This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: 5.005 - a sneak preview
[perl5.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
11This document attempts to begin to describe some of the
12considerations involved in patching and maintaining perl.
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
AD
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 $] < 5.002;
61
aa689395 62You can also require particular version (or later) with
63
64 use 5.002;
65
7b5757d1
AD
66At some point in the future, we may need to decide what to call the
67next big revision. In the .package file used by metaconfig to
68generate Configure, there are two variables that might be relevant:
69$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
76In addition, there may be "developer" sub-versions available. These
77are not official releases. They may contain unstable experimental
78features, and are subject to rapid change. Such developer
79sub-versions are numbered with sub-version numbers. For example,
fb73857a 80version 5.003_04 is the 4'th developer version built on top of
815.003. It might include the _01, _02, and _03 changes, but it
82also might not. Sub-versions are allowed to be subversive. (But see
83the next section for recent changes.)
aa689395 84
85These sub-versions can also be used as floating point numbers, so
86you can do things such as
87
7b5757d1 88 print "You've got an unstable perl\n" if $] == 5.00303;
aa689395 89
90You can also require particular version (or later) with
91
7b5757d1 92 use 5.003_03; # the "_" is optional
aa689395 93
94Sub-versions produced by the members of perl5-porters are usually
95available on CPAN in the F<src/5.0/unsupported> directory.
96
7b5757d1
AD
97=head2 Maintenance and Development Subversions
98
99As an experiment, starting with version 5.004, subversions _01 through
100_49 will be reserved for bug-fix maintenance releases, and subversions
101_50 through _99 will be available for unstable development versions.
102
103The separate bug-fix track is being established to allow us an easy
104way to distribute important bug fixes without waiting for the
105developers to untangle all the other problems in the current
106developer's release.
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
111patch' entry in patchlevel.h.
112
7b5757d1
AD
113Watch for announcements of maintenance subversions in
114comp.lang.perl.announce.
115
20f245af
AD
116The first rule of maintenance work is "First, do no harm."
117
aa689395 118=head2 Why such a complicated scheme?
119
120Two reasons, really. At least.
121
7b5757d1
AD
122First, we need some way to identify and release collections of patches
123that are known to have new features that need testing and exploration. The
aa689395 124subversion scheme does that nicely while fitting into the
125C<use 5.004;> mold.
126
127Second, since most of the folks who help maintain perl do so on a
128free-time voluntary basis, perl development does not proceed at a
129precise pace, though it always seems to be moving ahead quickly.
130We needed some way to pass around the "patch pumpkin" to allow
131different people chances to work on different aspects of the
132distribution without getting in each other's way. It wouldn't be
133constructive to have multiple people working on incompatible
134implementations of the same idea. Instead what was needed was
135some kind of "baton" or "token" to pass around so everyone knew
136whose turn was next.
137
138=head2 Why is it called the patch pumpkin?
139
140Chip Salzenberg gets credit for that, with a nod to his cow orker,
141David Croy. We had passed around various names (baton, token, hot
142potato) but none caught on. Then, Chip asked:
143
144[begin quote]
145
146 Who has the patch pumpkin?
147
148To explain: David Croy once told me once that at a previous job,
149there was one tape drive and multiple systems that used it for backups.
150But instead of some high-tech exclusion software, they used a low-tech
151method to prevent multiple simultaneous backups: a stuffed pumpkin.
152No one was allowed to make backups unless they had the "backup pumpkin".
153
154[end quote]
155
156The name has stuck.
157
158=head1 Philosophical Issues in Patching Perl
159
160There are no absolute rules, but there are some general guidelines I
161have tried to follow as I apply patches to the perl sources.
162(This section is still under construction.)
163
164=head2 Solve problems as generally as possible
165
7b5757d1
AD
166Never implement a specific restricted solution to a problem when you
167can solve the same problem in a more general, flexible way.
168
169For example, for dynamic loading to work on some SVR4 systems, we had
170to build a shared libperl.so library. In order to build "FAT" binaries
171on NeXT 4.0 systems, we had to build a special libperl library. Rather
172than continuing to build a contorted nest of special cases, I
173generalized the process of building libperl so that NeXT and SVR4 users
174could still get their work done, but others could build a shared
175libperl if they wanted to as well.
aa689395 176
177=head2 Seek consensus on major changes
178
179If you are making big changes, don't do it in secret. Discuss the
180ideas in advance on perl5-porters.
181
182=head2 Keep the documentation up-to-date
183
184If your changes may affect how users use perl, then check to be sure
185that the documentation is in sync with your changes. Be sure to
186check all the files F<pod/*.pod> and also the F<INSTALL> document.
187
188Consider writing the appropriate documentation first and then
7b5757d1 189implementing your change to correspond to the documentation.
aa689395 190
191=head2 Avoid machine-specific #ifdef's
192
193To the extent reasonable, try to avoid machine-specific #ifdef's in
194the sources. Instead, use feature-specific #ifdef's. The reason is
195that the machine-specific #ifdef's may not be valid across major
196releases of the operating system. Further, the feature-specific tests
197may help out folks on another platform who have the same problem.
198
199=head2 Allow for lots of testing
200
201We should never release a main version without testing it as a
202subversion first.
203
6877a1cf
AD
204=head2 Test popular applications and modules.
205
206We should never release a main version without testing whether or not
207it breaks various popular modules and applications. A partial list of
208such things would include majordomo, metaconfig, apache, Tk, CGI,
209libnet, and libwww, to name just a few. Of course it's quite possible
210that some of those things will be just plain broken and need to be fixed,
211but, in general, we ought to try to avoid breaking widely-installed
212things.
213
7b5757d1 214=head2 Automate generation of derivative files
aa689395 215
216The F<embed.h>, F<keywords.h>, F<opcode.h>, and F<perltoc.pod> files
217are all automatically generated by perl scripts. In general, don't
218patch these directly; patch the data files instead.
