This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
minor tweaks in Porting/pumpkin.pod
[perl5.git] / Porting / pumpkin.pod
CommitLineData
aa689395 1=head1 NAME
2
e25f343d 3Pumpkin - Notes on handling the Perl Patch Pumpkin And Porting Perl
aa689395 4
5=head1 SYNOPSIS
6
7There is no simple synopsis, yet.
8
9=head1 DESCRIPTION
10
98dddfbd
JH
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
a93751fa 27http://www.cpan.org/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
f38c94f4 46 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
fb73857a 47
aa689395 48=head1 How are Perl Releases Numbered?
49
f5a32c7f
GS
50Beginning with v5.6.0, even versions will stand for maintenance releases
51and odd versions for development releases, i.e., v5.6.x for maintenance
52releases, and v5.7.x for development releases. Before v5.6.0, subversions
53_01 through _49 were reserved for bug-fix maintenance releases, and
54subversions _50 through _99 for unstable development versions.
7b5757d1 55
f5a32c7f
GS
56For example, in v5.6.1, the revision number is 5, the version is 6,
57and 1 is the subversion.
aa689395 58
f5a32c7f
GS
59For compatibility with the older numbering scheme the composite floating
60point version number continues to be available as the magic variable $],
76ba0908 61and amounts to C<$revision + $version/1000 + $subversion/100000>. This
f5a32c7f 62can still be used in comparisons.
aa689395 63
f5a32c7f 64 print "You've got an old perl\n" if $] < 5.005_03;
aa689395 65
f5a32c7f 66In addition, the version is also available as a string in $^V.
aa689395 67
f5a32c7f 68 print "You've got a new perl\n" if $^V and $^V ge v5.6.0;
7b5757d1 69
f5a32c7f 70You can also require particular version (or later) with:
aa689395 71
f5a32c7f 72 use 5.006;
aa689395 73
f5a32c7f 74or using the new syntax available only from v5.6 onward:
aa689395 75
f5a32c7f 76 use v5.6.0;
aa689395 77
f5a32c7f
GS
78At some point in the future, we may need to decide what to call the
79next big revision. In the .package file used by metaconfig to
80generate Configure, there are two variables that might be relevant:
81$baserev=5 and $package=perl5.
aa689395 82
f5a32c7f 83Perl releases produced by the members of perl5-porters are usually
e04b929a
GS
84available on CPAN in the F<src/5.0/maint> and F<src/5.0/devel>
85directories.
aa689395 86
7b5757d1
AD
87=head2 Maintenance and Development Subversions
88
f5a32c7f 89The first rule of maintenance work is "First, do no harm."
7b5757d1 90
fb73857a 91Trial releases of bug-fix maintenance releases are announced on
92perl5-porters. Trial releases use the new subversion number (to avoid
93testers installing it over the previous release) and include a 'local
e04b929a
GS
94patch' entry in patchlevel.h. The distribution file contains the
95string C<MAINT_TRIAL> to make clear that the file is not meant for
96public consumption.
fb73857a 97
e04b929a 98In general, the names of official distribution files for the public
f5a32c7f 99always match the regular expression:
e04b929a 100
f5a32c7f 101 ^perl\d+\.(\d+)\.\d+(-MAINT_TRIAL_\d+)\.tar\.gz$
e04b929a 102
f5a32c7f
GS
103C<$1> in the pattern is always an even number for maintenance
104versions, and odd for developer releases.
e04b929a 105
efc41c8e 106In the past it has been observed that pumpkings tend to invent new
e04b929a
GS
107naming conventions on the fly. If you are a pumpking, before you
108invent a new name for any of the three types of perl distributions,
109please inform the guys from the CPAN who are doing indexing and
110provide the trees of symlinks and the like. They will have to know
111I<in advance> what you decide.
20f245af 112
aa689395 113=head2 Why is it called the patch pumpkin?
114
115Chip Salzenberg gets credit for that, with a nod to his cow orker,
116David Croy. We had passed around various names (baton, token, hot
117potato) but none caught on. Then, Chip asked:
118
119[begin quote]
120
121 Who has the patch pumpkin?
122
123To explain: David Croy once told me once that at a previous job,
124there was one tape drive and multiple systems that used it for backups.
125But instead of some high-tech exclusion software, they used a low-tech
126method to prevent multiple simultaneous backups: a stuffed pumpkin.
127No one was allowed to make backups unless they had the "backup pumpkin".
128
129[end quote]
130
131The name has stuck.
132
a6968aa6 133=head1 Philosophical Issues in Patching and Porting Perl
aa689395 134
135There are no absolute rules, but there are some general guidelines I
136have tried to follow as I apply patches to the perl sources.
137(This section is still under construction.)
138
139=head2 Solve problems as generally as possible
140
7b5757d1
AD
141Never implement a specific restricted solution to a problem when you
142can solve the same problem in a more general, flexible way.
143
144For example, for dynamic loading to work on some SVR4 systems, we had
145to build a shared libperl.so library. In order to build "FAT" binaries
146on NeXT 4.0 systems, we had to build a special libperl library. Rather
147than continuing to build a contorted nest of special cases, I
148generalized the process of building libperl so that NeXT and SVR4 users
149could still get their work done, but others could build a shared
150libperl if they wanted to as well.
aa689395 151
a6968aa6
JH
152Contain your changes carefully. Assume nothing about other operating
153systems, not even closely related ones. Your changes must not affect
154other platforms.
155
156Spy shamelessly on how similar patching or porting issues have been
157settled elsewhere.
158
159If feasible, try to keep filenames 8.3-compliant to humor those poor
160souls that get joy from running Perl under such dire limitations.
9e371ce5 161There's a script, check83.pl, for keeping your nose 8.3-clean.
efc41c8e
MB
162In a similar vein, do not create files or directories which differ only
163in case (upper versus lower).
a6968aa6 164
aa689395 165=head2 Seek consensus on major changes
166
167If you are making big changes, don't do it in secret. Discuss the
168ideas in advance on perl5-porters.
169
170=head2 Keep the documentation up-to-date
171
172If your changes may affect how users use perl, then check to be sure
173that the documentation is in sync with your changes. Be sure to
174check all the files F<pod/*.pod> and also the F<INSTALL> document.
175
176Consider writing the appropriate documentation first and then
7b5757d1 177implementing your change to correspond to the documentation.
aa689395 178
179=head2 Avoid machine-specific #ifdef's
180
181To the extent reasonable, try to avoid machine-specific #ifdef's in
182the sources. Instead, use feature-specific #ifdef's. The reason is
183that the machine-specific #ifdef's may not be valid across major
184releases of the operating system. Further, the feature-specific tests
185may help out folks on another platform who have the same problem.
186
a6968aa6
JH
187=head2 Machine-specific files
188
98dddfbd
JH
189=over 4
190
191=item source code
192
a6968aa6
JH
193If you have many machine-specific #defines or #includes, consider
194creating an "osish.h" (os2ish.h, vmsish.h, and so on) and including
195that in perl.h. If you have several machine-specific files (function
196emulations, function stubs, build utility wrappers) you may create a
197separate subdirectory (djgpp, win32) and put the files in there.
98dddfbd 198Remember to update C<MANIFEST> when you add files.
a6968aa6 199
ff935051 200If your system supports dynamic loading but none of the existing
98dddfbd
JH
201methods at F<ext/DynaLoader/dl_*.xs> work for you, you must write
202a new one. Study the existing ones to see what kind of interface
203you must supply.
204
205=item build hints
a6968aa6
JH
206
207There are two kinds of hints: hints for building Perl and hints for
208extensions. The former live in the C<hints> subdirectory, the latter
209in C<ext/*/hints> subdirectories.
