distribution. If you
489 have byacc-1.9, the patch won't apply cleanly. Changes to the printf
490 output statements mean the patch won't apply cleanly. Long ago I
491 started to fix F<perly.fixer> to detect this, but I never completed the
494 If C<perly.c> or C<perly.h> changes, make sure you run C<perl vms/vms_yfix.pl>
495 to update the corresponding VMS files. This could be taken care of by
496 the regen_all target in the Unix Makefile. See also
497 L<VMS-specific updates>.
499 Some additional notes from Larry on this:
501 Don't forget to regenerate perly_c.diff.
505 patch perly.c <perly_c.diff
506 # manually apply any failed hunks
507 diff -c perly.c.orig perly.c >perly_c.diff
509 One chunk of lines that often fails begins with
513 and ends one line before
515 #define YYERRCODE 256
517 This only happens when you add or remove a token type. I suppose this
518 could be automated, but it doesn't happen very often nowadays.
522 =head2 make regen_all
524 This target takes care of the PERLYVMS, regen_headers, and regen_pods
527 =head2 make regen_headers
529 The F<embed.h>, F<keywords.h>, and F<opcode.h> files are all automatically
530 generated by perl scripts. Since the user isn't guaranteed to have a
531 working perl, we can't require the user to generate them. Hence you have
532 to, if you're making a distribution.
534 I used to include rules like the following in the makefile:
536 # The following three header files are generated automatically
537 # The correct versions should be already supplied with the perl kit,
538 # in case you don't have perl or 'sh' available.
539 # The - is to ignore error return codes in case you have the source
540 # installed read-only or you don't have perl yet.
541 keywords.h: keywords.pl
542 @echo "Don't worry if this fails."
546 However, I got B<lots> of mail consisting of people worrying because the
547 command failed. I eventually decided that I would save myself time
548 and effort by manually running C<make regen_headers> myself rather
549 than answering all the questions and complaints about the failing
552 =head2 make regen_pods
554 Will run `make regen_pods` in the pod directory for indexing.
556 =head2 global.sym, interp.sym and perlio.sym
558 Make sure these files are up-to-date. Read the comments in these
559 files and in perl_exp.SH to see what to do.
561 =head2 Binary compatibility
563 If you do change F<global.sym> or F<interp.sym>, think carefully about
564 what you are doing. To the extent reasonable, we'd like to maintain
565 source and binary compatibility with older releases of perl. That way,
566 extensions built under one version of perl will continue to work with
567 new versions of perl.
569 Of course, some incompatible changes may well be necessary. I'm just
570 suggesting that we not make any such changes without thinking carefully
571 about them first. If possible, we should provide
572 backwards-compatibility stubs. There's a lot of XS code out there.
573 Let's not force people to keep changing it.
577 Be sure to update the F<Changes> file. Try to include both an overall
578 summary as well as detailed descriptions of the changes. Your
579 audience will include other developers and users, so describe
580 user-visible changes (if any) in terms they will understand, not in
581 code like "initialize foo variable in bar function".
583 There are differing opinions on whether the detailed descriptions
584 ought to go in the Changes file or whether they ought to be available
585 separately in the patch file (or both). There is no disagreement that
586 detailed descriptions ought to be easily available somewhere.
588 If you update the subversion number in F<patchlevel.h>, you may need
589 to change the version number near the top of the F<Changes> file.
593 The F<Todo> file contains a roughly-catgorized unordered list of
594 aspects of Perl that could use enhancement, features that could be
595 added, areas that could be cleaned up, and so on. During your term as
596 pumpkin-holder, you will probably address some of these issues, and
597 perhaps identify others which, while you decide not to address them
598 this time around, may be tackled in the future. Update the file
599 reflect the situation as it stands when you hand over the pumpkin.
601 You might like, early in your pumpkin-holding career, to see if you
602 can find champions for partiticular issues on the to-do list: an issue
603 owned is an issue more likely to be resolved.