219
220F<Configure> and F<config_h.SH> are also automatically generated by
221B<metaconfig>. In general, you should patch the metaconfig units
dfe9444c 222instead of patching these files directly. However, very minor changes to
aa689395 223F<Configure> may be made in between major sync-ups with the metaconfig
dfe9444c
AD
224units, which tends to be complicated operations. But be careful, this
225can quickly spiral out of control. Running metaconfig is not really
226hard.
aa689395 227
a8119d38
AD
228Finally, the sample files in the F<Porting/> subdirectory are
229generated automatically by the script F<U/mksample> included
230with the metaconfig units. See L<"run metaconfig"> below for
231information on obtaining the metaconfig units.
232
aa689395 233=head1 How to Make a Distribution
234
235There really ought to be a 'make dist' target, but there isn't.
236The 'dist' suite of tools also contains a number of tools that I haven't
237learned how to use yet. Some of them may make this all a bit easier.
238
239Here are the steps I go through to prepare a patch & distribution.
240
3e3baf6d
TB
241Lots of it could doubtless be automated but isn't. The Porting/makerel
242(make release) perl script does now help automate some parts of it.
aa689395 243
244=head2 Announce your intentions
245
246First, you should volunteer out loud to take the patch pumpkin. It's
247generally counter-productive to have multiple people working in secret
248on the same thing.
249
250At the same time, announce what you plan to do with the patch pumpkin,
251to allow folks a chance to object or suggest alternatives, or do it for
252you. Naturally, the patch pumpkin holder ought to incorporate various
253bug fixes and documentation improvements that are posted while he or
254she has the pumpkin, but there might also be larger issues at stake.
255
256One of the precepts of the subversion idea is that we shouldn't give
7b5757d1
AD
257the patch pumpkin to anyone unless we have some idea what he or she
258is going to do with it.
aa689395 259
260=head2 refresh pod/perltoc.pod
261
262Presumably, you have done a full C<make> in your working source
263directory. Before you C<make spotless> (if you do), and if you have
264changed any documentation in any module or pod file, change to the
265F<pod> directory and run C<make toc>.
266
3e3baf6d
TB
267=head2 run installhtml to check the validity of the pod files
268
aa689395 269=head2 update patchlevel.h
270
271Don't be shy about using the subversion number, even for a relatively
272modest patch. We've never even come close to using all 99 subversions,
273and it's better to have a distinctive number for your patch. If you
274need feedback on your patch, go ahead and issue it and promise to
275incorporate that feedback quickly (e.g. within 1 week) and send out a
276second patch.
277
278=head2 run metaconfig
279
280If you need to make changes to Configure or config_h.SH, it may be best to
281change the appropriate metaconfig units instead, and regenerate Configure.
282
283 metaconfig -m
284
20f245af
AD
285will regenerate Configure and config_h.SH. Much more information
286on obtaining and running metaconfig is in the F<U/README> file
287that comes with Perl's metaconfig units. Perl's metaconfig units
288should be available on CPAN. A set of units that will work with
289perl5.005 is in the file F<mc_units-5.005_00-01.tar.gz> under
290http://www.perl.com/CPAN/authors/id/ANDYD/ . The mc_units tar file
291should be unpacked in your main perl source directory. Note: those
292units were for use with 5.005. There may have been changes since then.
d562869c 293Check for later versions or contact perl5-porters@perl.org to obtain a
20f245af 294pointer to the current version.
aa689395 295
296Alternatively, do consider if the F<*ish.h> files might be a better
297place for your changes.
298
299=head2 MANIFEST
300
301Make sure the MANIFEST is up-to-date. You can use dist's B<manicheck>
302program for this. You can also use
303
3e3baf6d 304 perl -w -MExtUtils::Manifest=fullcheck -e fullcheck
aa689395 305
3e3baf6d
TB
306Both commands will also list extra files in the directory that are not
307listed in MANIFEST.
aa689395 308
bfb7748a 309The MANIFEST is normally sorted.
aa689395 310
311If you are using metaconfig to regenerate Configure, then you should note
312that metaconfig actually uses MANIFEST.new, so you want to be sure
313MANIFEST.new is up-to-date too. I haven't found the MANIFEST/MANIFEST.new
314distinction particularly useful, but that's probably because I still haven't
315learned how to use the full suite of tools in the dist distribution.
316
317=head2 Check permissions
318
319All the tests in the t/ directory ought to be executable. The
320main makefile used to do a 'chmod t/*/*.t', but that resulted in
321a self-modifying distribution--something some users would strongly
d562869c
AD
322prefer to avoid. The F<t/TEST> script will check for this
323and do the chmod if needed, but the tests still ought to be
324executable.
aa689395 325
326In all, the following files should probably be executable:
327
328 Configure
329 configpm
32fcaa0b 330 configure.gnu
aa689395 331 embed.pl
332 installperl
333 installman
334 keywords.pl
aa689395 335 myconfig
336 opcode.pl
337 perly.fixer
338 t/TEST
339 t/*/*.t
340 *.SH
341 vms/ext/Stdio/test.pl
342 vms/ext/filespec.t
aa689395 343 x2p/*.SH
344
345Other things ought to be readable, at least :-).
346
347Probably, the permissions for the files could be encoded in MANIFEST
348somehow, but I'm reluctant to change MANIFEST itself because that
349could break old scripts that use MANIFEST.
350
351I seem to recall that some SVR3 systems kept some sort of file that listed
352permissions for system files; something like that might be appropriate.
353
354=head2 Run Configure
355
356This will build a config.sh and config.h. You can skip this if you haven't
693762b4 357changed Configure or config_h.SH at all. I use the following command
aa689395 358
693762b4
AD
359 sh Configure -Dprefix=/opt/perl -Doptimize=-O -Dusethreads \
360 -Dcf_by='yourname' \
361 -Dcf_email='yourname@yourhost.yourplace.com' \
362 -Dperladmin='yourname@yourhost.yourplace.com' \
363 -Dmydomain='.yourplace.com' \
364 -Dmyhostname='yourhost' \
365 -des
aa689395 366
693762b4 367=head2 Update Porting/config.sh and Porting/config_H
dfe9444c 368
693762b4
AD
369[XXX
370This section needs revision. We're currently working on easing
371the task of keeping the vms, win32, and plan9 config.sh info
372up-to-date. The plan is to use keep up-to-date 'canned' config.sh
373files in the appropriate subdirectories and then generate 'canned'
374config.h files for vms, win32, etc. from the generic config.sh file.
