Regen Configure.
[p5sagit/p5-mst-13.2.git] / Porting / pumpkin.pod
CommitLineData
aa689395 1=head1 NAME
2
3Pumpkin - Notes on handling the Perl Patch Pumpkin
4
5=head1 SYNOPSIS
6
7There is no simple synopsis, yet.
8
9=head1 DESCRIPTION
10
98dddfbd 11This document attempts to begin to describe some of the considerations
12involved in patching, porting, and maintaining perl.
aa689395 13
14This document is still under construction, and still subject to
15significant changes. Still, I hope parts of it will be useful,
16so I'm releasing it even though it's not done.
17
18For the most part, it's a collection of anecdotal information that
19already assumes some familiarity with the Perl sources. I really need
20an introductory section that describes the organization of the sources
21and all the various auxiliary files that are part of the distribution.
22
23=head1 Where Do I Get Perl Sources and Related Material?
24
25The Comprehensive Perl Archive Network (or CPAN) is the place to go.
26There are many mirrors, but the easiest thing to use is probably
7b5757d1 27http://www.perl.com/CPAN/README.html , which automatically points you to a
aa689395 28mirror site "close" to you.
29
30=head2 Perl5-porters mailing list
31
32The mailing list perl5-porters@perl.org
33is the main group working with the development of perl. If you're
34interested in all the latest developments, you should definitely
35subscribe. The list is high volume, but generally has a
36fairly low noise level.
37
38Subscribe by sending the message (in the body of your letter)
39
40 subscribe perl5-porters
41
42to perl5-porters-request@perl.org .
43
fb73857a 44Archives of the list are held at:
45
46 http://www.rosat.mpe-garching.mpg.de/mailing-lists/perl-porters/
47
aa689395 48=head1 How are Perl Releases Numbered?
49
7b5757d1 50Perl version numbers are floating point numbers, such as 5.004.
51(Observations about the imprecision of floating point numbers for
52representing reality probably have more relevance than you might
53imagine :-) The major version number is 5 and the '004' is the
54patchlevel. (Questions such as whether or not '004' is really a minor
55version number can safely be ignored.:)
56
57The version number is available as the magic variable $],
aa689395 58and can be used in comparisons, e.g.
59
60 print "You've got an old perl\n" if $] < 5.002;
61
aa689395 62You can also require particular version (or later) with
63
64 use 5.002;
65
7b5757d1 66At some point in the future, we may need to decide what to call the
67next big revision. In the .package file used by metaconfig to
68generate Configure, there are two variables that might be relevant:
69$baserev=5.0 and $package=perl5. At various times, I have suggested
70we might change them to $baserev=5.1 and $package=perl5.1 if want
71to signify a fairly major update. Or, we might want to jump to perl6.
72Let's worry about that problem when we get there.
73
aa689395 74=head2 Subversions
75
e04b929a 76In addition, there usually are sub-versions available. Sub-versions
77are numbered with sub-version numbers. For example, version 5.003_04
78is the 4'th developer version built on top of 5.003. It might include
79the _01, _02, and _03 changes, but it also might not. Sub-versions are
80allowed to be subversive. (But see the next section for recent
81changes.)
aa689395 82
83These sub-versions can also be used as floating point numbers, so
84you can do things such as
85
7b5757d1 86 print "You've got an unstable perl\n" if $] == 5.00303;
aa689395 87
88You can also require particular version (or later) with
89
7b5757d1 90 use 5.003_03; # the "_" is optional
aa689395 91
92Sub-versions produced by the members of perl5-porters are usually
e04b929a 93available on CPAN in the F<src/5.0/maint> and F<src/5.0/devel>
94directories.
aa689395 95
7b5757d1 96=head2 Maintenance and Development Subversions
97
e04b929a 98Starting with version 5.004, subversions _01 through _49 is reserved
99for bug-fix maintenance releases, and subversions _50 through _99 for
100unstable development versions.
7b5757d1 101
102The separate bug-fix track is being established to allow us an easy
103way to distribute important bug fixes without waiting for the
104developers to untangle all the other problems in the current
e04b929a 105developer's release. The first rule of maintenance work is "First, do
106no harm."
7b5757d1 107
fb73857a 108Trial releases of bug-fix maintenance releases are announced on
109perl5-porters. Trial releases use the new subversion number (to avoid
110testers installing it over the previous release) and include a 'local
e04b929a 111patch' entry in patchlevel.h. The distribution file contains the
112string C<MAINT_TRIAL> to make clear that the file is not meant for
113public consumption.
fb73857a 114
e04b929a 115In general, the names of official distribution files for the public
116always match the regular expression
7b5757d1 117
e04b929a 118 ^perl5\.\d{3}(_[0-4]\d)?\.tar\.gz$
119
120Developer releases always match
121
122 ^perl5\.\d{3}(_[5-9]\d)?\.tar\.gz$
123
124And the trial versions for a new maintainance release match
125
126 ^perl5\.\d{3}(_[0-4]\d)-MAINT_TRIAL_\d+\.tar\.gz$
127
128In the past it has been observed that pumkings tend to invent new
129naming conventions on the fly. If you are a pumpking, before you
130invent a new name for any of the three types of perl distributions,
131please inform the guys from the CPAN who are doing indexing and
132provide the trees of symlinks and the like. They will have to know
133I<in advance> what you decide.
20f245af 134
aa689395 135=head2 Why such a complicated scheme?
136
137Two reasons, really. At least.
138
7b5757d1 139First, we need some way to identify and release collections of patches
140that are known to have new features that need testing and exploration. The
aa689395 141subversion scheme does that nicely while fitting into the
142C<use 5.004;> mold.
143
144Second, since most of the folks who help maintain perl do so on a
145free-time voluntary basis, perl development does not proceed at a
146precise pace, though it always seems to be moving ahead quickly.
147We needed some way to pass around the "patch pumpkin" to allow
148different people chances to work on different aspects of the
149distribution without getting in each other's way. It wouldn't be
150constructive to have multiple people working on incompatible
151implementations of the same idea. Instead what was needed was
152some kind of "baton" or "token" to pass around so everyone knew
153whose turn was next.
154
155=head2 Why is it called the patch pumpkin?
156
157Chip Salzenberg gets credit for that, with a nod to his cow orker,
158David Croy. We had passed around various names (baton, token, hot
159potato) but none caught on. Then, Chip asked:
160
161[begin quote]
162
163 Who has the patch pumpkin?
164
165To explain: David Croy once told me once that at a previous job,
166there was one tape drive and multiple systems that used it for backups.
167But instead of some high-tech exclusion software, they used a low-tech
168method to prevent multiple simultaneous backups: a stuffed pumpkin.
169No one was allowed to make backups unless they had the "backup pumpkin".
170
171[end quote]
172
173The name has stuck.
174
a6968aa6 175=head1 Philosophical Issues in Patching and Porting Perl
aa689395 176
177There are no absolute rules, but there are some general guidelines I
178have tried to follow as I apply patches to the perl sources.
179(This section is still under construction.)