210
211The top level hints are Bourne-shell scripts that set, modify and
212unset appropriate Configure variables, based on the Configure command
213line options and possibly existing config.sh and Policy.sh files from
214previous Configure runs.
215
76ba0908 216The extension hints are written in Perl (by the time they are used
a6968aa6
JH
217miniperl has been built) and control the building of their respective
218extensions. They can be used to for example manipulate compilation
219and linking flags.
220
98dddfbd
JH
221=item build and installation Makefiles, scripts, and so forth
222
223Sometimes you will also need to tweak the Perl build and installation
224procedure itself, like for example F<Makefile.SH> and F<installperl>.
225Tread very carefully, even more than usual. Contain your changes
226with utmost care.
a6968aa6 227
98dddfbd
JH
228=item test suite
229
230Many of the tests in C<t> subdirectory assume machine-specific things
a6968aa6
JH
231like existence of certain functions, something about filesystem
232semantics, certain external utilities and their error messages. Use
233the C<$^O> and the C<Config> module (which contains the results of the
234Configure run, in effect the C<config.sh> converted to Perl) to either
98dddfbd
JH
235skip (preferably not) or customize (preferable) the tests for your
236platform.
237
238=item modules
239
240Certain standard modules may need updating if your operating system
241sports for example a native filesystem naming. You may want to update
242some or all of the modules File::Basename, File::Spec, File::Path, and
243File::Copy to become aware of your native filesystem syntax and
244peculiarities.
245
b972f109
JH
246Remember to have a $VERSION in the modules. You can use the
247Porting/checkVERSION.pl script for checking this.
248
98dddfbd
JH
249=item documentation
250
251If your operating system comes from outside UNIX you almost certainly
252will have differences in the available operating system functionality
253(missing system calls, different semantics, whatever). Please
254document these at F<pod/perlport.pod>. If your operating system is
255the first B<not> to have a system call also update the list of
256"portability-bewares" at the beginning of F<pod/perlfunc.pod>.
257
258A file called F<README.youros> at the top level that explains things
259like how to install perl at this platform, where to get any possibly
260required additional software, and for example what test suite errors
76ba0908
PK
261to expect, is nice too. Such files are in the process of being written
262in pod format and will eventually be renamed F<INSTALL.youros>.
98dddfbd
JH
263
264You may also want to write a separate F<.pod> file for your operating
265system to tell about existing mailing lists, os-specific modules,
266documentation, whatever. Please name these along the lines of
267F<perl>I<youros>.pod. [unfinished: where to put this file (the pod/
268subdirectory, of course: but more importantly, which/what index files
269should be updated?)]
270
271=back
a6968aa6 272
aa689395 273=head2 Allow for lots of testing
274
275We should never release a main version without testing it as a
276subversion first.
277
6877a1cf
AD
278=head2 Test popular applications and modules.
279
280We should never release a main version without testing whether or not
281it breaks various popular modules and applications. A partial list of
282such things would include majordomo, metaconfig, apache, Tk, CGI,
283libnet, and libwww, to name just a few. Of course it's quite possible
284that some of those things will be just plain broken and need to be fixed,
285but, in general, we ought to try to avoid breaking widely-installed
286things.
287
98dddfbd 288=head2 Automated generation of derivative files
aa689395 289
290The F<embed.h>, F<keywords.h>, F<opcode.h>, and F<perltoc.pod> files
291are all automatically generated by perl scripts. In general, don't
292patch these directly; patch the data files instead.
293
294F<Configure> and F<config_h.SH> are also automatically generated by
295B<metaconfig>. In general, you should patch the metaconfig units
a6968aa6
JH
296instead of patching these files directly. However, very minor changes
297to F<Configure> may be made in between major sync-ups with the
298metaconfig units, which tends to be complicated operations. But be
299careful, this can quickly spiral out of control. Running metaconfig
300is not really hard.
aa689395 301
98dddfbd
JH
302Also F<Makefile> is automatically produced from F<Makefile.SH>.
303In general, look out for all F<*.SH> files.
304
a8119d38
AD
305Finally, the sample files in the F<Porting/> subdirectory are
306generated automatically by the script F<U/mksample> included
307with the metaconfig units. See L<"run metaconfig"> below for
308information on obtaining the metaconfig units.
309
aa689395 310=head1 How to Make a Distribution
311
312There really ought to be a 'make dist' target, but there isn't.
313The 'dist' suite of tools also contains a number of tools that I haven't
314learned how to use yet. Some of them may make this all a bit easier.
315
316Here are the steps I go through to prepare a patch & distribution.
317
3e3baf6d
TB
318Lots of it could doubtless be automated but isn't. The Porting/makerel
319(make release) perl script does now help automate some parts of it.
aa689395 320
321=head2 Announce your intentions
322
323First, you should volunteer out loud to take the patch pumpkin. It's
324generally counter-productive to have multiple people working in secret
325on the same thing.
326
327At the same time, announce what you plan to do with the patch pumpkin,
328to allow folks a chance to object or suggest alternatives, or do it for
329you. Naturally, the patch pumpkin holder ought to incorporate various
330bug fixes and documentation improvements that are posted while he or
331she has the pumpkin, but there might also be larger issues at stake.
332
333One of the precepts of the subversion idea is that we shouldn't give
7b5757d1
AD
334the patch pumpkin to anyone unless we have some idea what he or she
335is going to do with it.
aa689395 336
337=head2 refresh pod/perltoc.pod
338
339Presumably, you have done a full C<make> in your working source
340directory. Before you C<make spotless> (if you do), and if you have
341changed any documentation in any module or pod file, change to the
342F<pod> directory and run C<make toc>.
343
3e3baf6d
TB
344=head2 run installhtml to check the validity of the pod files
345
aa689395 346=head2 update patchlevel.h
347
348Don't be shy about using the subversion number, even for a relatively
349modest patch. We've never even come close to using all 99 subversions,
350and it's better to have a distinctive number for your patch. If you
351need feedback on your patch, go ahead and issue it and promise to
352incorporate that feedback quickly (e.g. within 1 week) and send out a
353second patch.
354
05ff1fbb
RB
355If you update the subversion number, you may need to change the version
356number near the top of the F<Changes> file.
357
aa689395 358=head2 run metaconfig
359
360If you need to make changes to Configure or config_h.SH, it may be best to
361change the appropriate metaconfig units instead, and regenerate Configure.
362
363 metaconfig -m
364
20f245af
AD
365will regenerate Configure and config_h.SH. Much more information
366on obtaining and running metaconfig is in the F<U/README> file
367that comes with Perl's metaconfig units. Perl's metaconfig units
368should be available on CPAN. A set of units that will work with
369perl5.005 is in the file F<mc_units-5.005_00-01.tar.gz> under
a93751fa 370http://www.cpan.org/authors/id/ANDYD/ . The mc_units tar file
20f245af
AD
371should be unpacked in your main perl source directory. Note: those
372units were for use with 5.005. There may have been changes since then.
d562869c 373Check for later versions or contact perl5-porters@perl.org to obtain a
20f245af 374pointer to the current version.
aa689395 375
376Alternatively, do consider if the F<*ish.h> files might be a better
377place for your changes.