375This is to ease maintenance. When Configure gets updated, the parts
376sometimes get scrambled around, and the changes in config_H can
377sometimes be very hard to follow. config.sh, on the other hand, can
378safely be sorted, so it's easy to track (typically very small) changes
379to config.sh and then propoagate them to a canned 'config.h' by any
380number of means, including a perl script in win32/ or carrying
381config.sh and config_h.SH to a Unix system and running sh
382config_h.SH.)
383XXX]
384
385The Porting/config.sh and Porting/config_H files are provided to
386help those folks who can't run Configure. It is important to keep
387them up-to-date. If you have changed config_h.SH, those changes must
388be reflected in config_H as well. (The name config_H was chosen to
389distinguish the file from config.h even on case-insensitive file systems.)
390Simply edit the existing config_H file; keep the first few explanatory
391lines and then copy your new config.h below.
aa689395 392
d562869c 393It may also be necessary to update win32/config.?c, vms/config.vms and
aa689395 394plan9/config.plan9, though you should be quite careful in doing so if
395you are not familiar with those systems. You might want to issue your
396patch with a promise to quickly issue a follow-up that handles those
397directories.
398
399=head2 make run_byacc
400
401If you have byacc-1.8.2 (available from CPAN), and if there have been
402changes to F<perly.y>, you can regenerate the F<perly.c> file. The
403run_byacc makefile target does this by running byacc and then applying
404some patches so that byacc dynamically allocates space, rather than
405having fixed limits. This patch is handled by the F<perly.fixer>
406script. Depending on the nature of the changes to F<perly.y>, you may
407or may not have to hand-edit the patch to apply correctly. If you do,
408you should include the edited patch in the new distribution. If you
409have byacc-1.9, the patch won't apply cleanly. Changes to the printf
410output statements mean the patch won't apply cleanly. Long ago I
411started to fix F<perly.fixer> to detect this, but I never completed the
412task.
413
414Some additional notes from Larry on this:
415
e262e9be 416Don't forget to regenerate perly_c.diff.
aa689395 417
7b5757d1 418 byacc -d perly.y
aa689395 419 mv y.tab.c perly.c
e262e9be 420 patch perly.c <perly_c.diff
aa689395 421 # manually apply any failed hunks
e262e9be 422 diff -c2 perly.c.orig perly.c >perly_c.diff
aa689395 423
424One chunk of lines that often fails begins with
425
426 #line 29 "perly.y"
427
428and ends one line before
429
430 #define YYERRCODE 256
431
432This only happens when you add or remove a token type. I suppose this
433could be automated, but it doesn't happen very often nowadays.
434
435Larry
436
437=head2 make regen_headers
438
439The F<embed.h>, F<keywords.h>, and F<opcode.h> files are all automatically
440generated by perl scripts. Since the user isn't guaranteed to have a
441working perl, we can't require the user to generate them. Hence you have
442to, if you're making a distribution.
443
444I used to include rules like the following in the makefile:
445
446 # The following three header files are generated automatically
447 # The correct versions should be already supplied with the perl kit,
448 # in case you don't have perl or 'sh' available.
449 # The - is to ignore error return codes in case you have the source
450 # installed read-only or you don't have perl yet.
451 keywords.h: keywords.pl
452 @echo "Don't worry if this fails."
453 - perl keywords.pl
454
455
7b5757d1 456However, I got B<lots> of mail consisting of people worrying because the
aa689395 457command failed. I eventually decided that I would save myself time
458and effort by manually running C<make regen_headers> myself rather
459than answering all the questions and complaints about the failing
460command.
461
3e3baf6d 462=head2 global.sym, interp.sym and perlio.sym
aa689395 463
464Make sure these files are up-to-date. Read the comments in these
465files and in perl_exp.SH to see what to do.
466
467=head2 Binary compatibility
468
469If you do change F<global.sym> or F<interp.sym>, think carefully about
470what you are doing. To the extent reasonable, we'd like to maintain
471souce and binary compatibility with older releases of perl. That way,
472extensions built under one version of perl will continue to work with
473new versions of perl.
474
475Of course, some incompatible changes may well be necessary. I'm just
476suggesting that we not make any such changes without thinking carefully
477about them first. If possible, we should provide
478backwards-compatibility stubs. There's a lot of XS code out there.
479Let's not force people to keep changing it.
480
481=head2 Changes
482
483Be sure to update the F<Changes> file. Try to include both an overall
484summary as well as detailed descriptions of the changes. Your
3e3baf6d 485audience will include other developers and users, so describe
aa689395 486user-visible changes (if any) in terms they will understand, not in
487code like "initialize foo variable in bar function".
488
489There are differing opinions on whether the detailed descriptions
490ought to go in the Changes file or whether they ought to be available
491separately in the patch file (or both). There is no disagreement that
492detailed descriptions ought to be easily available somewhere.
493
2a26e2f1
DD
494=head2 Todo
495
496The F<Todo> file contains a roughly-catgorized unordered list of
497aspects of Perl that could use enhancement, features that could be
498added, areas that could be cleaned up, and so on. During your term as
499pumpkin-holder, you will probably address some of these issues, and
500perhaps identify others which, while you decide not to address them
501this time around, may be tackled in the future. Update the file
502reflect the situation as it stands when you hand over the pumpkin.
503
504You might like, early in your pumpkin-holding career, to see if you
505can find champions for partiticular issues on the to-do list: an issue
506owned is an issue more likely to be resolved.
507
c4f23d77
AD
508There are also some more porting-specific L<Todo> items later in this
509file.