180
181=head2 Solve problems as generally as possible
182
7b5757d1 183Never implement a specific restricted solution to a problem when you
184can solve the same problem in a more general, flexible way.
185
186For example, for dynamic loading to work on some SVR4 systems, we had
187to build a shared libperl.so library. In order to build "FAT" binaries
188on NeXT 4.0 systems, we had to build a special libperl library. Rather
189than continuing to build a contorted nest of special cases, I
190generalized the process of building libperl so that NeXT and SVR4 users
191could still get their work done, but others could build a shared
192libperl if they wanted to as well.
aa689395 193
a6968aa6 194Contain your changes carefully. Assume nothing about other operating
195systems, not even closely related ones. Your changes must not affect
196other platforms.
197
198Spy shamelessly on how similar patching or porting issues have been
199settled elsewhere.
200
201If feasible, try to keep filenames 8.3-compliant to humor those poor
202souls that get joy from running Perl under such dire limitations.
203
aa689395 204=head2 Seek consensus on major changes
205
206If you are making big changes, don't do it in secret. Discuss the
207ideas in advance on perl5-porters.
208
209=head2 Keep the documentation up-to-date
210
211If your changes may affect how users use perl, then check to be sure
212that the documentation is in sync with your changes. Be sure to
213check all the files F<pod/*.pod> and also the F<INSTALL> document.
214
215Consider writing the appropriate documentation first and then
7b5757d1 216implementing your change to correspond to the documentation.
aa689395 217
218=head2 Avoid machine-specific #ifdef's
219
220To the extent reasonable, try to avoid machine-specific #ifdef's in
221the sources. Instead, use feature-specific #ifdef's. The reason is
222that the machine-specific #ifdef's may not be valid across major
223releases of the operating system. Further, the feature-specific tests
224may help out folks on another platform who have the same problem.
225
a6968aa6 226=head2 Machine-specific files
227
98dddfbd 228=over 4
229
230=item source code
231
a6968aa6 232If you have many machine-specific #defines or #includes, consider
233creating an "osish.h" (os2ish.h, vmsish.h, and so on) and including
234that in perl.h. If you have several machine-specific files (function
235emulations, function stubs, build utility wrappers) you may create a
236separate subdirectory (djgpp, win32) and put the files in there.
98dddfbd 237Remember to update C<MANIFEST> when you add files.
a6968aa6 238
98dddfbd 239If your system support dynamic loading but none of the existing
240methods at F<ext/DynaLoader/dl_*.xs> work for you, you must write
241a new one. Study the existing ones to see what kind of interface
242you must supply.
243
244=item build hints
a6968aa6 245
246There are two kinds of hints: hints for building Perl and hints for
247extensions. The former live in the C<hints> subdirectory, the latter
248in C<ext/*/hints> subdirectories.
249
250The top level hints are Bourne-shell scripts that set, modify and
251unset appropriate Configure variables, based on the Configure command
252line options and possibly existing config.sh and Policy.sh files from
253previous Configure runs.
254
255The extension hints are written Perl (by the time they are used
256miniperl has been built) and control the building of their respective
257extensions. They can be used to for example manipulate compilation
258and linking flags.
259
98dddfbd 260=item build and installation Makefiles, scripts, and so forth
261
262Sometimes you will also need to tweak the Perl build and installation
263procedure itself, like for example F<Makefile.SH> and F<installperl>.
264Tread very carefully, even more than usual. Contain your changes
265with utmost care.
a6968aa6 266
98dddfbd 267=item test suite
268
269Many of the tests in C<t> subdirectory assume machine-specific things
a6968aa6 270like existence of certain functions, something about filesystem
271semantics, certain external utilities and their error messages. Use
272the C<$^O> and the C<Config> module (which contains the results of the
273Configure run, in effect the C<config.sh> converted to Perl) to either
98dddfbd 274skip (preferably not) or customize (preferable) the tests for your
275platform.
276
277=item modules
278
279Certain standard modules may need updating if your operating system
280sports for example a native filesystem naming. You may want to update
281some or all of the modules File::Basename, File::Spec, File::Path, and
282File::Copy to become aware of your native filesystem syntax and
283peculiarities.
284
285=item documentation
286
287If your operating system comes from outside UNIX you almost certainly
288will have differences in the available operating system functionality
289(missing system calls, different semantics, whatever). Please
290document these at F<pod/perlport.pod>. If your operating system is
291the first B<not> to have a system call also update the list of
292"portability-bewares" at the beginning of F<pod/perlfunc.pod>.
293
294A file called F<README.youros> at the top level that explains things
295like how to install perl at this platform, where to get any possibly
296required additional software, and for example what test suite errors
297to expect, is nice too.
298
299You may also want to write a separate F<.pod> file for your operating
300system to tell about existing mailing lists, os-specific modules,
301documentation, whatever. Please name these along the lines of
302F<perl>I<youros>.pod. [unfinished: where to put this file (the pod/
303subdirectory, of course: but more importantly, which/what index files
304should be updated?)]
305
306=back
a6968aa6 307
aa689395 308=head2 Allow for lots of testing
309
310We should never release a main version without testing it as a
311subversion first.
312
6877a1cf 313=head2 Test popular applications and modules.
314
315We should never release a main version without testing whether or not
316it breaks various popular modules and applications. A partial list of
317such things would include majordomo, metaconfig, apache, Tk, CGI,
318libnet, and libwww, to name just a few. Of course it's quite possible
319that some of those things will be just plain broken and need to be fixed,
320but, in general, we ought to try to avoid breaking widely-installed
321things.
322
98dddfbd 323=head2 Automated generation of derivative files
aa689395 324
325The F<embed.h>, F<keywords.h>, F<opcode.h>, and F<perltoc.pod> files
326are all automatically generated by perl scripts. In general, don't
327patch these directly; patch the data files instead.
328
329F<Configure> and F<config_h.SH> are also automatically generated by
330B<metaconfig>. In general, you should patch the metaconfig units
a6968aa6 331instead of patching these files directly. However, very minor changes
332to F<Configure> may be made in between major sync-ups with the
333metaconfig units, which tends to be complicated operations. But be
334careful, this can quickly spiral out of control. Running metaconfig
335is not really hard.
aa689395 336
98dddfbd 337Also F<Makefile> is automatically produced from F<Makefile.SH>.
338In general, look out for all F<*.SH> files.
339
a8119d38 340Finally, the sample files in the F<Porting/> subdirectory are
341generated automatically by the script F<U/mksample> included
342with the metaconfig units. See L<"run metaconfig"> below for
343information on obtaining the metaconfig units.
344
aa689395 345=head1 How to Make a Distribution
346
347There really ought to be a 'make dist' target, but there isn't.
348The 'dist' suite of tools also contains a number of tools that I haven't
349learned how to use yet. Some of them may make this all a bit easier.
350
351Here are the steps I go through to prepare a patch & distribution.
352
3e3baf6d 353Lots of it could doubtless be automated but isn't. The Porting/makerel
354(make release) perl script does now help automate some parts of it.
aa689395 355
356=head2 Announce your intentions
357
358First, you should volunteer out loud to take the patch pumpkin. It's
359generally counter-productive to have multiple people working in secret
360on the same thing.