605 There are also some more porting-specific L<Todo> items later in this
608 =head2 OS/2-specific updates
610 In the os2 directory is F<diff.configure>, a set of OS/2-specific
611 diffs against B<Configure>. If you make changes to Configure, you may
612 want to consider regenerating this diff file to save trouble for the
615 You can also consider the OS/2 diffs as reminders of portability
616 things that need to be fixed in Configure.
618 =head2 VMS-specific updates
620 If you have changed F<perly.y> or F<perly.c>, then you most probably want
621 to update F<vms/perly_{h,c}.vms> by running C<perl vms/vms_yfix.pl>, or
622 by running `make regen_all` which will run that script for you.
624 The Perl revision number appears as "perl5" in configure.com.
625 It is courteous to update that if necessary.
627 =head2 Making the new distribution
629 Suppose, for example, that you want to make version 5.004_08. Then you can
630 do something like the following
632 mkdir ../perl5.004_08
633 awk '{print $1}' MANIFEST | cpio -pdm ../perl5.004_08
635 tar cf perl5.004_08.tar perl5.004_08
636 gzip --best perl5.004_08.tar
638 These steps, with extra checks, are automated by the Porting/makerel
641 =head2 Making a new patch
643 I find the F<makepatch> utility quite handy for making patches.
644 You can obtain it from any CPAN archive under
645 http://www.cpan.org/authors/Johan_Vromans/ . There are a couple
646 of differences between my version and the standard one. I have mine do
649 # Print a reassuring "End of Patch" note so people won't
650 # wonder if their mailer truncated patches.
651 print "\n\nEnd of Patch.\n";
653 at the end. That's because I used to get questions from people asking
654 if their mail was truncated.
656 It also writes Index: lines which include the new directory prefix
657 (change Index: print, approx line 294 or 310 depending on the version,
658 to read: print PATCH ("Index: $newdir$new\n");). That helps patches
659 work with more POSIX conformant patch programs.
661 Here's how I generate a new patch. I'll use the hypothetical
662 5.004_07 to 5.004_08 patch as an example.
664 # unpack perl5.004_07/
665 gzip -d -c perl5.004_07.tar.gz | tar -xof -
666 # unpack perl5.004_08/
667 gzip -d -c perl5.004_08.tar.gz | tar -xof -
668 makepatch perl5.004_07 perl5.004_08 > perl5.004_08.pat
670 Makepatch will automatically generate appropriate B<rm> commands to remove
671 deleted files. Unfortunately, it will not correctly set permissions
672 for newly created files, so you may have to do so manually. For example,
673 patch 5.003_04 created a new test F<t/op/gv.t> which needs to be executable,
674 so at the top of the patch, I inserted the following lines:
680 Now, of course, my patch is now wrong because makepatch didn't know I
681 was going to do that command, and it patched against /dev/null.
683 So, what I do is sort out all such shell commands that need to be in the
684 patch (including possible mv-ing of files, if needed) and put that in the
685 shell commands at the top of the patch. Next, I delete all the patch parts
686 of perl5.004_08.pat, leaving just the shell commands. Then, I do the
690 sh ../perl5.004_08.pat
692 makepatch perl5.004_07 perl5.004_08 >> perl5.004_08.pat
694 (Note the append to preserve my shell commands.)
695 Now, my patch will line up with what the end users are going to do.
697 =head2 Testing your patch
699 It seems obvious, but be sure to test your patch. That is, verify that
700 it produces exactly the same thing as your full distribution.
703 gzip -d -c perl5.004_07.tar.gz | tar -xf -
705 sh ../perl5.004_08.pat
706 patch -p1 -N < ../perl5.004_08.pat
708 gdiff -r perl5.004_07 perl5.004_08
710 where B<gdiff> is GNU diff. Other diff's may also do recursive checking.
714 Again, it's obvious, but you should test your new version as widely as you
715 can. You can be sure you'll hear about it quickly if your version doesn't
716 work on both ANSI and pre-ANSI compilers, and on common systems such as
717 SunOS 4.1.[34], Solaris, and Linux.
719 If your changes include conditional code, try to test the different
720 branches as thoroughly as you can. For example, if your system
721 supports dynamic loading, you can also test static loading with
725 You can also hand-tweak your config.h to try out different #ifdef
734 To test the correct use of printf-style arguments, C<Configure> with
735 S<-Dccflags='-DCHECK_FORMAT -Wformat'> and run C<make>. The compiler
736 will produce warning of incorrect use of format arguments. CHECK_FORMAT
737 changes perl-defined formats to common formats, so DO NOT USE the executable
738 produced by this process.