378
379=head2 MANIFEST
380
381Make sure the MANIFEST is up-to-date. You can use dist's B<manicheck>
382program for this. You can also use
383
3e3baf6d 384 perl -w -MExtUtils::Manifest=fullcheck -e fullcheck
aa689395 385
3e3baf6d
TB
386Both commands will also list extra files in the directory that are not
387listed in MANIFEST.
aa689395 388
bfb7748a 389The MANIFEST is normally sorted.
aa689395 390
391If you are using metaconfig to regenerate Configure, then you should note
392that metaconfig actually uses MANIFEST.new, so you want to be sure
393MANIFEST.new is up-to-date too. I haven't found the MANIFEST/MANIFEST.new
394distinction particularly useful, but that's probably because I still haven't
395learned how to use the full suite of tools in the dist distribution.
396
397=head2 Check permissions
398
399All the tests in the t/ directory ought to be executable. The
400main makefile used to do a 'chmod t/*/*.t', but that resulted in
401a self-modifying distribution--something some users would strongly
d562869c
AD
402prefer to avoid. The F<t/TEST> script will check for this
403and do the chmod if needed, but the tests still ought to be
404executable.
aa689395 405
406In all, the following files should probably be executable:
407
408 Configure
409 configpm
32fcaa0b 410 configure.gnu
aa689395 411 embed.pl
412 installperl
413 installman
414 keywords.pl
aa689395 415 myconfig
416 opcode.pl
417 perly.fixer
418 t/TEST
419 t/*/*.t
420 *.SH
421 vms/ext/Stdio/test.pl
422 vms/ext/filespec.t
aa689395 423 x2p/*.SH
424
425Other things ought to be readable, at least :-).
426
427Probably, the permissions for the files could be encoded in MANIFEST
428somehow, but I'm reluctant to change MANIFEST itself because that
429could break old scripts that use MANIFEST.
430
431I seem to recall that some SVR3 systems kept some sort of file that listed
432permissions for system files; something like that might be appropriate.
433
434=head2 Run Configure
435
436This will build a config.sh and config.h. You can skip this if you haven't
693762b4 437changed Configure or config_h.SH at all. I use the following command
aa689395 438
693762b4
AD
439 sh Configure -Dprefix=/opt/perl -Doptimize=-O -Dusethreads \
440 -Dcf_by='yourname' \
441 -Dcf_email='yourname@yourhost.yourplace.com' \
442 -Dperladmin='yourname@yourhost.yourplace.com' \
443 -Dmydomain='.yourplace.com' \
444 -Dmyhostname='yourhost' \
445 -des
aa689395 446
693762b4 447=head2 Update Porting/config.sh and Porting/config_H
dfe9444c 448
693762b4
AD
449[XXX
450This section needs revision. We're currently working on easing
451the task of keeping the vms, win32, and plan9 config.sh info
452up-to-date. The plan is to use keep up-to-date 'canned' config.sh
453files in the appropriate subdirectories and then generate 'canned'
454config.h files for vms, win32, etc. from the generic config.sh file.
455This is to ease maintenance. When Configure gets updated, the parts
456sometimes get scrambled around, and the changes in config_H can
457sometimes be very hard to follow. config.sh, on the other hand, can
458safely be sorted, so it's easy to track (typically very small) changes
459to config.sh and then propoagate them to a canned 'config.h' by any
460number of means, including a perl script in win32/ or carrying
461config.sh and config_h.SH to a Unix system and running sh
76ba0908
PK
462config_h.SH.) Vms uses configure.com to generate its own config.sh
463and config.h. If you want to add a new variable to config.sh check
464with vms folk how to add it to configure.com too.
693762b4
AD
465XXX]
466
467The Porting/config.sh and Porting/config_H files are provided to
468help those folks who can't run Configure. It is important to keep
469them up-to-date. If you have changed config_h.SH, those changes must
470be reflected in config_H as well. (The name config_H was chosen to
471distinguish the file from config.h even on case-insensitive file systems.)
472Simply edit the existing config_H file; keep the first few explanatory
473lines and then copy your new config.h below.
aa689395 474
76ba0908 475It may also be necessary to update win32/config.?c, and
aa689395 476plan9/config.plan9, though you should be quite careful in doing so if
477you are not familiar with those systems. You might want to issue your
478patch with a promise to quickly issue a follow-up that handles those
479directories.
480
481=head2 make run_byacc
482
483If you have byacc-1.8.2 (available from CPAN), and if there have been
484changes to F<perly.y>, you can regenerate the F<perly.c> file. The
485run_byacc makefile target does this by running byacc and then applying
486some patches so that byacc dynamically allocates space, rather than
487having fixed limits. This patch is handled by the F<perly.fixer>
488script. Depending on the nature of the changes to F<perly.y>, you may
489or may not have to hand-edit the patch to apply correctly. If you do,
490you should include the edited patch in the new distribution. If you
491have byacc-1.9, the patch won't apply cleanly. Changes to the printf
492output statements mean the patch won't apply cleanly. Long ago I
493started to fix F<perly.fixer> to detect this, but I never completed the
494task.
495
76ba0908
PK
496If C<perly.c> or C<perly.h> changes, make sure you run C<perl vms/vms_yfix.pl>
497to update the corresponding VMS files. This could be taken care of by
498the regen_all target in the Unix Makefile. See also
499L<VMS-specific updates>.
ebb99254 500
aa689395 501Some additional notes from Larry on this:
502
e262e9be 503Don't forget to regenerate perly_c.diff.
aa689395 504
7b5757d1 505 byacc -d perly.y
aa689395 506 mv y.tab.c perly.c
e262e9be 507 patch perly.c <perly_c.diff
aa689395 508 # manually apply any failed hunks
eade9b71 509 diff -c perly.c.orig perly.c >perly_c.diff
aa689395 510
511One chunk of lines that often fails begins with
512
513 #line 29 "perly.y"
514
515and ends one line before
516
517 #define YYERRCODE 256
518
519This only happens when you add or remove a token type. I suppose this
520could be automated, but it doesn't happen very often nowadays.
521
522Larry
523
76ba0908
PK
524=head2 make regen_all
525
526This target takes care of the PERLYVMS, regen_headers, and regen_pods
527targets.
528
aa689395 529=head2 make regen_headers
530
531The F<embed.h>, F<keywords.h>, and F<opcode.h> files are all automatically
532generated by perl scripts. Since the user isn't guaranteed to have a
533working perl, we can't require the user to generate them. Hence you have
534to, if you're making a distribution.
535
536I used to include rules like the following in the makefile:
537
538 # The following three header files are generated automatically
539 # The correct versions should be already supplied with the perl kit,
540 # in case you don't have perl or 'sh' available.
541 # The - is to ignore error return codes in case you have the source
542 # installed read-only or you don't have perl yet.
543 keywords.h: keywords.pl
544 @echo "Don't worry if this fails."
545 - perl keywords.pl
546
547
7b5757d1 548However, I got B<lots> of mail consisting of people worrying because the
aa689395 549command failed. I eventually decided that I would save myself time
550and effort by manually running C<make regen_headers> myself rather
551than answering all the questions and complaints about the failing
552command.
553
76ba0908
PK
554=head2 make regen_pods
555
556Will run `make regen_pods` in the pod directory for indexing.
557
3e3baf6d 558=head2 global.sym, interp.sym and perlio.sym
aa689395 559
560Make sure these files are up-to-date. Read the comments in these
561files and in perl_exp.SH to see what to do.
562
563=head2 Binary compatibility
564
565If you do change F<global.sym> or F<interp.sym>, think carefully about
566what you are doing. To the extent reasonable, we'd like to maintain
76ba0908 567source and binary compatibility with older releases of perl. That way,
aa689395 568extensions built under one version of perl will continue to work with
569new versions of perl.
570
571Of course, some incompatible changes may well be necessary. I'm just
572suggesting that we not make any such changes without thinking carefully
573about them first. If possible, we should provide
574backwards-compatibility stubs. There's a lot of XS code out there.