361
362At the same time, announce what you plan to do with the patch pumpkin,
363to allow folks a chance to object or suggest alternatives, or do it for
364you. Naturally, the patch pumpkin holder ought to incorporate various
365bug fixes and documentation improvements that are posted while he or
366she has the pumpkin, but there might also be larger issues at stake.
367
368One of the precepts of the subversion idea is that we shouldn't give
7b5757d1 369the patch pumpkin to anyone unless we have some idea what he or she
370is going to do with it.
aa689395 371
372=head2 refresh pod/perltoc.pod
373
374Presumably, you have done a full C<make> in your working source
375directory. Before you C<make spotless> (if you do), and if you have
376changed any documentation in any module or pod file, change to the
377F<pod> directory and run C<make toc>.
378
3e3baf6d 379=head2 run installhtml to check the validity of the pod files
380
aa689395 381=head2 update patchlevel.h
382
383Don't be shy about using the subversion number, even for a relatively
384modest patch. We've never even come close to using all 99 subversions,
385and it's better to have a distinctive number for your patch. If you
386need feedback on your patch, go ahead and issue it and promise to
387incorporate that feedback quickly (e.g. within 1 week) and send out a
388second patch.
389
390=head2 run metaconfig
391
392If you need to make changes to Configure or config_h.SH, it may be best to
393change the appropriate metaconfig units instead, and regenerate Configure.
394
395 metaconfig -m
396
20f245af 397will regenerate Configure and config_h.SH. Much more information
398on obtaining and running metaconfig is in the F<U/README> file
399that comes with Perl's metaconfig units. Perl's metaconfig units
400should be available on CPAN. A set of units that will work with
401perl5.005 is in the file F<mc_units-5.005_00-01.tar.gz> under
402http://www.perl.com/CPAN/authors/id/ANDYD/ . The mc_units tar file
403should be unpacked in your main perl source directory. Note: those
404units were for use with 5.005. There may have been changes since then.
d562869c 405Check for later versions or contact perl5-porters@perl.org to obtain a
20f245af 406pointer to the current version.
aa689395 407
408Alternatively, do consider if the F<*ish.h> files might be a better
409place for your changes.
410
411=head2 MANIFEST
412
413Make sure the MANIFEST is up-to-date. You can use dist's B<manicheck>
414program for this. You can also use
415
3e3baf6d 416 perl -w -MExtUtils::Manifest=fullcheck -e fullcheck
aa689395 417
3e3baf6d 418Both commands will also list extra files in the directory that are not
419listed in MANIFEST.
aa689395 420
bfb7748a 421The MANIFEST is normally sorted.
aa689395 422
423If you are using metaconfig to regenerate Configure, then you should note
424that metaconfig actually uses MANIFEST.new, so you want to be sure
425MANIFEST.new is up-to-date too. I haven't found the MANIFEST/MANIFEST.new
426distinction particularly useful, but that's probably because I still haven't
427learned how to use the full suite of tools in the dist distribution.
428
429=head2 Check permissions
430
431All the tests in the t/ directory ought to be executable. The
432main makefile used to do a 'chmod t/*/*.t', but that resulted in
433a self-modifying distribution--something some users would strongly
d562869c 434prefer to avoid. The F<t/TEST> script will check for this
435and do the chmod if needed, but the tests still ought to be
436executable.
aa689395 437
438In all, the following files should probably be executable:
439
440 Configure
441 configpm
32fcaa0b 442 configure.gnu
aa689395 443 embed.pl
444 installperl
445 installman
446 keywords.pl
aa689395 447 myconfig
448 opcode.pl
449 perly.fixer
450 t/TEST
451 t/*/*.t
452 *.SH
453 vms/ext/Stdio/test.pl
454 vms/ext/filespec.t
aa689395 455 x2p/*.SH
456
457Other things ought to be readable, at least :-).
458
459Probably, the permissions for the files could be encoded in MANIFEST
460somehow, but I'm reluctant to change MANIFEST itself because that
461could break old scripts that use MANIFEST.
462
463I seem to recall that some SVR3 systems kept some sort of file that listed
464permissions for system files; something like that might be appropriate.
465
466=head2 Run Configure
467
468This will build a config.sh and config.h. You can skip this if you haven't
693762b4 469changed Configure or config_h.SH at all. I use the following command
aa689395 470
693762b4 471 sh Configure -Dprefix=/opt/perl -Doptimize=-O -Dusethreads \
472 -Dcf_by='yourname' \
473 -Dcf_email='yourname@yourhost.yourplace.com' \
474 -Dperladmin='yourname@yourhost.yourplace.com' \
475 -Dmydomain='.yourplace.com' \
476 -Dmyhostname='yourhost' \
477 -des
aa689395 478
693762b4 479=head2 Update Porting/config.sh and Porting/config_H
dfe9444c 480
693762b4 481[XXX
482This section needs revision. We're currently working on easing
483the task of keeping the vms, win32, and plan9 config.sh info
484up-to-date. The plan is to use keep up-to-date 'canned' config.sh
485files in the appropriate subdirectories and then generate 'canned'
486config.h files for vms, win32, etc. from the generic config.sh file.
487This is to ease maintenance. When Configure gets updated, the parts
488sometimes get scrambled around, and the changes in config_H can
489sometimes be very hard to follow. config.sh, on the other hand, can
490safely be sorted, so it's easy to track (typically very small) changes
491to config.sh and then propoagate them to a canned 'config.h' by any
492number of means, including a perl script in win32/ or carrying
493config.sh and config_h.SH to a Unix system and running sh
494config_h.SH.)
495XXX]
496
497The Porting/config.sh and Porting/config_H files are provided to
498help those folks who can't run Configure. It is important to keep
499them up-to-date. If you have changed config_h.SH, those changes must
500be reflected in config_H as well. (The name config_H was chosen to
501distinguish the file from config.h even on case-insensitive file systems.)
502Simply edit the existing config_H file; keep the first few explanatory
503lines and then copy your new config.h below.
aa689395 504
d562869c 505It may also be necessary to update win32/config.?c, vms/config.vms and
aa689395 506plan9/config.plan9, though you should be quite careful in doing so if
507you are not familiar with those systems. You might want to issue your
508patch with a promise to quickly issue a follow-up that handles those
509directories.
510
511=head2 make run_byacc
512
513If you have byacc-1.8.2 (available from CPAN), and if there have been
514changes to F<perly.y>, you can regenerate the F<perly.c> file. The
515run_byacc makefile target does this by running byacc and then applying
516some patches so that byacc dynamically allocates space, rather than
517having fixed limits. This patch is handled by the F<perly.fixer>
518script. Depending on the nature of the changes to F<perly.y>, you may
519or may not have to hand-edit the patch to apply correctly. If you do,
520you should include the edited patch in the new distribution. If you
521have byacc-1.9, the patch won't apply cleanly. Changes to the printf
522output statements mean the patch won't apply cleanly. Long ago I
523started to fix F<perly.fixer> to detect this, but I never completed the
524task.