740 A more accurate approach is the following commands:
746 build miniperl with -DCHECK_FORMAT
749 make miniperl OPTIMIZE=-DCHECK_FORMAT >& mini.log
753 build a clean miniperl,
754 and build everything else from that with -DCHECK_FORMAT
758 make all OPTIMIZE='-DCHECK_FORMAT -Wformat' >& make.log
762 clean up, and print warnings from the log files
765 perl -nwe 'print if /^\S+:/ and not /^make\b/' \
770 (-Wformat support by Robin Barker.)
774 =head1 Running Purify
776 Purify is a commercial tool that is helpful in identifying memory
777 overruns, wild pointers, memory leaks and other such badness. Perl
778 must be compiled in a specific way for optimal testing with Purify.
780 Use the following commands to test perl with Purify:
782 sh Configure -des -Doptimize=-g -Uusemymalloc -Dusemultiplicity \
784 setenv PURIFYOPTIONS "-chain-length=25"
787 ln -s ../pureperl perl
788 setenv PERL_DESTRUCT_LEVEL 2
791 Disabling Perl's malloc allows Purify to monitor allocations and leaks
792 more closely; using Perl's malloc will make Purify report most leaks
793 in the "potential" leaks category. Enabling the multiplicity option
794 allows perl to clean up thoroughly when the interpreter shuts down, which
795 reduces the number of bogus leak reports from Purify. The -DPURIFY
796 enables any Purify-specific debugging code in the sources.
798 Purify outputs messages in "Viewer" windows by default. If you don't have
799 a windowing environment or if you simply want the Purify output to
800 unobtrusively go to a log file instead of to the interactive window,
801 use the following options instead:
803 setenv PURIFYOPTIONS "-chain-length=25 -windows=no -log-file=perl.log \
806 The only currently known leaks happen when there are compile-time errors
807 within eval or require. (Fixing these is non-trivial, unfortunately, but
808 they must be fixed eventually.)
810 =head1 Common Gotcha's
816 The '#elif' preprocessor directive is not understood on all systems.
817 Specifically, I know that Pyramids don't understand it. Thus instead of the
828 You have to do the more Byzantine
840 Incidentally, whitespace between the leading '#' and the preprocessor
841 command is not guaranteed, but is very portable and you may use it freely.
842 I think it makes things a bit more readable, especially once things get
843 rather deeply nested. I also think that things should almost never get
844 too deeply nested, so it ought to be a moot point :-)
846 =item Probably Prefer POSIX
848 It's often the case that you'll need to choose whether to do
849 something the BSD-ish way or the POSIX-ish way. It's usually not
850 a big problem when the two systems use different names for similar
851 functions, such as memcmp() and bcmp(). The perl.h header file
852 handles these by appropriate #defines, selecting the POSIX mem*()
853 functions if available, but falling back on the b*() functions, if
856 More serious is the case where some brilliant person decided to
857 use the same function name but give it a different meaning or
858 calling sequence :-). getpgrp() and setpgrp() come to mind.
859 These are a real problem on systems that aim for conformance to
860 one standard (e.g. POSIX), but still try to support the other way
861 of doing things (e.g. BSD). My general advice (still not really
862 implemented in the source) is to do something like the following.
863 Suppose there are two alternative versions, fooPOSIX() and
867 /* use fooPOSIX(); */
870 /* try to emulate fooPOSIX() with fooBSD();
871 perhaps with the following: */
872 # define fooPOSIX fooBSD
874 # /* Uh, oh. We have to supply our own. */
875 # define fooPOSIX Perl_fooPOSIX
879 =item Think positively
881 If you need to add an #ifdef test, it is usually easier to follow if you
882 think positively, e.g.