510
aa689395 511=head2 OS/2-specific updates
512
513In the os2 directory is F<diff.configure>, a set of OS/2-specific
514diffs against B<Configure>. If you make changes to Configure, you may
515want to consider regenerating this diff file to save trouble for the
516OS/2 maintainer.
517
7b5757d1
AD
518You can also consider the OS/2 diffs as reminders of portability
519things that need to be fixed in Configure.
520
aa689395 521=head2 VMS-specific updates
522
523If you have changed F<perly.y>, then you may want to update
524F<vms/perly_{h,c}.vms> by running C<perl vms/vms_yfix.pl>.
525
526The Perl version number appears in several places under F<vms>.
527It is courteous to update these versions. For example, if you are
528making 5.004_42, replace "5.00441" with "5.00442".
529
530=head2 Making the new distribution
531
532Suppose, for example, that you want to make version 5.004_08. Then you can
533do something like the following
534
535 mkdir ../perl5.004_08
536 awk '{print $1}' MANIFEST | cpio -pdm ../perl5.004_08
537 cd ../
538 tar cf perl5.004_08.tar perl5.004_08
539 gzip --best perl5.004_08.tar
540
3e3baf6d
TB
541These steps, with extra checks, are automated by the Porting/makerel
542script.
543
aa689395 544=head2 Making a new patch
545
546I find the F<makepatch> utility quite handy for making patches.
547You can obtain it from any CPAN archive under
3e3baf6d
TB
548http://www.perl.com/CPAN/authors/Johan_Vromans/ . There are a couple
549of differences between my version and the standard one. I have mine do
550a
aa689395 551
552 # Print a reassuring "End of Patch" note so people won't
553 # wonder if their mailer truncated patches.
554 print "\n\nEnd of Patch.\n";
555
3e3baf6d
TB
556at the end. That's because I used to get questions from people asking
557if their mail was truncated.
558
559It also writes Index: lines which include the new directory prefix
560(change Index: print, approx line 294 or 310 depending on the version,
561to read: print PATCH ("Index: $newdir$new\n");). That helps patches
562work with more POSIX conformant patch programs.
aa689395 563
564Here's how I generate a new patch. I'll use the hypothetical
5655.004_07 to 5.004_08 patch as an example.
566
567 # unpack perl5.004_07/
568 gzip -d -c perl5.004_07.tar.gz | tar -xof -
569 # unpack perl5.004_08/
570 gzip -d -c perl5.004_08.tar.gz | tar -xof -
571 makepatch perl5.004_07 perl5.004_08 > perl5.004_08.pat
572
573Makepatch will automatically generate appropriate B<rm> commands to remove
574deleted files. Unfortunately, it will not correctly set permissions
575for newly created files, so you may have to do so manually. For example,
576patch 5.003_04 created a new test F<t/op/gv.t> which needs to be executable,
577so at the top of the patch, I inserted the following lines:
578
579 # Make a new test
580 touch t/op/gv.t
581 chmod +x t/opt/gv.t
582
583Now, of course, my patch is now wrong because makepatch didn't know I
584was going to do that command, and it patched against /dev/null.
585
586So, what I do is sort out all such shell commands that need to be in the
587patch (including possible mv-ing of files, if needed) and put that in the
588shell commands at the top of the patch. Next, I delete all the patch parts
589of perl5.004_08.pat, leaving just the shell commands. Then, I do the
590following:
591
7b5757d1
AD
592 cd perl5.004_07
593 sh ../perl5.004_08.pat
aa689395 594 cd ..
7b5757d1 595 makepatch perl5.004_07 perl5.004_08 >> perl5.004_08.pat
aa689395 596
597(Note the append to preserve my shell commands.)
598Now, my patch will line up with what the end users are going to do.
599
600=head2 Testing your patch
601
602It seems obvious, but be sure to test your patch. That is, verify that
603it produces exactly the same thing as your full distribution.
604
7b5757d1
AD
605 rm -rf perl5.004_07
606 gzip -d -c perl5.004_07.tar.gz | tar -xf -
607 cd perl5.004_07
608 sh ../perl5.004_08.pat
609 patch -p1 -N < ../perl5.004_08.pat
aa689395 610 cd ..
7b5757d1 611 gdiff -r perl5.004_07 perl5.004_08
aa689395 612
613where B<gdiff> is GNU diff. Other diff's may also do recursive checking.
614
615=head2 More testing
616
617Again, it's obvious, but you should test your new version as widely as you
618can. You can be sure you'll hear about it quickly if your version doesn't
619work on both ANSI and pre-ANSI compilers, and on common systems such as
620SunOS 4.1.[34], Solaris, and Linux.
621
622If your changes include conditional code, try to test the different
623branches as thoroughly as you can. For example, if your system
624supports dynamic loading, you can also test static loading with
625
626 sh Configure -Uusedl
627
628You can also hand-tweak your config.h to try out different #ifdef
629branches.
630
631=head1 Common Gotcha's
632
633=over 4
634
635=item #elif
636
637The '#elif' preprocessor directive is not understood on all systems.
638Specifically, I know that Pyramids don't understand it. Thus instead of the
639simple
640
641 #if defined(I_FOO)
642 # include <foo.h>
643 #elif defined(I_BAR)
644 # include <bar.h>
645 #else
646 # include <fubar.h>
647 #endif
648
649You have to do the more Byzantine
650
651 #if defined(I_FOO)
652 # include <foo.h>
653 #else
654 # if defined(I_BAR)
655 # include <bar.h>
656 # else
657 # include <fubar.h>
658 # endif
659 #endif
660
661Incidentally, whitespace between the leading '#' and the preprocessor
662command is not guaranteed, but is very portable and you may use it freely.
663I think it makes things a bit more readable, especially once things get
664rather deeply nested. I also think that things should almost never get
665too deeply nested, so it ought to be a moot point :-)
666
667=item Probably Prefer POSIX
668
669It's often the case that you'll need to choose whether to do
670something the BSD-ish way or the POSIX-ish way. It's usually not
671a big problem when the two systems use different names for similar
672functions, such as memcmp() and bcmp(). The perl.h header file
673handles these by appropriate #defines, selecting the POSIX mem*()
674functions if available, but falling back on the b*() functions, if
675need be.