525
ebb99254 526If C<perly.c> changes, make sure you run C<perl vms/vms_yfix.pl> to
527update the corresponding VMS files. See L<VMS-specific updates>.
528
aa689395 529Some additional notes from Larry on this:
530
e262e9be 531Don't forget to regenerate perly_c.diff.
aa689395 532
7b5757d1 533 byacc -d perly.y
aa689395 534 mv y.tab.c perly.c
e262e9be 535 patch perly.c <perly_c.diff
aa689395 536 # manually apply any failed hunks
e262e9be 537 diff -c2 perly.c.orig perly.c >perly_c.diff
aa689395 538
539One chunk of lines that often fails begins with
540
541 #line 29 "perly.y"
542
543and ends one line before
544
545 #define YYERRCODE 256
546
547This only happens when you add or remove a token type. I suppose this
548could be automated, but it doesn't happen very often nowadays.
549
550Larry
551
552=head2 make regen_headers
553
554The F<embed.h>, F<keywords.h>, and F<opcode.h> files are all automatically
555generated by perl scripts. Since the user isn't guaranteed to have a
556working perl, we can't require the user to generate them. Hence you have
557to, if you're making a distribution.
558
559I used to include rules like the following in the makefile:
560
561 # The following three header files are generated automatically
562 # The correct versions should be already supplied with the perl kit,
563 # in case you don't have perl or 'sh' available.
564 # The - is to ignore error return codes in case you have the source
565 # installed read-only or you don't have perl yet.
566 keywords.h: keywords.pl
567 @echo "Don't worry if this fails."
568 - perl keywords.pl
569
570
7b5757d1 571However, I got B<lots> of mail consisting of people worrying because the
aa689395 572command failed. I eventually decided that I would save myself time
573and effort by manually running C<make regen_headers> myself rather
574than answering all the questions and complaints about the failing
575command.
576
3e3baf6d 577=head2 global.sym, interp.sym and perlio.sym
aa689395 578
579Make sure these files are up-to-date. Read the comments in these
580files and in perl_exp.SH to see what to do.
581
582=head2 Binary compatibility
583
584If you do change F<global.sym> or F<interp.sym>, think carefully about
585what you are doing. To the extent reasonable, we'd like to maintain
586souce and binary compatibility with older releases of perl. That way,
587extensions built under one version of perl will continue to work with
588new versions of perl.
589
590Of course, some incompatible changes may well be necessary. I'm just
591suggesting that we not make any such changes without thinking carefully
592about them first. If possible, we should provide
593backwards-compatibility stubs. There's a lot of XS code out there.
594Let's not force people to keep changing it.
595
596=head2 Changes
597
598Be sure to update the F<Changes> file. Try to include both an overall
599summary as well as detailed descriptions of the changes. Your
3e3baf6d 600audience will include other developers and users, so describe
aa689395 601user-visible changes (if any) in terms they will understand, not in
602code like "initialize foo variable in bar function".
603
604There are differing opinions on whether the detailed descriptions
605ought to go in the Changes file or whether they ought to be available
606separately in the patch file (or both). There is no disagreement that
607detailed descriptions ought to be easily available somewhere.
608
2a26e2f1 609=head2 Todo
610
611The F<Todo> file contains a roughly-catgorized unordered list of
612aspects of Perl that could use enhancement, features that could be
613added, areas that could be cleaned up, and so on. During your term as
614pumpkin-holder, you will probably address some of these issues, and
615perhaps identify others which, while you decide not to address them
616this time around, may be tackled in the future. Update the file
617reflect the situation as it stands when you hand over the pumpkin.
618
619You might like, early in your pumpkin-holding career, to see if you
620can find champions for partiticular issues on the to-do list: an issue
621owned is an issue more likely to be resolved.
622
c4f23d77 623There are also some more porting-specific L<Todo> items later in this
624file.
625
aa689395 626=head2 OS/2-specific updates
627
628In the os2 directory is F<diff.configure>, a set of OS/2-specific
629diffs against B<Configure>. If you make changes to Configure, you may
630want to consider regenerating this diff file to save trouble for the
631OS/2 maintainer.
632
7b5757d1 633You can also consider the OS/2 diffs as reminders of portability
634things that need to be fixed in Configure.
635
aa689395 636=head2 VMS-specific updates
637
ebb99254 638If you have changed F<perly.y> or F<perly.c>, then you most probably want
639to update F<vms/perly_{h,c}.vms> by running C<perl vms/vms_yfix.pl>.
aa689395 640
641The Perl version number appears in several places under F<vms>.
642It is courteous to update these versions. For example, if you are
643making 5.004_42, replace "5.00441" with "5.00442".
644
645=head2 Making the new distribution
646
647Suppose, for example, that you want to make version 5.004_08. Then you can
648do something like the following
649
650 mkdir ../perl5.004_08
651 awk '{print $1}' MANIFEST | cpio -pdm ../perl5.004_08
652 cd ../
653 tar cf perl5.004_08.tar perl5.004_08
654 gzip --best perl5.004_08.tar
655
3e3baf6d 656These steps, with extra checks, are automated by the Porting/makerel
657script.
658
aa689395 659=head2 Making a new patch
660
661I find the F<makepatch> utility quite handy for making patches.
662You can obtain it from any CPAN archive under
3e3baf6d 663http://www.perl.com/CPAN/authors/Johan_Vromans/ . There are a couple
664of differences between my version and the standard one. I have mine do
665a
aa689395 666
667 # Print a reassuring "End of Patch" note so people won't
668 # wonder if their mailer truncated patches.
669 print "\n\nEnd of Patch.\n";
670
3e3baf6d 671at the end. That's because I used to get questions from people asking
672if their mail was truncated.
673
674It also writes Index: lines which include the new directory prefix
675(change Index: print, approx line 294 or 310 depending on the version,
676to read: print PATCH ("Index: $newdir$new\n");). That helps patches
677work with more POSIX conformant patch programs.
aa689395 678
679Here's how I generate a new patch. I'll use the hypothetical
6805.004_07 to 5.004_08 patch as an example.
681
682 # unpack perl5.004_07/
683 gzip -d -c perl5.004_07.tar.gz | tar -xof -
684 # unpack perl5.004_08/
685 gzip -d -c perl5.004_08.tar.gz | tar -xof -
686 makepatch perl5.004_07 perl5.004_08 > perl5.004_08.pat
687
688Makepatch will automatically generate appropriate B<rm> commands to remove
689deleted files. Unfortunately, it will not correctly set permissions
690for newly created files, so you may have to do so manually. For example,
691patch 5.003_04 created a new test F<t/op/gv.t> which needs to be executable,
692so at the top of the patch, I inserted the following lines:
693
694 # Make a new test
695 touch t/op/gv.t
696 chmod +x t/opt/gv.t
697
698Now, of course, my patch is now wrong because makepatch didn't know I
699was going to do that command, and it patched against /dev/null.
700
701So, what I do is sort out all such shell commands that need to be in the
702patch (including possible mv-ing of files, if needed) and put that in the
703shell commands at the top of the patch. Next, I delete all the patch parts
704of perl5.004_08.pat, leaving just the shell commands. Then, I do the
705following:
706
7b5757d1 707 cd perl5.004_07
708 sh ../perl5.004_08.pat
aa689395 709 cd ..