884 #ifdef HAS_NEATO_FEATURE
885 /* use neato feature */
887 /* use some fallback mechanism */
890 rather than the more impenetrable
892 #ifndef MISSING_NEATO_FEATURE
893 /* Not missing it, so we must have it, so use it */
895 /* Are missing it, so fall back on something else. */
898 Of course for this toy example, there's not much difference. But when
899 the #ifdef's start spanning a couple of screen fulls, and the #else's
900 are marked something like
902 #else /* !MISSING_NEATO_FEATURE */
904 I find it easy to get lost.
906 =item Providing Missing Functions -- Problem
908 Not all systems have all the neat functions you might want or need, so
909 you might decide to be helpful and provide an emulation. This is
910 sound in theory and very kind of you, but please be careful about what
911 you name the function. Let me use the C<pause()> function as an
914 Perl5.003 has the following in F<perl.h>
917 #define pause() sleep((32767<<16)+32767)
920 Configure sets HAS_PAUSE if the system has the pause() function, so
921 this #define only kicks in if the pause() function is missing.
924 Unfortunately, some systems apparently have a prototype for pause()
925 in F<unistd.h>, but don't actually have the function in the library.
926 (Or maybe they do have it in a library we're not using.)
928 Thus, the compiler sees something like
930 extern int pause(void);
932 #define pause() sleep((32767<<16)+32767)
934 and dies with an error message. (Some compilers don't mind this;
935 others apparently do.)
937 To work around this, 5.003_03 and later have the following in perl.h:
939 /* Some unistd.h's give a prototype for pause() even though
940 HAS_PAUSE ends up undefined. This causes the #define
941 below to be rejected by the compiler. Sigh.
946 # define Pause() sleep((32767<<16)+32767)
951 The curious reader may wonder why I didn't do the following in
957 sleep((32767<<16)+32767);
961 That is, since the function is missing, just provide it.
962 Then things would probably be been alright, it would seem.
964 Well, almost. It could be made to work. The problem arises from the
965 conflicting needs of dynamic loading and namespace protection.
967 For dynamic loading to work on AIX (and VMS) we need to provide a list
968 of symbols to be exported. This is done by the script F<perl_exp.SH>,
969 which reads F<global.sym> and F<interp.sym>. Thus, the C<pause>
970 symbol would have to be added to F<global.sym> So far, so good.
972 On the other hand, one of the goals of Perl5 is to make it easy to
973 either extend or embed perl and link it with other libraries. This
974 means we have to be careful to keep the visible namespace "clean".
975 That is, we don't want perl's global variables to conflict with
976 those in the other application library. Although this work is still
977 in progress, the way it is currently done is via the F<embed.h> file.
978 This file is built from the F<global.sym> and F<interp.sym> files,
979 since those files already list the globally visible symbols. If we
980 had added C<pause> to global.sym, then F<embed.h> would contain the
983 #define pause Perl_pause
985 and calls to C<pause> in the perl sources would now point to
986 C<Perl_pause>. Now, when B<ld> is run to build the F<perl> executable,
987 it will go looking for C<perl_pause>, which probably won't exist in any
988 of the standard libraries. Thus the build of perl will fail.
990 Those systems where C<HAS_PAUSE> is not defined would be ok, however,
991 since they would get a C<Perl_pause> function in util.c. The rest of
992 the world would be in trouble.
994 And yes, this scenario has happened. On SCO, the function C<chsize>
995 is available. (I think it's in F<-lx>, the Xenix compatibility
996 library.) Since the perl4 days (and possibly before), Perl has
997 included a C<chsize> function that gets called something akin to
1000 I32 chsize(fd, length)
1006 #define chsize Perl_chsize
1008 to F<embed.h>, the compile started failing on SCO systems.
1010 The "fix" is to give the function a different name. The one
1011 implemented in 5.003_05 isn't optimal, but here's what was done:
1014 # ifdef my_chsize /* Probably #defined to Perl_my_chsize in embed.h */
1017 # define my_chsize chsize
1020 My explanatory comment in patch 5.003_05 said:
1022 Undef and then re-define my_chsize from Perl_my_chsize to
1023 just plain chsize if this system HAS_CHSIZE. This probably only
1024 applies to SCO. This shows the perils of having internal
1025 functions with the same name as external library functions :-).