676
677More serious is the case where some brilliant person decided to
678use the same function name but give it a different meaning or
679calling sequence :-). getpgrp() and setpgrp() come to mind.
680These are a real problem on systems that aim for conformance to
681one standard (e.g. POSIX), but still try to support the other way
682of doing things (e.g. BSD). My general advice (still not really
683implemented in the source) is to do something like the following.
684Suppose there are two alternative versions, fooPOSIX() and
685fooBSD().
686
687 #ifdef HAS_FOOPOSIX
688 /* use fooPOSIX(); */
689 #else
690 # ifdef HAS_FOOBSD
691 /* try to emulate fooPOSIX() with fooBSD();
692 perhaps with the following: */
693 # define fooPOSIX fooBSD
694 # else
695 # /* Uh, oh. We have to supply our own. */
696 # define fooPOSIX Perl_fooPOSIX
697 # endif
698 #endif
699
700=item Think positively
701
702If you need to add an #ifdef test, it is usually easier to follow if you
703think positively, e.g.
704
705 #ifdef HAS_NEATO_FEATURE
706 /* use neato feature */
707 #else
708 /* use some fallback mechanism */
709 #endif
710
711rather than the more impenetrable
712
713 #ifndef MISSING_NEATO_FEATURE
714 /* Not missing it, so we must have it, so use it */
715 #else
716 /* Are missing it, so fall back on something else. */
717 #endif
718
719Of course for this toy example, there's not much difference. But when
720the #ifdef's start spanning a couple of screen fulls, and the #else's
721are marked something like
722
723 #else /* !MISSING_NEATO_FEATURE */
724
725I find it easy to get lost.
726
727=item Providing Missing Functions -- Problem
728
729Not all systems have all the neat functions you might want or need, so
730you might decide to be helpful and provide an emulation. This is
731sound in theory and very kind of you, but please be careful about what
732you name the function. Let me use the C<pause()> function as an
733illustration.
734
735Perl5.003 has the following in F<perl.h>
736
737 #ifndef HAS_PAUSE
738 #define pause() sleep((32767<<16)+32767)
739 #endif
740
741Configure sets HAS_PAUSE if the system has the pause() function, so
742this #define only kicks in if the pause() function is missing.
743Nice idea, right?
744
745Unfortunately, some systems apparently have a prototype for pause()
746in F<unistd.h>, but don't actually have the function in the library.
747(Or maybe they do have it in a library we're not using.)
748
749Thus, the compiler sees something like
750
751 extern int pause(void);
752 /* . . . */
753 #define pause() sleep((32767<<16)+32767)
754
755and dies with an error message. (Some compilers don't mind this;
756others apparently do.)
757
758To work around this, 5.003_03 and later have the following in perl.h:
759
760 /* Some unistd.h's give a prototype for pause() even though
761 HAS_PAUSE ends up undefined. This causes the #define
762 below to be rejected by the compiler. Sigh.
763 */
764 #ifdef HAS_PAUSE
765 # define Pause pause
766 #else
767 # define Pause() sleep((32767<<16)+32767)
768 #endif
769
770This works.
771
772The curious reader may wonder why I didn't do the following in
773F<util.c> instead:
774
775 #ifndef HAS_PAUSE
776 void pause()
777 {
778 sleep((32767<<16)+32767);
779 }
780 #endif
781
782That is, since the function is missing, just provide it.
783Then things would probably be been alright, it would seem.
784
785Well, almost. It could be made to work. The problem arises from the
786conflicting needs of dynamic loading and namespace protection.
787
788For dynamic loading to work on AIX (and VMS) we need to provide a list
789of symbols to be exported. This is done by the script F<perl_exp.SH>,
790which reads F<global.sym> and F<interp.sym>. Thus, the C<pause>
791symbol would have to be added to F<global.sym> So far, so good.
792
793On the other hand, one of the goals of Perl5 is to make it easy to
794either extend or embed perl and link it with other libraries. This
795means we have to be careful to keep the visible namespace "clean".
796That is, we don't want perl's global variables to conflict with
797those in the other application library. Although this work is still
798in progress, the way it is currently done is via the F<embed.h> file.
799This file is built from the F<global.sym> and F<interp.sym> files,
800since those files already list the globally visible symbols. If we
801had added C<pause> to global.sym, then F<embed.h> would contain the
802line
803
804 #define pause Perl_pause
805
806and calls to C<pause> in the perl sources would now point to
807C<Perl_pause>. Now, when B<ld> is run to build the F<perl> executable,
808it will go looking for C<perl_pause>, which probably won't exist in any
809of the standard libraries. Thus the build of perl will fail.
810
811Those systems where C<HAS_PAUSE> is not defined would be ok, however,
812since they would get a C<Perl_pause> function in util.c. The rest of
813the world would be in trouble.
814
815And yes, this scenario has happened. On SCO, the function C<chsize>
816is available. (I think it's in F<-lx>, the Xenix compatibility
817library.) Since the perl4 days (and possibly before), Perl has
818included a C<chsize> function that gets called something akin to
819
820 #ifndef HAS_CHSIZE
821 I32 chsize(fd, length)
822 /* . . . */
823 #endif
824
825When 5.003 added
826
827 #define chsize Perl_chsize
828
829to F<embed.h>, the compile started failing on SCO systems.
830
831The "fix" is to give the function a different name. The one
832implemented in 5.003_05 isn't optimal, but here's what was done:
833
834 #ifdef HAS_CHSIZE
835 # ifdef my_chsize /* Probably #defined to Perl_my_chsize in embed.h */
836 # undef my_chsize
837 # endif
838 # define my_chsize chsize
839 #endif
840
841My explanatory comment in patch 5.003_05 said:
842
843 Undef and then re-define my_chsize from Perl_my_chsize to
844 just plain chsize if this system HAS_CHSIZE. This probably only
845 applies to SCO. This shows the perils of having internal
846 functions with the same name as external library functions :-).