7b5757d1 710 makepatch perl5.004_07 perl5.004_08 >> perl5.004_08.pat
aa689395 711
712(Note the append to preserve my shell commands.)
713Now, my patch will line up with what the end users are going to do.
714
715=head2 Testing your patch
716
717It seems obvious, but be sure to test your patch. That is, verify that
718it produces exactly the same thing as your full distribution.
719
7b5757d1 720 rm -rf perl5.004_07
721 gzip -d -c perl5.004_07.tar.gz | tar -xf -
722 cd perl5.004_07
723 sh ../perl5.004_08.pat
724 patch -p1 -N < ../perl5.004_08.pat
aa689395 725 cd ..
7b5757d1 726 gdiff -r perl5.004_07 perl5.004_08
aa689395 727
728where B<gdiff> is GNU diff. Other diff's may also do recursive checking.
729
730=head2 More testing
731
732Again, it's obvious, but you should test your new version as widely as you
733can. You can be sure you'll hear about it quickly if your version doesn't
734work on both ANSI and pre-ANSI compilers, and on common systems such as
735SunOS 4.1.[34], Solaris, and Linux.
736
737If your changes include conditional code, try to test the different
738branches as thoroughly as you can. For example, if your system
739supports dynamic loading, you can also test static loading with
740
741 sh Configure -Uusedl
742
743You can also hand-tweak your config.h to try out different #ifdef
744branches.
745
746=head1 Common Gotcha's
747
748=over 4
749
750=item #elif
751
752The '#elif' preprocessor directive is not understood on all systems.
753Specifically, I know that Pyramids don't understand it. Thus instead of the
754simple
755
756 #if defined(I_FOO)
757 # include <foo.h>
758 #elif defined(I_BAR)
759 # include <bar.h>
760 #else
761 # include <fubar.h>
762 #endif
763
764You have to do the more Byzantine
765
766 #if defined(I_FOO)
767 # include <foo.h>
768 #else
769 # if defined(I_BAR)
770 # include <bar.h>
771 # else
772 # include <fubar.h>
773 # endif
774 #endif
775
776Incidentally, whitespace between the leading '#' and the preprocessor
777command is not guaranteed, but is very portable and you may use it freely.
778I think it makes things a bit more readable, especially once things get
779rather deeply nested. I also think that things should almost never get
780too deeply nested, so it ought to be a moot point :-)
781
782=item Probably Prefer POSIX
783
784It's often the case that you'll need to choose whether to do
785something the BSD-ish way or the POSIX-ish way. It's usually not
786a big problem when the two systems use different names for similar
787functions, such as memcmp() and bcmp(). The perl.h header file
788handles these by appropriate #defines, selecting the POSIX mem*()
789functions if available, but falling back on the b*() functions, if
790need be.
791
792More serious is the case where some brilliant person decided to
793use the same function name but give it a different meaning or
794calling sequence :-). getpgrp() and setpgrp() come to mind.
795These are a real problem on systems that aim for conformance to
796one standard (e.g. POSIX), but still try to support the other way
797of doing things (e.g. BSD). My general advice (still not really
798implemented in the source) is to do something like the following.
799Suppose there are two alternative versions, fooPOSIX() and
800fooBSD().
801
802 #ifdef HAS_FOOPOSIX
803 /* use fooPOSIX(); */
804 #else
805 # ifdef HAS_FOOBSD
806 /* try to emulate fooPOSIX() with fooBSD();
807 perhaps with the following: */
808 # define fooPOSIX fooBSD
809 # else
810 # /* Uh, oh. We have to supply our own. */
811 # define fooPOSIX Perl_fooPOSIX
812 # endif
813 #endif
814
815=item Think positively
816
817If you need to add an #ifdef test, it is usually easier to follow if you
818think positively, e.g.
819
820 #ifdef HAS_NEATO_FEATURE
821 /* use neato feature */
822 #else
823 /* use some fallback mechanism */
824 #endif
825
826rather than the more impenetrable
827
828 #ifndef MISSING_NEATO_FEATURE
829 /* Not missing it, so we must have it, so use it */
830 #else
831 /* Are missing it, so fall back on something else. */
832 #endif
833
834Of course for this toy example, there's not much difference. But when
835the #ifdef's start spanning a couple of screen fulls, and the #else's
836are marked something like
837
838 #else /* !MISSING_NEATO_FEATURE */
839
840I find it easy to get lost.
841
842=item Providing Missing Functions -- Problem
843
844Not all systems have all the neat functions you might want or need, so
845you might decide to be helpful and provide an emulation. This is
846sound in theory and very kind of you, but please be careful about what
847you name the function. Let me use the C<pause()> function as an
848illustration.
849
850Perl5.003 has the following in F<perl.h>
851
852 #ifndef HAS_PAUSE
853 #define pause() sleep((32767<<16)+32767)
854 #endif
855
856Configure sets HAS_PAUSE if the system has the pause() function, so
857this #define only kicks in if the pause() function is missing.
858Nice idea, right?
859
860Unfortunately, some systems apparently have a prototype for pause()
861in F<unistd.h>, but don't actually have the function in the library.
862(Or maybe they do have it in a library we're not using.)
863
864Thus, the compiler sees something like
865
866 extern int pause(void);
867 /* . . . */
868 #define pause() sleep((32767<<16)+32767)
869
870and dies with an error message. (Some compilers don't mind this;
871others apparently do.)
872
873To work around this, 5.003_03 and later have the following in perl.h:
874
875 /* Some unistd.h's give a prototype for pause() even though
876 HAS_PAUSE ends up undefined. This causes the #define
877 below to be rejected by the compiler. Sigh.
878 */
879 #ifdef HAS_PAUSE
880 # define Pause pause
881 #else
882 # define Pause() sleep((32767<<16)+32767)
883 #endif
884
885This works.
886
887The curious reader may wonder why I didn't do the following in
888F<util.c> instead:
889
890 #ifndef HAS_PAUSE
891 void pause()
892 {
893 sleep((32767<<16)+32767);
894 }
895 #endif
896
897That is, since the function is missing, just provide it.
898Then things would probably be been alright, it would seem.
899
900Well, almost. It could be made to work. The problem arises from the
901conflicting needs of dynamic loading and namespace protection.
902
903For dynamic loading to work on AIX (and VMS) we need to provide a list
904of symbols to be exported. This is done by the script F<perl_exp.SH>,
905which reads F<global.sym> and F<interp.sym>. Thus, the C<pause>
906symbol would have to be added to F<global.sym> So far, so good.
907
908On the other hand, one of the goals of Perl5 is to make it easy to
909either extend or embed perl and link it with other libraries. This
910means we have to be careful to keep the visible namespace "clean".
911That is, we don't want perl's global variables to conflict with
912those in the other application library. Although this work is still
913in progress, the way it is currently done is via the F<embed.h> file.