1027 Now, we can safely put C<my_chsize> in F<global.sym>, export it, and
1028 hide it with F<embed.h>.
1030 To be consistent with what I did for C<pause>, I probably should have
1031 called the new function C<Chsize>, rather than C<my_chsize>.
1032 However, the perl sources are quite inconsistent on this (Consider
1033 New, Mymalloc, and Myremalloc, to name just a few.)
1035 There is a problem with this fix, however, in that C<Perl_chsize>
1036 was available as a F<libperl.a> library function in 5.003, but it
1037 isn't available any more (as of 5.003_07). This means that we've
1038 broken binary compatibility. This is not good.
1040 =item Providing missing functions -- some ideas
1042 We currently don't have a standard way of handling such missing
1043 function names. Right now, I'm effectively thinking aloud about a
1044 solution. Some day, I'll try to formally propose a solution.
1046 Part of the problem is that we want to have some functions listed as
1047 exported but not have their names mangled by embed.h or possibly
1048 conflict with names in standard system headers. We actually already
1049 have such a list at the end of F<perl_exp.SH> (though that list is
1052 # extra globals not included above.
1053 cat <<END >> perl.exp
1077 This still needs much thought, but I'm inclined to think that one
1078 possible solution is to prefix all such functions with C<perl_> in the
1079 source and list them along with the other C<perl_*> functions in
1082 Thus, for C<chsize>, we'd do something like the following:
1086 # define perl_chsize chsize
1089 then in some file (e.g. F<util.c> or F<doio.c>) do
1092 I32 perl_chsize(fd, length)
1093 /* implement the function here . . . */
1096 Alternatively, we could just always use C<chsize> everywhere and move
1097 C<chsize> from F<global.sym> to the end of F<perl_exp.SH>. That would
1098 probably be fine as long as our C<chsize> function agreed with all the
1099 C<chsize> function prototypes in the various systems we'll be using.
1100 As long as the prototypes in actual use don't vary that much, this is
1101 probably a good alternative. (As a counter-example, note how Configure
1102 and perl have to go through hoops to find and use get Malloc_t and
1103 Free_t for C<malloc> and C<free>.)
1105 At the moment, this latter option is what I tend to prefer.
1107 =item All the world's a VAX
1109 Sorry, showing my age:-). Still, all the world is not BSD 4.[34],
1110 SVR4, or POSIX. Be aware that SVR3-derived systems are still quite
1111 common (do you have any idea how many systems run SCO?) If you don't
1112 have a bunch of v7 manuals handy, the metaconfig units (by default
1113 installed in F</usr/local/lib/dist/U>) are a good resource to look at
1118 =head1 Miscellaneous Topics
1122 Why does perl use a metaconfig-generated Configure script instead of an
1123 autoconf-generated configure script?
1125 Metaconfig and autoconf are two tools with very similar purposes.
1126 Metaconfig is actually the older of the two, and was originally written
1127 by Larry Wall, while autoconf is probably now used in a wider variety of
1128 packages. The autoconf info file discusses the history of autoconf and
1129 how it came to be. The curious reader is referred there for further
1132 Overall, both tools are quite good, I think, and the choice of which one
1133 to use could be argued either way. In March, 1994, when I was just
1134 starting to work on Configure support for Perl5, I considered both
1135 autoconf and metaconfig, and eventually decided to use metaconfig for the
1140 =item Compatibility with Perl4
1142 Perl4 used metaconfig, so many of the #ifdef's were already set up for
1143 metaconfig. Of course metaconfig had evolved some since Perl4's days,
1144 but not so much that it posed any serious problems.
1146 =item Metaconfig worked for me
1148 My system at the time was Interactive 2.2, an SVR3.2/386 derivative that
1149 also had some POSIX support. Metaconfig-generated Configure scripts
1150 worked fine for me on that system. On the other hand, autoconf-generated
1151 scripts usually didn't. (They did come quite close, though, in some
1152 cases.) At the time, I actually fetched a large number of GNU packages
1153 and checked. Not a single one configured and compiled correctly
1154 out-of-the-box with the system's cc compiler.