575Let's not force people to keep changing it.
576
577=head2 Changes
578
579Be sure to update the F<Changes> file. Try to include both an overall
580summary as well as detailed descriptions of the changes. Your
3e3baf6d 581audience will include other developers and users, so describe
aa689395 582user-visible changes (if any) in terms they will understand, not in
583code like "initialize foo variable in bar function".
584
585There are differing opinions on whether the detailed descriptions
586ought to go in the Changes file or whether they ought to be available
587separately in the patch file (or both). There is no disagreement that
588detailed descriptions ought to be easily available somewhere.
589
05ff1fbb
RB
590If you update the subversion number in F<patchlevel.h>, you may need
591to change the version number near the top of the F<Changes> file.
592
2a26e2f1
DD
593=head2 Todo
594
efc41c8e
MB
595The F<pod/perltodo.pod> file contains a roughly-categorized unordered
596list of aspects of Perl that could use enhancement, features that could
597be added, areas that could be cleaned up, and so on. During your term
598as pumpkin-holder, you will probably address some of these issues, and
599perhaps identify others which, while you decide not to address them this
600time around, may be tackled in the future. Update the file to reflect
601the situation as it stands when you hand over the pumpkin.
2a26e2f1
DD
602
603You might like, early in your pumpkin-holding career, to see if you
604can find champions for partiticular issues on the to-do list: an issue
605owned is an issue more likely to be resolved.
606
94655993 607There are also some more porting-specific L</Todo> items later in this
c4f23d77
AD
608file.
609
aa689395 610=head2 OS/2-specific updates
611
612In the os2 directory is F<diff.configure>, a set of OS/2-specific
613diffs against B<Configure>. If you make changes to Configure, you may
614want to consider regenerating this diff file to save trouble for the
615OS/2 maintainer.
616
7b5757d1
AD
617You can also consider the OS/2 diffs as reminders of portability
618things that need to be fixed in Configure.
619
aa689395 620=head2 VMS-specific updates
621
ebb99254 622If you have changed F<perly.y> or F<perly.c>, then you most probably want
76ba0908
PK
623to update F<vms/perly_{h,c}.vms> by running C<perl vms/vms_yfix.pl>, or
624by running `make regen_all` which will run that script for you.
aa689395 625
76ba0908
PK
626The Perl revision number appears as "perl5" in configure.com.
627It is courteous to update that if necessary.
aa689395 628
629=head2 Making the new distribution
630
631Suppose, for example, that you want to make version 5.004_08. Then you can
632do something like the following
633
634 mkdir ../perl5.004_08
635 awk '{print $1}' MANIFEST | cpio -pdm ../perl5.004_08
636 cd ../
637 tar cf perl5.004_08.tar perl5.004_08
638 gzip --best perl5.004_08.tar
639
3e3baf6d
TB
640These steps, with extra checks, are automated by the Porting/makerel
641script.
642
aa689395 643=head2 Making a new patch
644
645I find the F<makepatch> utility quite handy for making patches.
646You can obtain it from any CPAN archive under
a93751fa 647http://www.cpan.org/authors/Johan_Vromans/ . There are a couple
3e3baf6d
TB
648of differences between my version and the standard one. I have mine do
649a
aa689395 650
651 # Print a reassuring "End of Patch" note so people won't
652 # wonder if their mailer truncated patches.
653 print "\n\nEnd of Patch.\n";
654
3e3baf6d
TB
655at the end. That's because I used to get questions from people asking
656if their mail was truncated.
657
658It also writes Index: lines which include the new directory prefix
659(change Index: print, approx line 294 or 310 depending on the version,
660to read: print PATCH ("Index: $newdir$new\n");). That helps patches
661work with more POSIX conformant patch programs.
aa689395 662
663Here's how I generate a new patch. I'll use the hypothetical
6645.004_07 to 5.004_08 patch as an example.
665
666 # unpack perl5.004_07/
667 gzip -d -c perl5.004_07.tar.gz | tar -xof -
668 # unpack perl5.004_08/
669 gzip -d -c perl5.004_08.tar.gz | tar -xof -
670 makepatch perl5.004_07 perl5.004_08 > perl5.004_08.pat
671
672Makepatch will automatically generate appropriate B<rm> commands to remove
673deleted files. Unfortunately, it will not correctly set permissions
674for newly created files, so you may have to do so manually. For example,
675patch 5.003_04 created a new test F<t/op/gv.t> which needs to be executable,
676so at the top of the patch, I inserted the following lines:
677
678 # Make a new test
679 touch t/op/gv.t
680 chmod +x t/opt/gv.t
681
682Now, of course, my patch is now wrong because makepatch didn't know I
683was going to do that command, and it patched against /dev/null.
684
685So, what I do is sort out all such shell commands that need to be in the
686patch (including possible mv-ing of files, if needed) and put that in the
687shell commands at the top of the patch. Next, I delete all the patch parts
688of perl5.004_08.pat, leaving just the shell commands. Then, I do the
689following:
690
7b5757d1
AD
691 cd perl5.004_07
692 sh ../perl5.004_08.pat
aa689395 693 cd ..
7b5757d1 694 makepatch perl5.004_07 perl5.004_08 >> perl5.004_08.pat
aa689395 695
696(Note the append to preserve my shell commands.)
697Now, my patch will line up with what the end users are going to do.
698
699=head2 Testing your patch
700
701It seems obvious, but be sure to test your patch. That is, verify that
702it produces exactly the same thing as your full distribution.
703
7b5757d1
AD
704 rm -rf perl5.004_07
705 gzip -d -c perl5.004_07.tar.gz | tar -xf -
706 cd perl5.004_07
707 sh ../perl5.004_08.pat
708 patch -p1 -N < ../perl5.004_08.pat
aa689395 709 cd ..
7b5757d1 710 gdiff -r perl5.004_07 perl5.004_08
aa689395 711
712where B<gdiff> is GNU diff. Other diff's may also do recursive checking.
713
714=head2 More testing
715
716Again, it's obvious, but you should test your new version as widely as you
717can. You can be sure you'll hear about it quickly if your version doesn't
718work on both ANSI and pre-ANSI compilers, and on common systems such as
719SunOS 4.1.[34], Solaris, and Linux.
720
721If your changes include conditional code, try to test the different
722branches as thoroughly as you can. For example, if your system
723supports dynamic loading, you can also test static loading with
724
725 sh Configure -Uusedl
726
727You can also hand-tweak your config.h to try out different #ifdef
728branches.
729
d2560b70
RB
730=head2 Other tests
731
732=over 4
733
734=item CHECK_FORMAT
735
736To test the correct use of printf-style arguments, C<Configure> with
737S<-Dccflags='-DCHECK_FORMAT -Wformat'> and run C<make>. The compiler
738will produce warning of incorrect use of format arguments. CHECK_FORMAT
739changes perl-defined formats to common formats, so DO NOT USE the executable
740produced by this process.
741
742A more accurate approach is the following commands:
743
b3fe4827
RB
744=over 4
745
746=item *
747
748build miniperl with -DCHECK_FORMAT
749
750 make clean
751 make miniperl OPTIMIZE=-DCHECK_FORMAT >& mini.log
752
753=item *
754
755build a clean miniperl,
756and build everything else from that with -DCHECK_FORMAT
757
d2560b70 758 make clean
b3fe4827 759 make miniperl
436c6dd3 760 make all OPTIMIZE='-DCHECK_FORMAT -Wformat' >& make.log
b3fe4827
RB
761
762=item *
763
764clean up, and print warnings from the log files
765
d2560b70 766 make clean
b3fe4827
RB
767 perl -nwe 'print if /^\S+:/ and not /^make\b/' \
768 mini.log make.log
769
770=back
d2560b70
RB
771
772(-Wformat support by Robin Barker.)