847
848Now, we can safely put C<my_chsize> in F<global.sym>, export it, and
849hide it with F<embed.h>.
850
851To be consistent with what I did for C<pause>, I probably should have
852called the new function C<Chsize>, rather than C<my_chsize>.
853However, the perl sources are quite inconsistent on this (Consider
854New, Mymalloc, and Myremalloc, to name just a few.)
855
856There is a problem with this fix, however, in that C<Perl_chsize>
857was available as a F<libperl.a> library function in 5.003, but it
858isn't available any more (as of 5.003_07). This means that we've
859broken binary compatibility. This is not good.
860
861=item Providing missing functions -- some ideas
862
863We currently don't have a standard way of handling such missing
864function names. Right now, I'm effectively thinking aloud about a
865solution. Some day, I'll try to formally propose a solution.
866
867Part of the problem is that we want to have some functions listed as
868exported but not have their names mangled by embed.h or possibly
869conflict with names in standard system headers. We actually already
870have such a list at the end of F<perl_exp.SH> (though that list is
871out-of-date):
872
873 # extra globals not included above.
874 cat <<END >> perl.exp
875 perl_init_ext
876 perl_init_fold
877 perl_init_i18nl14n
878 perl_alloc
879 perl_construct
880 perl_destruct
881 perl_free
882 perl_parse
883 perl_run
884 perl_get_sv
885 perl_get_av
886 perl_get_hv
887 perl_get_cv
888 perl_call_argv
889 perl_call_pv
890 perl_call_method
891 perl_call_sv
892 perl_requirepv
893 safecalloc
894 safemalloc
895 saferealloc
896 safefree
897
898This still needs much thought, but I'm inclined to think that one
899possible solution is to prefix all such functions with C<perl_> in the
900source and list them along with the other C<perl_*> functions in
901F<perl_exp.SH>.
902
903Thus, for C<chsize>, we'd do something like the following:
904
905 /* in perl.h */
906 #ifdef HAS_CHSIZE
907 # define perl_chsize chsize
908 #endif
909
910then in some file (e.g. F<util.c> or F<doio.c>) do
911
912 #ifndef HAS_CHSIZE
913 I32 perl_chsize(fd, length)
914 /* implement the function here . . . */
915 #endif
916
917Alternatively, we could just always use C<chsize> everywhere and move
918C<chsize> from F<global.sym> to the end of F<perl_exp.SH>. That would
919probably be fine as long as our C<chsize> function agreed with all the
920C<chsize> function prototypes in the various systems we'll be using.
921As long as the prototypes in actual use don't vary that much, this is
922probably a good alternative. (As a counter-example, note how Configure
923and perl have to go through hoops to find and use get Malloc_t and
924Free_t for C<malloc> and C<free>.)
925
926At the moment, this latter option is what I tend to prefer.
927
928=item All the world's a VAX
929
930Sorry, showing my age:-). Still, all the world is not BSD 4.[34],
931SVR4, or POSIX. Be aware that SVR3-derived systems are still quite
932common (do you have any idea how many systems run SCO?) If you don't
933have a bunch of v7 manuals handy, the metaconfig units (by default
934installed in F</usr/local/lib/dist/U>) are a good resource to look at
935for portability.
936
937=back
938
939=head1 Miscellaneous Topics
940
941=head2 Autoconf
942
943Why does perl use a metaconfig-generated Configure script instead of an
944autoconf-generated configure script?
945
946Metaconfig and autoconf are two tools with very similar purposes.
947Metaconfig is actually the older of the two, and was originally written
948by Larry Wall, while autoconf is probably now used in a wider variety of
949packages. The autoconf info file discusses the history of autoconf and
950how it came to be. The curious reader is referred there for further
951information.
952
953Overall, both tools are quite good, I think, and the choice of which one
954to use could be argued either way. In March, 1994, when I was just
955starting to work on Configure support for Perl5, I considered both
956autoconf and metaconfig, and eventually decided to use metaconfig for the
957following reasons:
958
959=over 4
960
961=item Compatibility with Perl4
962
963Perl4 used metaconfig, so many of the #ifdef's were already set up for
964metaconfig. Of course metaconfig had evolved some since Perl4's days,
965but not so much that it posed any serious problems.
966
967=item Metaconfig worked for me
968
969My system at the time was Interactive 2.2, a SVR3.2/386 derivative that
970also had some POSIX support. Metaconfig-generated Configure scripts
971worked fine for me on that system. On the other hand, autoconf-generated
972scripts usually didn't. (They did come quite close, though, in some
973cases.) At the time, I actually fetched a large number of GNU packages
974and checked. Not a single one configured and compiled correctly
975out-of-the-box with the system's cc compiler.
976
977=item Configure can be interactive
978
979With both autoconf and metaconfig, if the script works, everything is
980fine. However, one of my main problems with autoconf-generated scripts
981was that if it guessed wrong about something, it could be B<very> hard to
982go back and fix it. For example, autoconf always insisted on passing the
983-Xp flag to cc (to turn on POSIX behavior), even when that wasn't what I
984wanted or needed for that package. There was no way short of editing the
985configure script to turn this off. You couldn't just edit the resulting
986Makefile at the end because the -Xp flag influenced a number of other
987configure tests.
988
989Metaconfig's Configure scripts, on the other hand, can be interactive.
990Thus if Configure is guessing things incorrectly, you can go back and fix
991them. This isn't as important now as it was when we were actively
992developing Configure support for new features such as dynamic loading,
993but it's still useful occasionally.
994
995=item GPL
996
997At the time, autoconf-generated scripts were covered under the GNU Public
998License, and hence weren't suitable for inclusion with Perl, which has a
999different licensing policy. (Autoconf's licensing has since changed.)