1156 =item Configure can be interactive
1158 With both autoconf and metaconfig, if the script works, everything is
1159 fine. However, one of my main problems with autoconf-generated scripts
1160 was that if it guessed wrong about something, it could be B<very> hard to
1161 go back and fix it. For example, autoconf always insisted on passing the
1162 -Xp flag to cc (to turn on POSIX behavior), even when that wasn't what I
1163 wanted or needed for that package. There was no way short of editing the
1164 configure script to turn this off. You couldn't just edit the resulting
1165 Makefile at the end because the -Xp flag influenced a number of other
1168 Metaconfig's Configure scripts, on the other hand, can be interactive.
1169 Thus if Configure is guessing things incorrectly, you can go back and fix
1170 them. This isn't as important now as it was when we were actively
1171 developing Configure support for new features such as dynamic loading,
1172 but it's still useful occasionally.
1176 At the time, autoconf-generated scripts were covered under the GNU Public
1177 License, and hence weren't suitable for inclusion with Perl, which has a
1178 different licensing policy. (Autoconf's licensing has since changed.)
1182 Metaconfig builds up Configure from a collection of discrete pieces
1183 called "units". You can override the standard behavior by supplying your
1184 own unit. With autoconf, you have to patch the standard files instead.
1185 I find the metaconfig "unit" method easier to work with. Others
1186 may find metaconfig's units clumsy to work with.
1190 =head2 Why isn't there a directory to override Perl's library?
1192 Mainly because no one's gotten around to making one. Note that
1193 "making one" involves changing perl.c, Configure, config_h.SH (and
1194 associated files, see above), and I<documenting> it all in the
1197 Apparently, most folks who want to override one of the standard library
1198 files simply do it by overwriting the standard library files.
1202 In the perl.c sources, you'll find an undocumented APPLLIB_EXP
1203 variable, sort of like PRIVLIB_EXP and ARCHLIB_EXP (which are
1204 documented in config_h.SH). Here's what APPLLIB_EXP is for, from
1205 a mail message from Larry:
1207 The main intent of APPLLIB_EXP is for folks who want to send out a
1208 version of Perl embedded in their product. They would set the symbol
1209 to be the name of the library containing the files needed to run or to
1210 support their particular application. This works at the "override"
1211 level to make sure they get their own versions of any library code that
1212 they absolutely must have configuration control over.
1214 As such, I don't see any conflict with a sysadmin using it for a
1215 override-ish sort of thing, when installing a generic Perl. It should
1216 probably have been named something to do with overriding though. Since
1217 it's undocumented we could still change it... :-)
1219 Given that it's already there, you can use it to override
1220 distribution modules. If you do
1222 sh Configure -Dccflags='-DAPPLLIB_EXP=/my/override'
1224 then perl.c will put /my/override ahead of ARCHLIB and PRIVLIB.
1226 =head2 Shared libperl.so location
1228 Why isn't the shared libperl.so installed in /usr/lib/ along
1229 with "all the other" shared libraries? Instead, it is installed
1230 in $archlib, which is typically something like
1232 /usr/local/lib/perl5/archname/5.00404
1234 and is architecture- and version-specific.
1236 The basic reason why a shared libperl.so gets put in $archlib is so that
1237 you can have more than one version of perl on the system at the same time,
1238 and have each refer to its own libperl.so.
1240 Three examples might help. All of these work now; none would work if you
1241 put libperl.so in /usr/lib.
1247 Suppose you want to have both threaded and non-threaded perl versions
1248 around. Configure will name both perl libraries "libperl.so" (so that
1249 you can link to them with -lperl). The perl binaries tell them apart
1250 by having looking in the appropriate $archlib directories.
1254 Suppose you have perl5.004_04 installed and you want to try to compile
1255 it again, perhaps with different options or after applying a patch.