773
93189314
JH
774=item gcc -ansi -pedantic
775
776Configure -Dgccansipedantic [ -Dcc=gcc ] will enable (via the cflags script,
777not $Config{ccflags}) the gcc strict ANSI C flags -ansi and -pedantic for
778the compilation of the core files on platforms where it knows it can
779do so (like Linux, see cflags.SH for the full list), and on some
780platforms only one (Solaris can do only -pedantic, not -ansi).
781The flag -DPERL_GCC_PEDANTIC also gets added, since gcc does not add
782any internal cpp flag to signify that -pedantic is being used, as it
783does for -ansi (__STRICT_ANSI__).
784
a0426075
MB
785Note that the -ansi and -pedantic are enabled only for version 3 (and
786later) of gcc, since even gcc version 2.95.4 finds lots of seemingly
787false "value computed not used" errors from Perl.
788
93189314
JH
789The -ansi and -pedantic are useful in catching at least the following
790nonportable practices:
791
792=over 4
793
794=item *
795
796gcc-specific extensions
797
798=item *
799
800lvalue casts
801
802=item *
803
804// C++ comments
805
806=item *
807
808enum trailing commas
809
810=back
811
812The -Dgccansipedantic should be used only when cleaning up the code,
813not for production builds, since otherwise gcc cannot inline certain
814things.
815
d2560b70
RB
816=back
817
d33b2eba 818=head1 Running Purify
f5a32c7f
GS
819
820Purify is a commercial tool that is helpful in identifying memory
821overruns, wild pointers, memory leaks and other such badness. Perl
822must be compiled in a specific way for optimal testing with Purify.
823
824Use the following commands to test perl with Purify:
825
826 sh Configure -des -Doptimize=-g -Uusemymalloc -Dusemultiplicity \
827 -Accflags=-DPURIFY
828 setenv PURIFYOPTIONS "-chain-length=25"
829 make all pureperl
830 cd t
831 ln -s ../pureperl perl
365a6279 832 setenv PERL_DESTRUCT_LEVEL 2
f5a32c7f
GS
833 ./perl TEST
834
835Disabling Perl's malloc allows Purify to monitor allocations and leaks
836more closely; using Perl's malloc will make Purify report most leaks
837in the "potential" leaks category. Enabling the multiplicity option
838allows perl to clean up thoroughly when the interpreter shuts down, which
839reduces the number of bogus leak reports from Purify. The -DPURIFY
840enables any Purify-specific debugging code in the sources.
841
842Purify outputs messages in "Viewer" windows by default. If you don't have
843a windowing environment or if you simply want the Purify output to
844unobtrusively go to a log file instead of to the interactive window,
845use the following options instead:
846
847 setenv PURIFYOPTIONS "-chain-length=25 -windows=no -log-file=perl.log \
848 -append-logfile=yes"
849
850The only currently known leaks happen when there are compile-time errors
851within eval or require. (Fixing these is non-trivial, unfortunately, but
852they must be fixed eventually.)
853
aa689395 854=head1 Common Gotcha's
855
856=over 4
857
858=item #elif
859
860The '#elif' preprocessor directive is not understood on all systems.
861Specifically, I know that Pyramids don't understand it. Thus instead of the
862simple
863
864 #if defined(I_FOO)
865 # include <foo.h>
866 #elif defined(I_BAR)
867 # include <bar.h>
868 #else
869 # include <fubar.h>
870 #endif
871
872You have to do the more Byzantine
873
874 #if defined(I_FOO)
875 # include <foo.h>
876 #else
877 # if defined(I_BAR)
878 # include <bar.h>
879 # else
880 # include <fubar.h>
881 # endif
882 #endif
883
884Incidentally, whitespace between the leading '#' and the preprocessor
885command is not guaranteed, but is very portable and you may use it freely.
886I think it makes things a bit more readable, especially once things get
887rather deeply nested. I also think that things should almost never get
888too deeply nested, so it ought to be a moot point :-)
889
890=item Probably Prefer POSIX
891
892It's often the case that you'll need to choose whether to do
893something the BSD-ish way or the POSIX-ish way. It's usually not
894a big problem when the two systems use different names for similar
895functions, such as memcmp() and bcmp(). The perl.h header file
896handles these by appropriate #defines, selecting the POSIX mem*()
897functions if available, but falling back on the b*() functions, if
898need be.
899
900More serious is the case where some brilliant person decided to
901use the same function name but give it a different meaning or
902calling sequence :-). getpgrp() and setpgrp() come to mind.
903These are a real problem on systems that aim for conformance to
904one standard (e.g. POSIX), but still try to support the other way
905of doing things (e.g. BSD). My general advice (still not really
906implemented in the source) is to do something like the following.
907Suppose there are two alternative versions, fooPOSIX() and
908fooBSD().
909
910 #ifdef HAS_FOOPOSIX
911 /* use fooPOSIX(); */
912 #else
913 # ifdef HAS_FOOBSD
914 /* try to emulate fooPOSIX() with fooBSD();
915 perhaps with the following: */
916 # define fooPOSIX fooBSD
917 # else
918 # /* Uh, oh. We have to supply our own. */
919 # define fooPOSIX Perl_fooPOSIX
920 # endif
921 #endif
922
923=item Think positively
924
925If you need to add an #ifdef test, it is usually easier to follow if you
926think positively, e.g.
927
928 #ifdef HAS_NEATO_FEATURE
929 /* use neato feature */
930 #else
931 /* use some fallback mechanism */
932 #endif
933
934rather than the more impenetrable
935
936 #ifndef MISSING_NEATO_FEATURE
937 /* Not missing it, so we must have it, so use it */
938 #else
939 /* Are missing it, so fall back on something else. */
940 #endif
941
942Of course for this toy example, there's not much difference. But when
943the #ifdef's start spanning a couple of screen fulls, and the #else's
944are marked something like
945
946 #else /* !MISSING_NEATO_FEATURE */
947
948I find it easy to get lost.
949
950=item Providing Missing Functions -- Problem
951
952Not all systems have all the neat functions you might want or need, so
953you might decide to be helpful and provide an emulation. This is
954sound in theory and very kind of you, but please be careful about what
955you name the function. Let me use the C<pause()> function as an
956illustration.
957
958Perl5.003 has the following in F<perl.h>
959
960 #ifndef HAS_PAUSE
961 #define pause() sleep((32767<<16)+32767)
962 #endif
963
964Configure sets HAS_PAUSE if the system has the pause() function, so
965this #define only kicks in if the pause() function is missing.
966Nice idea, right?
967
968Unfortunately, some systems apparently have a prototype for pause()
969in F<unistd.h>, but don't actually have the function in the library.
970(Or maybe they do have it in a library we're not using.)
971
972Thus, the compiler sees something like
973
974 extern int pause(void);
975 /* . . . */
976 #define pause() sleep((32767<<16)+32767)
977
978and dies with an error message. (Some compilers don't mind this;
979others apparently do.)
980
981To work around this, 5.003_03 and later have the following in perl.h:
982
983 /* Some unistd.h's give a prototype for pause() even though
984 HAS_PAUSE ends up undefined. This causes the #define
985 below to be rejected by the compiler. Sigh.
986 */
987 #ifdef HAS_PAUSE
988 # define Pause pause
989 #else
990 # define Pause() sleep((32767<<16)+32767)
991 #endif
992
993This works.