1000
1001=item Modularity
1002
1003Metaconfig builds up Configure from a collection of discrete pieces
1004called "units". You can override the standard behavior by supplying your
1005own unit. With autoconf, you have to patch the standard files instead.
1006I find the metaconfig "unit" method easier to work with. Others
1007may find metaconfig's units clumsy to work with.
1008
1009=back
1010
1011=head2 @INC search order
1012
1013By default, the list of perl library directories in @INC is the
1014following:
1015
1016 $archlib
1017 $privlib
1018 $sitearch
1019 $sitelib
1020
1021Specifically, on my Solaris/x86 system, I run
1022B<sh Configure -Dprefix=/opt/perl> and I have the following
1023directories:
1024
1025 /opt/perl/lib/i86pc-solaris/5.00307
1026 /opt/perl/lib
1027 /opt/perl/lib/site_perl/i86pc-solaris
1028 /opt/perl/lib/site_perl
1029
1030That is, perl's directories come first, followed by the site-specific
1031directories.
1032
1033The site libraries come second to support the usage of extensions
1034across perl versions. Read the relevant section in F<INSTALL> for
1035more information. If we ever make $sitearch version-specific, this
1036topic could be revisited.
1037
1038=head2 Why isn't there a directory to override Perl's library?
1039
1040Mainly because no one's gotten around to making one. Note that
1041"making one" involves changing perl.c, Configure, config_h.SH (and
1042associated files, see above), and I<documenting> it all in the
1043INSTALL file.
1044
1045Apparently, most folks who want to override one of the standard library
1046files simply do it by overwriting the standard library files.
1047
1048=head2 APPLLIB
1049
1050In the perl.c sources, you'll find an undocumented APPLLIB_EXP
1051variable, sort of like PRIVLIB_EXP and ARCHLIB_EXP (which are
1052documented in config_h.SH). Here's what APPLLIB_EXP is for, from
1053a mail message from Larry:
1054
1055 The main intent of APPLLIB_EXP is for folks who want to send out a
1056 version of Perl embedded in their product. They would set the symbol
1057 to be the name of the library containing the files needed to run or to
1058 support their particular application. This works at the "override"
1059 level to make sure they get their own versions of any library code that
1060 they absolutely must have configuration control over.
1061
1062 As such, I don't see any conflict with a sysadmin using it for a
1063 override-ish sort of thing, when installing a generic Perl. It should
1064 probably have been named something to do with overriding though. Since
1065 it's undocumented we could still change it... :-)
1066
1067Given that it's already there, you can use it to override
1068distribution modules. If you do
1069
1070 sh Configure -Dccflags='-DAPPLLIB_EXP=/my/override'
1071
1072then perl.c will put /my/override ahead of ARCHLIB and PRIVLIB.
1073
c4f23d77
AD
1074=head2 Shared libperl.so location
1075
1076Why isn't the shared libperl.so installed in /usr/lib/ along
1077with "all the other" shared libraries? Instead, it is installed
1078in $archlib, which is typically something like
1079
1080 /usr/local/lib/perl5/archname/5.00404
1081
1082and is architecture- and version-specific.
1083
1084The basic reason why a shared libperl.so gets put in $archlib is so that
1085you can have more than one version of perl on the system at the same time,
1086and have each refer to its own libperl.so.
1087
1088Three examples might help. All of these work now; none would work if you
1089put libperl.so in /usr/lib.
1090
1091=over
1092
1093=item 1.
1094
1095Suppose you want to have both threaded and non-threaded perl versions
1096around. Configure will name both perl libraries "libperl.so" (so that
1097you can link to them with -lperl). The perl binaries tell them apart
1098by having looking in the appropriate $archlib directories.
1099
1100=item 2.
1101
1102Suppose you have perl5.004_04 installed and you want to try to compile
1103it again, perhaps with different options or after applying a patch.
1104If you already have libperl.so installed in /usr/lib/, then it may be
1105either difficult or impossible to get ld.so to find the new libperl.so
1106that you're trying to build. If, instead, libperl.so is tucked away in
1107$archlib, then you can always just change $archlib in the current perl
1108you're trying to build so that ld.so won't find your old libperl.so.
1109(The INSTALL file suggests you do this when building a debugging perl.)
1110
1111=item 3.
1112
1113The shared perl library is not a "well-behaved" shared library with
1114proper major and minor version numbers, so you can't necessarily
1115have perl5.004_04 and perl5.004_05 installed simultaneously. Suppose
1116perl5.004_04 were to install /usr/lib/libperl.so.4.4, and perl5.004_05
1117were to install /usr/lib/libperl.so.4.5. Now, when you try to run
1118perl5.004_04, ld.so might try to load libperl.so.4.5, since it has
1119the right "major version" number. If this works at all, it almost
1120certainly defeats the reason for keeping perl5.004_04 around. Worse,
1121with development subversions, you certaily can't guarantee that
1122libperl.so.4.4 and libperl.so.4.55 will be compatible.
1123
1124Anyway, all this leads to quite obscure failures that are sure to drive
1125casual users crazy. Even experienced users will get confused :-). Upon
1126reflection, I'd say leave libperl.so in $archlib.
1127
1128=back
1129
aa689395 1130=head1 Upload Your Work to CPAN
1131
1132You can upload your work to CPAN if you have a CPAN id. Check out
1133http://www.perl.com/CPAN/modules/04pause.html for information on
1134_PAUSE_, the Perl Author's Upload Server.
1135
1136I typically upload both the patch file, e.g. F<perl5.004_08.pat.gz>
1137and the full tar file, e.g. F<perl5.004_08.tar.gz>.
1138
1139If you want your patch to appear in the F<src/5.0/unsupported>
1140directory on CPAN, send e-mail to the CPAN master librarian. (Check
7b5757d1 1141out http://www.perl.com/CPAN/CPAN.html ).
aa689395 1142
1143=head1 Help Save the World
1144
1145You should definitely announce your patch on the perl5-porters list.
1146You should also consider announcing your patch on
1147comp.lang.perl.announce, though you should make it quite clear that a
1148subversion is not a production release, and be prepared to deal with
1149people who will not read your disclaimer.