1256 If you already have libperl.so installed in /usr/lib/, then it may be
1257 either difficult or impossible to get ld.so to find the new libperl.so
1258 that you're trying to build. If, instead, libperl.so is tucked away in
1259 $archlib, then you can always just change $archlib in the current perl
1260 you're trying to build so that ld.so won't find your old libperl.so.
1261 (The INSTALL file suggests you do this when building a debugging perl.)
1265 The shared perl library is not a "well-behaved" shared library with
1266 proper major and minor version numbers, so you can't necessarily
1267 have perl5.004_04 and perl5.004_05 installed simultaneously. Suppose
1268 perl5.004_04 were to install /usr/lib/libperl.so.4.4, and perl5.004_05
1269 were to install /usr/lib/libperl.so.4.5. Now, when you try to run
1270 perl5.004_04, ld.so might try to load libperl.so.4.5, since it has
1271 the right "major version" number. If this works at all, it almost
1272 certainly defeats the reason for keeping perl5.004_04 around. Worse,
1273 with development subversions, you certaily can't guarantee that
1274 libperl.so.4.4 and libperl.so.4.55 will be compatible.
1276 Anyway, all this leads to quite obscure failures that are sure to drive
1277 casual users crazy. Even experienced users will get confused :-). Upon
1278 reflection, I'd say leave libperl.so in $archlib.
1282 =head1 Upload Your Work to CPAN
1284 You can upload your work to CPAN if you have a CPAN id. Check out
1285 http://www.cpan.org/modules/04pause.html for information on
1286 _PAUSE_, the Perl Author's Upload Server.
1288 I typically upload both the patch file, e.g. F<perl5.004_08.pat.gz>
1289 and the full tar file, e.g. F<perl5.004_08.tar.gz>.
1291 If you want your patch to appear in the F<src/5.0/unsupported>
1292 directory on CPAN, send e-mail to the CPAN master librarian. (Check
1293 out http://www.cpan.org/CPAN.html ).
1295 =head1 Help Save the World
1297 You should definitely announce your patch on the perl5-porters list.
1298 You should also consider announcing your patch on
1299 comp.lang.perl.announce, though you should make it quite clear that a
1300 subversion is not a production release, and be prepared to deal with
1301 people who will not read your disclaimer.
1305 Here, in no particular order, are some Configure and build-related
1306 items that merit consideration. This list isn't exhaustive, it's just
1307 what I came up with off the top of my head.
1309 =head2 Adding missing library functions to Perl
1311 The perl Configure script automatically determines which headers and
1312 functions you have available on your system and arranges for them to be
1313 included in the compilation and linking process. Occasionally, when porting
1314 perl to an operating system for the first time, you may find that the
1315 operating system is missing a key function. While perl may still build
1316 without this function, no perl program will be able to reference the missing
1317 function. You may be able to write the missing function yourself, or you
1318 may be able to find the missing function in the distribution files for
1319 another software package. In this case, you need to instruct the perl
1320 configure-and-build process to use your function. Perform these steps.
1326 Code and test the function you wish to add. Test it carefully; you will
1327 have a much easier time debugging your code independently than when it is a
1332 Here is an implementation of the POSIX truncate function for an operating
1333 system (VOS) that does not supply one, but which does supply the ftruncate()
1336 /* Beginning of modification history */
1337 /* Written 02-01-02 by Nick Ing-Simmons (nick@ing-simmons.net) */
1338 /* End of modification history */
1340 /* VOS doesn't supply a truncate function, so we build one up
1341 from the available POSIX functions. */
1344 #include <sys/types.h>
1348 truncate(const char *path, off_t len)
1350 int fd = open(path,O_WRONLY);
1353 code = ftruncate(fd,len);
1359 Place this file into a subdirectory that has the same name as the operating
1360 system. This file is named perl/vos/vos.c
1364 If your operating system has a hints file (in perl/hints/XXX.sh for an
1365 operating system named XXX), then start with it. If your operating system
1366 has no hints file, then create one. You can use a hints file for a similar
1367 operating system, if one exists, as a template.