994
995The curious reader may wonder why I didn't do the following in
996F<util.c> instead:
997
998 #ifndef HAS_PAUSE
999 void pause()
1000 {
1001 sleep((32767<<16)+32767);
1002 }
1003 #endif
1004
1005That is, since the function is missing, just provide it.
1006Then things would probably be been alright, it would seem.
1007
1008Well, almost. It could be made to work. The problem arises from the
1009conflicting needs of dynamic loading and namespace protection.
1010
1011For dynamic loading to work on AIX (and VMS) we need to provide a list
1012of symbols to be exported. This is done by the script F<perl_exp.SH>,
1013which reads F<global.sym> and F<interp.sym>. Thus, the C<pause>
1014symbol would have to be added to F<global.sym> So far, so good.
1015
1016On the other hand, one of the goals of Perl5 is to make it easy to
1017either extend or embed perl and link it with other libraries. This
1018means we have to be careful to keep the visible namespace "clean".
1019That is, we don't want perl's global variables to conflict with
1020those in the other application library. Although this work is still
1021in progress, the way it is currently done is via the F<embed.h> file.
1022This file is built from the F<global.sym> and F<interp.sym> files,
1023since those files already list the globally visible symbols. If we
1024had added C<pause> to global.sym, then F<embed.h> would contain the
1025line
1026
1027 #define pause Perl_pause
1028
1029and calls to C<pause> in the perl sources would now point to
1030C<Perl_pause>. Now, when B<ld> is run to build the F<perl> executable,
1031it will go looking for C<perl_pause>, which probably won't exist in any
1032of the standard libraries. Thus the build of perl will fail.
1033
1034Those systems where C<HAS_PAUSE> is not defined would be ok, however,
1035since they would get a C<Perl_pause> function in util.c. The rest of
1036the world would be in trouble.
1037
1038And yes, this scenario has happened. On SCO, the function C<chsize>
1039is available. (I think it's in F<-lx>, the Xenix compatibility
1040library.) Since the perl4 days (and possibly before), Perl has
1041included a C<chsize> function that gets called something akin to
1042
1043 #ifndef HAS_CHSIZE
1044 I32 chsize(fd, length)
1045 /* . . . */
1046 #endif
1047
1048When 5.003 added
1049
1050 #define chsize Perl_chsize
1051
1052to F<embed.h>, the compile started failing on SCO systems.
1053
1054The "fix" is to give the function a different name. The one
1055implemented in 5.003_05 isn't optimal, but here's what was done:
1056
1057 #ifdef HAS_CHSIZE
1058 # ifdef my_chsize /* Probably #defined to Perl_my_chsize in embed.h */
1059 # undef my_chsize
1060 # endif
1061 # define my_chsize chsize
1062 #endif
1063
1064My explanatory comment in patch 5.003_05 said:
1065
1066 Undef and then re-define my_chsize from Perl_my_chsize to
1067 just plain chsize if this system HAS_CHSIZE. This probably only
1068 applies to SCO. This shows the perils of having internal
1069 functions with the same name as external library functions :-).
1070
1071Now, we can safely put C<my_chsize> in F<global.sym>, export it, and
1072hide it with F<embed.h>.
1073
1074To be consistent with what I did for C<pause>, I probably should have
1075called the new function C<Chsize>, rather than C<my_chsize>.
1076However, the perl sources are quite inconsistent on this (Consider
1077New, Mymalloc, and Myremalloc, to name just a few.)
1078
1079There is a problem with this fix, however, in that C<Perl_chsize>
1080was available as a F<libperl.a> library function in 5.003, but it
1081isn't available any more (as of 5.003_07). This means that we've
1082broken binary compatibility. This is not good.
1083
1084=item Providing missing functions -- some ideas
1085
1086We currently don't have a standard way of handling such missing
1087function names. Right now, I'm effectively thinking aloud about a
1088solution. Some day, I'll try to formally propose a solution.
1089
1090Part of the problem is that we want to have some functions listed as
1091exported but not have their names mangled by embed.h or possibly
1092conflict with names in standard system headers. We actually already
1093have such a list at the end of F<perl_exp.SH> (though that list is
1094out-of-date):
1095
1096 # extra globals not included above.
1097 cat <<END >> perl.exp
1098 perl_init_ext
1099 perl_init_fold
1100 perl_init_i18nl14n
1101 perl_alloc
1102 perl_construct
1103 perl_destruct
1104 perl_free
1105 perl_parse
1106 perl_run
1107 perl_get_sv
1108 perl_get_av
1109 perl_get_hv
1110 perl_get_cv
1111 perl_call_argv
1112 perl_call_pv
1113 perl_call_method
1114 perl_call_sv
1115 perl_requirepv
1116 safecalloc
1117 safemalloc
1118 saferealloc
1119 safefree
1120
1121This still needs much thought, but I'm inclined to think that one
1122possible solution is to prefix all such functions with C<perl_> in the
1123source and list them along with the other C<perl_*> functions in
1124F<perl_exp.SH>.
1125
1126Thus, for C<chsize>, we'd do something like the following:
1127
1128 /* in perl.h */
1129 #ifdef HAS_CHSIZE
1130 # define perl_chsize chsize
1131 #endif
1132
1133then in some file (e.g. F<util.c> or F<doio.c>) do
1134
1135 #ifndef HAS_CHSIZE
1136 I32 perl_chsize(fd, length)
1137 /* implement the function here . . . */
1138 #endif
1139
1140Alternatively, we could just always use C<chsize> everywhere and move
1141C<chsize> from F<global.sym> to the end of F<perl_exp.SH>. That would
1142probably be fine as long as our C<chsize> function agreed with all the
1143C<chsize> function prototypes in the various systems we'll be using.
1144As long as the prototypes in actual use don't vary that much, this is
1145probably a good alternative. (As a counter-example, note how Configure
1146and perl have to go through hoops to find and use get Malloc_t and
1147Free_t for C<malloc> and C<free>.)
1148
1149At the moment, this latter option is what I tend to prefer.
1150
1151=item All the world's a VAX
1152
1153Sorry, showing my age:-). Still, all the world is not BSD 4.[34],
1154SVR4, or POSIX. Be aware that SVR3-derived systems are still quite
1155common (do you have any idea how many systems run SCO?) If you don't
1156have a bunch of v7 manuals handy, the metaconfig units (by default
1157installed in F</usr/local/lib/dist/U>) are a good resource to look at
1158for portability.
1159
1160=back
1161
1162=head1 Miscellaneous Topics
1163
1164=head2 Autoconf
1165
1166Why does perl use a metaconfig-generated Configure script instead of an
1167autoconf-generated configure script?
1168
1169Metaconfig and autoconf are two tools with very similar purposes.
1170Metaconfig is actually the older of the two, and was originally written
1171by Larry Wall, while autoconf is probably now used in a wider variety of
1172packages. The autoconf info file discusses the history of autoconf and
1173how it came to be. The curious reader is referred there for further
1174information.
1175
1176Overall, both tools are quite good, I think, and the choice of which one
1177to use could be argued either way. In March, 1994, when I was just
1178starting to work on Configure support for Perl5, I considered both
1179autoconf and metaconfig, and eventually decided to use metaconfig for the
1180following reasons:
1181
1182=over 4
1183
1184=item Compatibility with Perl4
1185
1186Perl4 used metaconfig, so many of the #ifdef's were already set up for
1187metaconfig. Of course metaconfig had evolved some since Perl4's days,
1188but not so much that it posed any serious problems.