914This file is built from the F<global.sym> and F<interp.sym> files,
915since those files already list the globally visible symbols. If we
916had added C<pause> to global.sym, then F<embed.h> would contain the
917line
918
919 #define pause Perl_pause
920
921and calls to C<pause> in the perl sources would now point to
922C<Perl_pause>. Now, when B<ld> is run to build the F<perl> executable,
923it will go looking for C<perl_pause>, which probably won't exist in any
924of the standard libraries. Thus the build of perl will fail.
925
926Those systems where C<HAS_PAUSE> is not defined would be ok, however,
927since they would get a C<Perl_pause> function in util.c. The rest of
928the world would be in trouble.
929
930And yes, this scenario has happened. On SCO, the function C<chsize>
931is available. (I think it's in F<-lx>, the Xenix compatibility
932library.) Since the perl4 days (and possibly before), Perl has
933included a C<chsize> function that gets called something akin to
934
935 #ifndef HAS_CHSIZE
936 I32 chsize(fd, length)
937 /* . . . */
938 #endif
939
940When 5.003 added
941
942 #define chsize Perl_chsize
943
944to F<embed.h>, the compile started failing on SCO systems.
945
946The "fix" is to give the function a different name. The one
947implemented in 5.003_05 isn't optimal, but here's what was done:
948
949 #ifdef HAS_CHSIZE
950 # ifdef my_chsize /* Probably #defined to Perl_my_chsize in embed.h */
951 # undef my_chsize
952 # endif
953 # define my_chsize chsize
954 #endif
955
956My explanatory comment in patch 5.003_05 said:
957
958 Undef and then re-define my_chsize from Perl_my_chsize to
959 just plain chsize if this system HAS_CHSIZE. This probably only
960 applies to SCO. This shows the perils of having internal
961 functions with the same name as external library functions :-).
962
963Now, we can safely put C<my_chsize> in F<global.sym>, export it, and
964hide it with F<embed.h>.
965
966To be consistent with what I did for C<pause>, I probably should have
967called the new function C<Chsize>, rather than C<my_chsize>.
968However, the perl sources are quite inconsistent on this (Consider
969New, Mymalloc, and Myremalloc, to name just a few.)
970
971There is a problem with this fix, however, in that C<Perl_chsize>
972was available as a F<libperl.a> library function in 5.003, but it
973isn't available any more (as of 5.003_07). This means that we've
974broken binary compatibility. This is not good.
975
976=item Providing missing functions -- some ideas
977
978We currently don't have a standard way of handling such missing
979function names. Right now, I'm effectively thinking aloud about a
980solution. Some day, I'll try to formally propose a solution.
981
982Part of the problem is that we want to have some functions listed as
983exported but not have their names mangled by embed.h or possibly
984conflict with names in standard system headers. We actually already
985have such a list at the end of F<perl_exp.SH> (though that list is
986out-of-date):
987
988 # extra globals not included above.
989 cat <<END >> perl.exp
990 perl_init_ext
991 perl_init_fold
992 perl_init_i18nl14n
993 perl_alloc
994 perl_construct
995 perl_destruct
996 perl_free
997 perl_parse
998 perl_run
999 perl_get_sv
1000 perl_get_av
1001 perl_get_hv
1002 perl_get_cv
1003 perl_call_argv
1004 perl_call_pv
1005 perl_call_method
1006 perl_call_sv
1007 perl_requirepv
1008 safecalloc
1009 safemalloc
1010 saferealloc
1011 safefree
1012
1013This still needs much thought, but I'm inclined to think that one
1014possible solution is to prefix all such functions with C<perl_> in the
1015source and list them along with the other C<perl_*> functions in
1016F<perl_exp.SH>.
1017
1018Thus, for C<chsize>, we'd do something like the following:
1019
1020 /* in perl.h */
1021 #ifdef HAS_CHSIZE
1022 # define perl_chsize chsize
1023 #endif
1024
1025then in some file (e.g. F<util.c> or F<doio.c>) do
1026
1027 #ifndef HAS_CHSIZE
1028 I32 perl_chsize(fd, length)
1029 /* implement the function here . . . */
1030 #endif
1031
1032Alternatively, we could just always use C<chsize> everywhere and move
1033C<chsize> from F<global.sym> to the end of F<perl_exp.SH>. That would
1034probably be fine as long as our C<chsize> function agreed with all the
1035C<chsize> function prototypes in the various systems we'll be using.
1036As long as the prototypes in actual use don't vary that much, this is
1037probably a good alternative. (As a counter-example, note how Configure
1038and perl have to go through hoops to find and use get Malloc_t and
1039Free_t for C<malloc> and C<free>.)
1040
1041At the moment, this latter option is what I tend to prefer.
1042
1043=item All the world's a VAX
1044
1045Sorry, showing my age:-). Still, all the world is not BSD 4.[34],
1046SVR4, or POSIX. Be aware that SVR3-derived systems are still quite
1047common (do you have any idea how many systems run SCO?) If you don't
1048have a bunch of v7 manuals handy, the metaconfig units (by default
1049installed in F</usr/local/lib/dist/U>) are a good resource to look at
1050for portability.
1051
1052=back
1053
1054=head1 Miscellaneous Topics
1055
1056=head2 Autoconf
1057
1058Why does perl use a metaconfig-generated Configure script instead of an
1059autoconf-generated configure script?
1060
1061Metaconfig and autoconf are two tools with very similar purposes.
1062Metaconfig is actually the older of the two, and was originally written
1063by Larry Wall, while autoconf is probably now used in a wider variety of
1064packages. The autoconf info file discusses the history of autoconf and
1065how it came to be. The curious reader is referred there for further
1066information.
1067
1068Overall, both tools are quite good, I think, and the choice of which one
1069to use could be argued either way. In March, 1994, when I was just
1070starting to work on Configure support for Perl5, I considered both
1071autoconf and metaconfig, and eventually decided to use metaconfig for the
1072following reasons:
1073
1074=over 4
1075
1076=item Compatibility with Perl4
1077
1078Perl4 used metaconfig, so many of the #ifdef's were already set up for
1079metaconfig. Of course metaconfig had evolved some since Perl4's days,
1080but not so much that it posed any serious problems.
1081
1082=item Metaconfig worked for me
1083
1084My system at the time was Interactive 2.2, a SVR3.2/386 derivative that
1085also had some POSIX support. Metaconfig-generated Configure scripts
1086worked fine for me on that system. On the other hand, autoconf-generated
1087scripts usually didn't. (They did come quite close, though, in some
1088cases.) At the time, I actually fetched a large number of GNU packages
1089and checked. Not a single one configured and compiled correctly
1090out-of-the-box with the system's cc compiler.
1091
1092=item Configure can be interactive
1093
1094With both autoconf and metaconfig, if the script works, everything is
1095fine. However, one of my main problems with autoconf-generated scripts
1096was that if it guessed wrong about something, it could be B<very> hard to
1097go back and fix it. For example, autoconf always insisted on passing the
1098-Xp flag to cc (to turn on POSIX behavior), even when that wasn't what I
1099wanted or needed for that package. There was no way short of editing the
1100configure script to turn this off. You couldn't just edit the resulting
1101Makefile at the end because the -Xp flag influenced a number of other
1102configure tests.