1150
1151=head1 Todo
1152
1153Here, in no particular order, are some Configure and build-related
1154items that merit consideration. This list isn't exhaustive, it's just
1155what I came up with off the top of my head.
1156
1157=head2 Good ideas waiting for round tuits
1158
1159=over 4
1160
1161=item installprefix
1162
1163I think we ought to support
1164
1165 Configure -Dinstallprefix=/blah/blah
1166
1167Currently, we support B<-Dprefix=/blah/blah>, but the changing the install
1168location has to be handled by something like the F<config.over> trick
1169described in F<INSTALL>. AFS users also are treated specially.
1170We should probably duplicate the metaconfig prefix stuff for an
1171install prefix.
1172
c4f23d77 1173=item Configure -Dsrc=/blah/blah
aa689395 1174
1175We should be able to emulate B<configure --srcdir>. Tom Tromey
1176tromey@creche.cygnus.com has submitted some patches to
c4f23d77
AD
1177the dist-users mailing list along these lines. They have been folded
1178back into the main distribution, but various parts of the perl
1179Configure/build/install process still assume src='.'.
aa689395 1180
1181=item Hint file fixes
1182
1183Various hint files work around Configure problems. We ought to fix
1184Configure so that most of them aren't needed.
1185
1186=item Hint file information
1187
1188Some of the hint file information (particularly dynamic loading stuff)
1189ought to be fed back into the main metaconfig distribution.
1190
c4f23d77
AD
1191=item Catch GNU Libc "Stub" functions
1192
1193Some functions (such as lchown()) are present in libc, but are
1194unimplmented. That is, they always fail and set errno=ENOSYS.
1195
1196Thomas Bushnell provided the following sample code and the explanation
1197that follows:
1198
1199 /* System header to define __stub macros and hopefully few prototypes,
1200 which can conflict with char FOO(); below. */
1201 #include <assert.h>
1202 /* Override any gcc2 internal prototype to avoid an error. */
1203 /* We use char because int might match the return type of a gcc2
1204 builtin and then its argument prototype would still apply. */
1205 char FOO();
1206
1207 int main() {
1208
1209 /* The GNU C library defines this for functions which it implements
1210 to always fail with ENOSYS. Some functions are actually named
1211 something starting with __ and the normal name is an alias. */
1212 #if defined (__stub_FOO) || defined (__stub___FOO)
1213 choke me
1214 #else
1215 FOO();
1216 #endif
1217
1218 ; return 0; }
1219
1220The choice of <assert.h> is essentially arbitrary. The GNU libc
1221macros are found in <gnu/stubs.h>. You can include that file instead
1222of <assert.h> (which itself includes <gnu/stubs.h>) if you test for
1223its existence first. <assert.h> is assumed to exist on every system,
1224which is why it's used here. Any GNU libc header file will include
1225the stubs macros. If either __stub_NAME or __stub___NAME is defined,
1226then the function doesn't actually exist. Tests using <assert.h> work
1227on every system around.
1228
1229The declaration of FOO is there to override builtin prototypes for
1230ANSI C functions.
1231
aa689395 1232=back
1233
1234=head2 Probably good ideas waiting for round tuits
1235
1236=over 4
1237
1238=item GNU configure --options
1239
1240I've received sensible suggestions for --exec_prefix and other
1241GNU configure --options. It's not always obvious exactly what is
1242intended, but this merits investigation.
1243
1244=item make clean
1245
1246Currently, B<make clean> isn't all that useful, though
1247B<make realclean> and B<make distclean> are. This needs a bit of
1248thought and documentation before it gets cleaned up.
1249
1250=item Try gcc if cc fails
1251
1252Currently, we just give up.
1253
1254=item bypassing safe*alloc wrappers
1255
1256On some systems, it may be safe to call the system malloc directly
1257without going through the util.c safe* layers. (Such systems would
1258accept free(0), for example.) This might be a time-saver for systems
1259that already have a good malloc. (Recent Linux libc's apparently have
1260a nice malloc that is well-tuned for the system.)
1261
1262=back
1263
1264=head2 Vague possibilities
1265
1266=over 4
1267
aa689395 1268=item MacPerl
1269
3e3baf6d 1270Get some of the Macintosh stuff folded back into the main distribution.
aa689395 1271
1272=item gconvert replacement
1273
1274Maybe include a replacement function that doesn't lose data in rare
1275cases of coercion between string and numerical values.
1276
aa689395 1277=item Improve makedepend
1278
1279The current makedepend process is clunky and annoyingly slow, but it
1280works for most folks. Alas, it assumes that there is a filename
1281$firstmakefile that the B<make> command will try to use before it uses
1282F<Makefile>. Such may not be the case for all B<make> commands,
1283particularly those on non-Unix systems.
1284
1285Probably some variant of the BSD F<.depend> file will be useful.
1286We ought to check how other packages do this, if they do it at all.
1287We could probably pre-generate the dependencies (with the exception of
1288malloc.o, which could probably be determined at F<Makefile.SH>
1289extraction time.
1290
1291=item GNU Makefile standard targets
1292
1293GNU software generally has standardized Makefile targets. Unless we
1294have good reason to do otherwise, I see no reason not to support them.
1295
1296=item File locking
1297
1298Somehow, straighten out, document, and implement lockf(), flock(),
1299and/or fcntl() file locking. It's a mess.
1300
1301=back
1302
fb73857a 1303=head1 AUTHORS
aa689395 1304
fb73857a 1305Original author: Andy Dougherty doughera@lafcol.lafayette.edu .
1306Additions by Chip Salzenberg chip@perl.com and
1307Tim Bunce Tim.Bunce@ig.co.uk .
aa689395 1308
1309All opinions expressed herein are those of the authorZ<>(s).
1310
1311=head1 LAST MODIFIED
1312
20f245af 1313$Id: pumpkin.pod,v 1.22 1998/07/22 16:33:55 doughera Released $