1189
1190=item Metaconfig worked for me
1191
d1be9408 1192My system at the time was Interactive 2.2, an SVR3.2/386 derivative that
aa689395 1193also had some POSIX support. Metaconfig-generated Configure scripts
1194worked fine for me on that system. On the other hand, autoconf-generated
1195scripts usually didn't. (They did come quite close, though, in some
1196cases.) At the time, I actually fetched a large number of GNU packages
1197and checked. Not a single one configured and compiled correctly
1198out-of-the-box with the system's cc compiler.
1199
1200=item Configure can be interactive
1201
1202With both autoconf and metaconfig, if the script works, everything is
1203fine. However, one of my main problems with autoconf-generated scripts
1204was that if it guessed wrong about something, it could be B<very> hard to
1205go back and fix it. For example, autoconf always insisted on passing the
1206-Xp flag to cc (to turn on POSIX behavior), even when that wasn't what I
1207wanted or needed for that package. There was no way short of editing the
1208configure script to turn this off. You couldn't just edit the resulting
1209Makefile at the end because the -Xp flag influenced a number of other
1210configure tests.
1211
1212Metaconfig's Configure scripts, on the other hand, can be interactive.
1213Thus if Configure is guessing things incorrectly, you can go back and fix
1214them. This isn't as important now as it was when we were actively
1215developing Configure support for new features such as dynamic loading,
1216but it's still useful occasionally.
1217
1218=item GPL
1219
1220At the time, autoconf-generated scripts were covered under the GNU Public
1221License, and hence weren't suitable for inclusion with Perl, which has a
1222different licensing policy. (Autoconf's licensing has since changed.)
1223
1224=item Modularity
1225
1226Metaconfig builds up Configure from a collection of discrete pieces
1227called "units". You can override the standard behavior by supplying your
1228own unit. With autoconf, you have to patch the standard files instead.
1229I find the metaconfig "unit" method easier to work with. Others
1230may find metaconfig's units clumsy to work with.
1231
1232=back
1233
aa689395 1234=head2 Why isn't there a directory to override Perl's library?
1235
1236Mainly because no one's gotten around to making one. Note that
1237"making one" involves changing perl.c, Configure, config_h.SH (and
1238associated files, see above), and I<documenting> it all in the
1239INSTALL file.
1240
1241Apparently, most folks who want to override one of the standard library
1242files simply do it by overwriting the standard library files.
1243
1244=head2 APPLLIB
1245
1246In the perl.c sources, you'll find an undocumented APPLLIB_EXP
1247variable, sort of like PRIVLIB_EXP and ARCHLIB_EXP (which are
1248documented in config_h.SH). Here's what APPLLIB_EXP is for, from
1249a mail message from Larry:
1250
1251 The main intent of APPLLIB_EXP is for folks who want to send out a
1252 version of Perl embedded in their product. They would set the symbol
1253 to be the name of the library containing the files needed to run or to
1254 support their particular application. This works at the "override"
1255 level to make sure they get their own versions of any library code that
1256 they absolutely must have configuration control over.
1257
1258 As such, I don't see any conflict with a sysadmin using it for a
1259 override-ish sort of thing, when installing a generic Perl. It should
1260 probably have been named something to do with overriding though. Since
1261 it's undocumented we could still change it... :-)
1262
24f415b4
AD
1263Given that it's already there, you can use it to override distribution modules.
1264One way to do that is to add
1265
453a1e5f 1266 ccflags="$ccflags -DAPPLLIB_EXP=\"/my/override\""
24f415b4
AD
1267
1268to your config.over file. (You have to be particularly careful to get the
453a1e5f
MB
1269double quotes in. APPLLIB_EXP must be a valid C string. It might
1270actually be easier to just #define it yourself in perl.c.)
24f415b4
AD
1271
1272Then perl.c will put /my/override ahead of ARCHLIB and PRIVLIB. Perl will
1273also search architecture-specific and version-specific subdirectories of
1274APPLLIB_EXP.
aa689395 1275
c4f23d77
AD
1276=head2 Shared libperl.so location
1277
1278Why isn't the shared libperl.so installed in /usr/lib/ along
1279with "all the other" shared libraries? Instead, it is installed
1280in $archlib, which is typically something like
1281
1282 /usr/local/lib/perl5/archname/5.00404
1283
1284and is architecture- and version-specific.
1285
1286The basic reason why a shared libperl.so gets put in $archlib is so that
1287you can have more than one version of perl on the system at the same time,
1288and have each refer to its own libperl.so.
1289
1290Three examples might help. All of these work now; none would work if you
1291put libperl.so in /usr/lib.
1292
1293=over
1294
1295=item 1.
1296
1297Suppose you want to have both threaded and non-threaded perl versions
1298around. Configure will name both perl libraries "libperl.so" (so that
1299you can link to them with -lperl). The perl binaries tell them apart
1300by having looking in the appropriate $archlib directories.
1301
1302=item 2.
1303
1304Suppose you have perl5.004_04 installed and you want to try to compile
1305it again, perhaps with different options or after applying a patch.
1306If you already have libperl.so installed in /usr/lib/, then it may be
1307either difficult or impossible to get ld.so to find the new libperl.so
1308that you're trying to build. If, instead, libperl.so is tucked away in
1309$archlib, then you can always just change $archlib in the current perl
1310you're trying to build so that ld.so won't find your old libperl.so.
1311(The INSTALL file suggests you do this when building a debugging perl.)
1312
1313=item 3.
1314
1315The shared perl library is not a "well-behaved" shared library with
1316proper major and minor version numbers, so you can't necessarily
1317have perl5.004_04 and perl5.004_05 installed simultaneously. Suppose
1318perl5.004_04 were to install /usr/lib/libperl.so.4.4, and perl5.004_05
1319were to install /usr/lib/libperl.so.4.5. Now, when you try to run
1320perl5.004_04, ld.so might try to load libperl.so.4.5, since it has
1321the right "major version" number. If this works at all, it almost
1322certainly defeats the reason for keeping perl5.004_04 around. Worse,
1323with development subversions, you certaily can't guarantee that
1324libperl.so.4.4 and libperl.so.4.55 will be compatible.
1325
1326Anyway, all this leads to quite obscure failures that are sure to drive
1327casual users crazy. Even experienced users will get confused :-). Upon
1328reflection, I'd say leave libperl.so in $archlib.
1329
94655993
SR
1330=back
1331
1332=head2 Indentation style
2032ff04 1333
94655993 1334Over the years Perl has become a mishmash of
2032ff04
JH
1335various indentation styles, but the original "Larry style" can
1336probably be restored with (GNU) indent somewhat like this:
1337
1338 indent -kr -nce -psl -sc
1339
55c0ed8c
JH
1340A more ambitious solution would also specify a list of Perl specific
1341types with -TSV -TAV -THV .. -TMAGIC -TPerlIO ... but that list would
1342be quite ungainly. Also note that GNU indent also doesn't do aligning
1343of consecutive assignments, which would truly wreck the layout in
1344places like sv.c:Perl_sv_upgrade() or sv.c:Perl_clone_using().
1345Similarly nicely aligned &&s, ||s and ==s would not be respected.
2032ff04 1346
aa689395 1347=head1 Upload Your Work to CPAN
1348
1349You can upload your work to CPAN if you have a CPAN id. Check out
a93751fa 1350http://www.cpan.org/modules/04pause.html for information on
aa689395 1351_PAUSE_, the Perl Author's Upload Server.
1352
1353I typically upload both the patch file, e.g. F<perl5.004_08.pat.gz>
1354and the full tar file, e.g. F<perl5.004_08.tar.gz>.