1371 Add lines like the following to your hints file. The first line
1372 (d_truncate="define") instructs Configure that the truncate() function
1373 exists. The second line (archobjs="vos.o") instructs the makefiles that the
1374 perl executable depends on the existence of a file named "vos.o". (Make
1375 will automatically look for "vos.c" and compile it with the same options as
1376 the perl source code). The final line ("test -h...") adds a symbolic link
1377 to the top-level directory so that make can find vos.c. Of course, you
1378 should use your own operating system name for the source file of extensions,
1381 # VOS does not have truncate() but we supply one in vos.c
1385 # Help gmake find vos.c
1386 test -h vos.c || ln -s vos/vos.c vos.c
1388 The hints file is a series of shell commands that are run in the top-level
1389 directory (the "perl" directory). Thus, these commands are simply executed
1390 by Configure at an appropriate place during its execution.
1394 At this point, you can run the Configure script and rebuild perl. Carefully
1395 test the newly-built perl to ensure that normal paths, and error paths,
1396 behave as you expect.
1400 =head2 Good ideas waiting for round tuits
1404 =item Configure -Dsrc=/blah/blah
1406 We should be able to emulate B<configure --srcdir>. Tom Tromey
1407 tromey@creche.cygnus.com has submitted some patches to
1408 the dist-users mailing list along these lines. They have been folded
1409 back into the main distribution, but various parts of the perl
1410 Configure/build/install process still assume src='.'.
1412 =item Hint file fixes
1414 Various hint files work around Configure problems. We ought to fix
1415 Configure so that most of them aren't needed.
1417 =item Hint file information
1419 Some of the hint file information (particularly dynamic loading stuff)
1420 ought to be fed back into the main metaconfig distribution.
1424 =head2 Probably good ideas waiting for round tuits
1428 =item GNU configure --options
1430 I've received sensible suggestions for --exec_prefix and other
1431 GNU configure --options. It's not always obvious exactly what is
1432 intended, but this merits investigation.
1436 Currently, B<make clean> isn't all that useful, though
1437 B<make realclean> and B<make distclean> are. This needs a bit of
1438 thought and documentation before it gets cleaned up.
1440 =item Try gcc if cc fails
1442 Currently, we just give up.
1444 =item bypassing safe*alloc wrappers
1446 On some systems, it may be safe to call the system malloc directly
1447 without going through the util.c safe* layers. (Such systems would
1448 accept free(0), for example.) This might be a time-saver for systems
1449 that already have a good malloc. (Recent Linux libc's apparently have
1450 a nice malloc that is well-tuned for the system.)
1454 =head2 Vague possibilities
1460 Get some of the Macintosh stuff folded back into the main distribution.
1462 =item gconvert replacement
1464 Maybe include a replacement function that doesn't lose data in rare
1465 cases of coercion between string and numerical values.
1467 =item Improve makedepend
1469 The current makedepend process is clunky and annoyingly slow, but it
1470 works for most folks. Alas, it assumes that there is a filename
1471 $firstmakefile that the B<make> command will try to use before it uses
1472 F<Makefile>. Such may not be the case for all B<make> commands,
1473 particularly those on non-Unix systems.
1475 Probably some variant of the BSD F<.depend> file will be useful.
1476 We ought to check how other packages do this, if they do it at all.
1477 We could probably pre-generate the dependencies (with the exception of
1478 malloc.o, which could probably be determined at F<Makefile.SH>
1481 =item GNU Makefile standard targets
1483 GNU software generally has standardized Makefile targets. Unless we
1484 have good reason to do otherwise, I see no reason not to support them.
1488 Somehow, straighten out, document, and implement lockf(), flock(),
1489 and/or fcntl() file locking. It's a mess. See $d_fcntl_can_lock
1490 in recent config.sh files though.
1496 Original author: Andy Dougherty doughera@lafcol.lafayette.edu .
1497 Additions by Chip Salzenberg chip@perl.com and
1498 Tim Bunce Tim.Bunce@ig.co.uk .
1500 All opinions expressed herein are those of the authorZ<>(s).
1502 =head1 LAST MODIFIED
1504 $Id: pumpkin.pod,v 1.23 2000/01/13 19:45:13 doughera Released $