1103
1104Metaconfig's Configure scripts, on the other hand, can be interactive.
1105Thus if Configure is guessing things incorrectly, you can go back and fix
1106them. This isn't as important now as it was when we were actively
1107developing Configure support for new features such as dynamic loading,
1108but it's still useful occasionally.
1109
1110=item GPL
1111
1112At the time, autoconf-generated scripts were covered under the GNU Public
1113License, and hence weren't suitable for inclusion with Perl, which has a
1114different licensing policy. (Autoconf's licensing has since changed.)
1115
1116=item Modularity
1117
1118Metaconfig builds up Configure from a collection of discrete pieces
1119called "units". You can override the standard behavior by supplying your
1120own unit. With autoconf, you have to patch the standard files instead.
1121I find the metaconfig "unit" method easier to work with. Others
1122may find metaconfig's units clumsy to work with.
1123
1124=back
1125
1126=head2 @INC search order
1127
1128By default, the list of perl library directories in @INC is the
1129following:
1130
1131 $archlib
1132 $privlib
1133 $sitearch
1134 $sitelib
1135
1136Specifically, on my Solaris/x86 system, I run
1137B<sh Configure -Dprefix=/opt/perl> and I have the following
1138directories:
1139
1140 /opt/perl/lib/i86pc-solaris/5.00307
1141 /opt/perl/lib
1142 /opt/perl/lib/site_perl/i86pc-solaris
1143 /opt/perl/lib/site_perl
1144
1145That is, perl's directories come first, followed by the site-specific
1146directories.
1147
1148The site libraries come second to support the usage of extensions
1149across perl versions. Read the relevant section in F<INSTALL> for
1150more information. If we ever make $sitearch version-specific, this
1151topic could be revisited.
1152
1153=head2 Why isn't there a directory to override Perl's library?
1154
1155Mainly because no one's gotten around to making one. Note that
1156"making one" involves changing perl.c, Configure, config_h.SH (and
1157associated files, see above), and I<documenting> it all in the
1158INSTALL file.
1159
1160Apparently, most folks who want to override one of the standard library
1161files simply do it by overwriting the standard library files.
1162
1163=head2 APPLLIB
1164
1165In the perl.c sources, you'll find an undocumented APPLLIB_EXP
1166variable, sort of like PRIVLIB_EXP and ARCHLIB_EXP (which are
1167documented in config_h.SH). Here's what APPLLIB_EXP is for, from
1168a mail message from Larry:
1169
1170 The main intent of APPLLIB_EXP is for folks who want to send out a
1171 version of Perl embedded in their product. They would set the symbol
1172 to be the name of the library containing the files needed to run or to
1173 support their particular application. This works at the "override"
1174 level to make sure they get their own versions of any library code that
1175 they absolutely must have configuration control over.
1176
1177 As such, I don't see any conflict with a sysadmin using it for a
1178 override-ish sort of thing, when installing a generic Perl. It should
1179 probably have been named something to do with overriding though. Since
1180 it's undocumented we could still change it... :-)
1181
1182Given that it's already there, you can use it to override
1183distribution modules. If you do
1184
1185 sh Configure -Dccflags='-DAPPLLIB_EXP=/my/override'
1186
1187then perl.c will put /my/override ahead of ARCHLIB and PRIVLIB.
1188
c4f23d77 1189=head2 Shared libperl.so location
1190
1191Why isn't the shared libperl.so installed in /usr/lib/ along
1192with "all the other" shared libraries? Instead, it is installed
1193in $archlib, which is typically something like
1194
1195 /usr/local/lib/perl5/archname/5.00404
1196
1197and is architecture- and version-specific.
1198
1199The basic reason why a shared libperl.so gets put in $archlib is so that
1200you can have more than one version of perl on the system at the same time,
1201and have each refer to its own libperl.so.
1202
1203Three examples might help. All of these work now; none would work if you
1204put libperl.so in /usr/lib.
1205
1206=over
1207
1208=item 1.
1209
1210Suppose you want to have both threaded and non-threaded perl versions
1211around. Configure will name both perl libraries "libperl.so" (so that
1212you can link to them with -lperl). The perl binaries tell them apart
1213by having looking in the appropriate $archlib directories.
1214
1215=item 2.
1216
1217Suppose you have perl5.004_04 installed and you want to try to compile
1218it again, perhaps with different options or after applying a patch.
1219If you already have libperl.so installed in /usr/lib/, then it may be
1220either difficult or impossible to get ld.so to find the new libperl.so
1221that you're trying to build. If, instead, libperl.so is tucked away in
1222$archlib, then you can always just change $archlib in the current perl
1223you're trying to build so that ld.so won't find your old libperl.so.
1224(The INSTALL file suggests you do this when building a debugging perl.)
1225
1226=item 3.
1227
1228The shared perl library is not a "well-behaved" shared library with
1229proper major and minor version numbers, so you can't necessarily
1230have perl5.004_04 and perl5.004_05 installed simultaneously. Suppose
1231perl5.004_04 were to install /usr/lib/libperl.so.4.4, and perl5.004_05
1232were to install /usr/lib/libperl.so.4.5. Now, when you try to run
1233perl5.004_04, ld.so might try to load libperl.so.4.5, since it has
1234the right "major version" number. If this works at all, it almost
1235certainly defeats the reason for keeping perl5.004_04 around. Worse,
1236with development subversions, you certaily can't guarantee that
1237libperl.so.4.4 and libperl.so.4.55 will be compatible.
1238
1239Anyway, all this leads to quite obscure failures that are sure to drive
1240casual users crazy. Even experienced users will get confused :-). Upon
1241reflection, I'd say leave libperl.so in $archlib.
1242
1243=back
1244
aa689395 1245=head1 Upload Your Work to CPAN
1246
1247You can upload your work to CPAN if you have a CPAN id. Check out
1248http://www.perl.com/CPAN/modules/04pause.html for information on
1249_PAUSE_, the Perl Author's Upload Server.
1250
1251I typically upload both the patch file, e.g. F<perl5.004_08.pat.gz>
1252and the full tar file, e.g. F<perl5.004_08.tar.gz>.
1253
1254If you want your patch to appear in the F<src/5.0/unsupported>
1255directory on CPAN, send e-mail to the CPAN master librarian. (Check
7b5757d1 1256out http://www.perl.com/CPAN/CPAN.html ).
aa689395 1257
1258=head1 Help Save the World
1259
1260You should definitely announce your patch on the perl5-porters list.
1261You should also consider announcing your patch on
1262comp.lang.perl.announce, though you should make it quite clear that a
1263subversion is not a production release, and be prepared to deal with
1264people who will not read your disclaimer.
1265
1266=head1 Todo
1267
1268Here, in no particular order, are some Configure and build-related
1269items that merit consideration. This list isn't exhaustive, it's just
1270what I came up with off the top of my head.