1355
1356If you want your patch to appear in the F<src/5.0/unsupported>
1357directory on CPAN, send e-mail to the CPAN master librarian. (Check
a93751fa 1358out http://www.cpan.org/CPAN.html ).
aa689395 1359
1360=head1 Help Save the World
1361
1362You should definitely announce your patch on the perl5-porters list.
1363You should also consider announcing your patch on
1364comp.lang.perl.announce, though you should make it quite clear that a
1365subversion is not a production release, and be prepared to deal with
1366people who will not read your disclaimer.
1367
1368=head1 Todo
1369
1370Here, in no particular order, are some Configure and build-related
1371items that merit consideration. This list isn't exhaustive, it's just
1372what I came up with off the top of my head.
1373
e25f343d
PG
1374=head2 Adding missing library functions to Perl
1375
1376The perl Configure script automatically determines which headers and
1377functions you have available on your system and arranges for them to be
1378included in the compilation and linking process. Occasionally, when porting
1379perl to an operating system for the first time, you may find that the
1380operating system is missing a key function. While perl may still build
1381without this function, no perl program will be able to reference the missing
1382function. You may be able to write the missing function yourself, or you
1383may be able to find the missing function in the distribution files for
1384another software package. In this case, you need to instruct the perl
1385configure-and-build process to use your function. Perform these steps.
1386
1387=over 3
1388
1389=item *
1390
2ecb232b 1391Code and test the function you wish to add. Test it carefully; you will
e25f343d
PG
1392have a much easier time debugging your code independently than when it is a
1393part of perl.
1394
1395=item *
1396
1397Here is an implementation of the POSIX truncate function for an operating
1398system (VOS) that does not supply one, but which does supply the ftruncate()
1399function.
1400
1401 /* Beginning of modification history */
1402 /* Written 02-01-02 by Nick Ing-Simmons (nick@ing-simmons.net) */
1403 /* End of modification history */
1404
1405 /* VOS doesn't supply a truncate function, so we build one up
1406 from the available POSIX functions. */
1407
1408 #include <fcntl.h>
1409 #include <sys/types.h>
1410 #include <unistd.h>
1411
1412 int
1413 truncate(const char *path, off_t len)
1414 {
1415 int fd = open(path,O_WRONLY);
1416 int code = -1;
1417 if (fd >= 0) {
1418 code = ftruncate(fd,len);
1419 close(fd);
1420 }
1421 return code;
1422 }
1423
1424Place this file into a subdirectory that has the same name as the operating
1425system. This file is named perl/vos/vos.c
1426
1427=item *
1428
1429If your operating system has a hints file (in perl/hints/XXX.sh for an
1430operating system named XXX), then start with it. If your operating system
1431has no hints file, then create one. You can use a hints file for a similar
1432operating system, if one exists, as a template.
1433
1434=item *
1435
1436Add lines like the following to your hints file. The first line
1437(d_truncate="define") instructs Configure that the truncate() function
1438exists. The second line (archobjs="vos.o") instructs the makefiles that the
1439perl executable depends on the existence of a file named "vos.o". (Make
1440will automatically look for "vos.c" and compile it with the same options as
1441the perl source code). The final line ("test -h...") adds a symbolic link
1442to the top-level directory so that make can find vos.c. Of course, you
1443should use your own operating system name for the source file of extensions,
1444not "vos.c".
1445
1446 # VOS does not have truncate() but we supply one in vos.c
1447 d_truncate="define"
1448 archobjs="vos.o"
1449
1450 # Help gmake find vos.c
1451 test -h vos.c || ln -s vos/vos.c vos.c
1452
1453The hints file is a series of shell commands that are run in the top-level
1454directory (the "perl" directory). Thus, these commands are simply executed
1455by Configure at an appropriate place during its execution.
1456
1457=item *
1458
1459At this point, you can run the Configure script and rebuild perl. Carefully
1460test the newly-built perl to ensure that normal paths, and error paths,
1461behave as you expect.
1462
1463=back
1464
aa689395 1465=head2 Good ideas waiting for round tuits
1466
1467=over 4
1468
c4f23d77 1469=item Configure -Dsrc=/blah/blah
aa689395 1470
1471We should be able to emulate B<configure --srcdir>. Tom Tromey
1472tromey@creche.cygnus.com has submitted some patches to
c4f23d77
AD
1473the dist-users mailing list along these lines. They have been folded
1474back into the main distribution, but various parts of the perl
1475Configure/build/install process still assume src='.'.
aa689395 1476
1477=item Hint file fixes
1478
1479Various hint files work around Configure problems. We ought to fix
1480Configure so that most of them aren't needed.
1481
1482=item Hint file information
1483
1484Some of the hint file information (particularly dynamic loading stuff)
1485ought to be fed back into the main metaconfig distribution.
1486
1487=back
1488
1489=head2 Probably good ideas waiting for round tuits
1490
1491=over 4
1492
1493=item GNU configure --options
1494
1495I've received sensible suggestions for --exec_prefix and other
1496GNU configure --options. It's not always obvious exactly what is
1497intended, but this merits investigation.
1498
1499=item make clean
1500
1501Currently, B<make clean> isn't all that useful, though
1502B<make realclean> and B<make distclean> are. This needs a bit of
1503thought and documentation before it gets cleaned up.
1504
1505=item Try gcc if cc fails
1506
1507Currently, we just give up.
1508
1509=item bypassing safe*alloc wrappers
1510
1511On some systems, it may be safe to call the system malloc directly
1512without going through the util.c safe* layers. (Such systems would
1513accept free(0), for example.) This might be a time-saver for systems
1514that already have a good malloc. (Recent Linux libc's apparently have
1515a nice malloc that is well-tuned for the system.)
1516
1517=back
1518
1519=head2 Vague possibilities
1520
1521=over 4
1522
aa689395 1523=item MacPerl
1524
3e3baf6d 1525Get some of the Macintosh stuff folded back into the main distribution.
aa689395 1526
1527=item gconvert replacement
1528
1529Maybe include a replacement function that doesn't lose data in rare
1530cases of coercion between string and numerical values.
1531
aa689395 1532=item Improve makedepend
1533
1534The current makedepend process is clunky and annoyingly slow, but it
1535works for most folks. Alas, it assumes that there is a filename
1536$firstmakefile that the B<make> command will try to use before it uses
1537F<Makefile>. Such may not be the case for all B<make> commands,
1538particularly those on non-Unix systems.
1539
1540Probably some variant of the BSD F<.depend> file will be useful.
1541We ought to check how other packages do this, if they do it at all.
1542We could probably pre-generate the dependencies (with the exception of
1543malloc.o, which could probably be determined at F<Makefile.SH>
1544extraction time.
1545
1546=item GNU Makefile standard targets
1547
1548GNU software generally has standardized Makefile targets. Unless we
1549have good reason to do otherwise, I see no reason not to support them.
1550
1551=item File locking
1552
1553Somehow, straighten out, document, and implement lockf(), flock(),
76ba0908
PK
1554and/or fcntl() file locking. It's a mess. See $d_fcntl_can_lock
1555in recent config.sh files though.
aa689395 1556
1557=back
1558
fb73857a 1559=head1 AUTHORS
aa689395 1560
fb73857a 1561Original author: Andy Dougherty doughera@lafcol.lafayette.edu .
1562Additions by Chip Salzenberg chip@perl.com and
1563Tim Bunce Tim.Bunce@ig.co.uk .
aa689395 1564
1565All opinions expressed herein are those of the authorZ<>(s).
1566
1567=head1 LAST MODIFIED
1568
ff935051 1569$Id: pumpkin.pod,v 1.23 2000/01/13 19:45:13 doughera Released $