1271
1272=head2 Good ideas waiting for round tuits
1273
1274=over 4
1275
1276=item installprefix
1277
1278I think we ought to support
1279
1280 Configure -Dinstallprefix=/blah/blah
1281
1282Currently, we support B<-Dprefix=/blah/blah>, but the changing the install
1283location has to be handled by something like the F<config.over> trick
1284described in F<INSTALL>. AFS users also are treated specially.
1285We should probably duplicate the metaconfig prefix stuff for an
1286install prefix.
1287
c4f23d77 1288=item Configure -Dsrc=/blah/blah
aa689395 1289
1290We should be able to emulate B<configure --srcdir>. Tom Tromey
1291tromey@creche.cygnus.com has submitted some patches to
c4f23d77 1292the dist-users mailing list along these lines. They have been folded
1293back into the main distribution, but various parts of the perl
1294Configure/build/install process still assume src='.'.
aa689395 1295
9cc29783 1296=item Directory for vendor-supplied modules?
1297
1298If a vendor supplies perl, but wants to leave $siteperl and $sitearch
1299for the local user to use, where should the vendor put vendor-supplied
67a4fa72 1300modules (such as Tk.so)? If the vendor puts them in the default $archlib,
1301then they need to be updated each time the perl version is updated.
1302Perhaps we need a set of libries $vendorlib and $vendorarch that
1303track $apiversion (like the $sitexxx directories do) rather than just
1304$version (like the main perl directory).
1305
1306An alternative (and perhaps even better) plan might be for the vendor
1307to select non-default $privlib and $archlib directories, perhaps using
1308$apiversion instead of $version (or even just /usr/lib/perl5 with no
1309version stuff at all), and put modules into those directories (with perl
1310Makefile.PL INSTALLDIRS=perl). This would be fine unless the vendor
1311wanted to support different versions of perl installed at the same time.
1312(How many vendors *really* want to do that?)
1313
1314=item Separate directories for Perl-supplied and add-on man pages
1315
1316Man pages supplied with the perl distribution proper ought to go in
1317an appropriate man directory. Perhaps man pages supplied with add-on
1318modules ought to (at least optionally) go into a $siteman[1-9] directory.
1319For example, suppose that $privlib is /usr/lib/perl5 and $man1dir
1320is /usr/man/man1. Also, suppose $sitelib is /usr/local/lib/perl5.
1321In this situation, it might make sense for man pages to go into
1322/usr/local/lib/man/man1.
9cc29783 1323
aa689395 1324=item Hint file fixes
1325
1326Various hint files work around Configure problems. We ought to fix
1327Configure so that most of them aren't needed.
1328
1329=item Hint file information
1330
1331Some of the hint file information (particularly dynamic loading stuff)
1332ought to be fed back into the main metaconfig distribution.
1333
c4f23d77 1334=item Catch GNU Libc "Stub" functions
1335
1336Some functions (such as lchown()) are present in libc, but are
1337unimplmented. That is, they always fail and set errno=ENOSYS.
1338
1339Thomas Bushnell provided the following sample code and the explanation
1340that follows:
1341
1342 /* System header to define __stub macros and hopefully few prototypes,
1343 which can conflict with char FOO(); below. */
1344 #include <assert.h>
1345 /* Override any gcc2 internal prototype to avoid an error. */
1346 /* We use char because int might match the return type of a gcc2
1347 builtin and then its argument prototype would still apply. */
1348 char FOO();
1349
1350 int main() {
1351
1352 /* The GNU C library defines this for functions which it implements
1353 to always fail with ENOSYS. Some functions are actually named
1354 something starting with __ and the normal name is an alias. */
1355 #if defined (__stub_FOO) || defined (__stub___FOO)
1356 choke me
1357 #else
1358 FOO();
1359 #endif
1360
1361 ; return 0; }
1362
1363The choice of <assert.h> is essentially arbitrary. The GNU libc
1364macros are found in <gnu/stubs.h>. You can include that file instead
1365of <assert.h> (which itself includes <gnu/stubs.h>) if you test for
1366its existence first. <assert.h> is assumed to exist on every system,
1367which is why it's used here. Any GNU libc header file will include
1368the stubs macros. If either __stub_NAME or __stub___NAME is defined,
1369then the function doesn't actually exist. Tests using <assert.h> work
1370on every system around.
1371
1372The declaration of FOO is there to override builtin prototypes for
1373ANSI C functions.
1374
aa689395 1375=back
1376
1377=head2 Probably good ideas waiting for round tuits
1378
1379=over 4
1380
1381=item GNU configure --options
1382
1383I've received sensible suggestions for --exec_prefix and other
1384GNU configure --options. It's not always obvious exactly what is
1385intended, but this merits investigation.
1386
1387=item make clean
1388
1389Currently, B<make clean> isn't all that useful, though
1390B<make realclean> and B<make distclean> are. This needs a bit of
1391thought and documentation before it gets cleaned up.
1392
1393=item Try gcc if cc fails
1394
1395Currently, we just give up.
1396
1397=item bypassing safe*alloc wrappers
1398
1399On some systems, it may be safe to call the system malloc directly
1400without going through the util.c safe* layers. (Such systems would
1401accept free(0), for example.) This might be a time-saver for systems
1402that already have a good malloc. (Recent Linux libc's apparently have
1403a nice malloc that is well-tuned for the system.)
1404
1405=back
1406
1407=head2 Vague possibilities
1408
1409=over 4
1410
aa689395 1411=item MacPerl
1412
3e3baf6d 1413Get some of the Macintosh stuff folded back into the main distribution.
aa689395 1414
1415=item gconvert replacement
1416
1417Maybe include a replacement function that doesn't lose data in rare
1418cases of coercion between string and numerical values.
1419
aa689395 1420=item Improve makedepend
1421
1422The current makedepend process is clunky and annoyingly slow, but it
1423works for most folks. Alas, it assumes that there is a filename
1424$firstmakefile that the B<make> command will try to use before it uses
1425F<Makefile>. Such may not be the case for all B<make> commands,
1426particularly those on non-Unix systems.
1427
1428Probably some variant of the BSD F<.depend> file will be useful.
1429We ought to check how other packages do this, if they do it at all.
1430We could probably pre-generate the dependencies (with the exception of
1431malloc.o, which could probably be determined at F<Makefile.SH>
1432extraction time.
1433
1434=item GNU Makefile standard targets
1435
1436GNU software generally has standardized Makefile targets. Unless we
1437have good reason to do otherwise, I see no reason not to support them.
1438
1439=item File locking
1440
1441Somehow, straighten out, document, and implement lockf(), flock(),
1442and/or fcntl() file locking. It's a mess.
1443
1444=back
1445
fb73857a 1446=head1 AUTHORS
aa689395 1447
fb73857a 1448Original author: Andy Dougherty doughera@lafcol.lafayette.edu .
1449Additions by Chip Salzenberg chip@perl.com and
1450Tim Bunce Tim.Bunce@ig.co.uk .
aa689395 1451
1452All opinions expressed herein are those of the authorZ<>(s).
1453
1454=head1 LAST MODIFIED
1455
20f245af 1456$Id: pumpkin.pod,v 1.22 1998/07/22 16:33:55 doughera Released $