This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
pumpkin-1.9.pod
[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
44=head1 How are Perl Releases Numbered?
45
7b5757d1
AD
46Perl version numbers are floating point numbers, such as 5.004.
47(Observations about the imprecision of floating point numbers for
48representing reality probably have more relevance than you might
49imagine :-) The major version number is 5 and the '004' is the
50patchlevel. (Questions such as whether or not '004' is really a minor
51version number can safely be ignored.:)
52
53The version number is available as the magic variable $],
aa689395 54and can be used in comparisons, e.g.
55
56 print "You've got an old perl\n" if $] < 5.002;
57
aa689395 58You can also require particular version (or later) with
59
60 use 5.002;
61
7b5757d1
AD
62At some point in the future, we may need to decide what to call the
63next big revision. In the .package file used by metaconfig to
64generate Configure, there are two variables that might be relevant:
65$baserev=5.0 and $package=perl5. At various times, I have suggested
66we might change them to $baserev=5.1 and $package=perl5.1 if want
67to signify a fairly major update. Or, we might want to jump to perl6.
68Let's worry about that problem when we get there.
69
aa689395 70=head2 Subversions
71
72In addition, there may be "developer" sub-versions available. These
73are not official releases. They may contain unstable experimental
74features, and are subject to rapid change. Such developer
75sub-versions are numbered with sub-version numbers. For example,
76version 5.004_04 is the 4'th developer version built on top of
775.004. It might include the _01, _02, and _03 changes, but it
78also might not. Sub-versions are allowed to be subversive.
79
80These sub-versions can also be used as floating point numbers, so
81you can do things such as
82
7b5757d1 83 print "You've got an unstable perl\n" if $] == 5.00303;
aa689395 84
85You can also require particular version (or later) with
86
7b5757d1 87 use 5.003_03; # the "_" is optional
aa689395 88
89Sub-versions produced by the members of perl5-porters are usually
90available on CPAN in the F<src/5.0/unsupported> directory.
91
7b5757d1
AD
92=head2 Maintenance and Development Subversions
93
94As an experiment, starting with version 5.004, subversions _01 through
95_49 will be reserved for bug-fix maintenance releases, and subversions
96_50 through _99 will be available for unstable development versions.
97
98The separate bug-fix track is being established to allow us an easy
99way to distribute important bug fixes without waiting for the
100developers to untangle all the other problems in the current
101developer's release.
102
103Watch for announcements of maintenance subversions in
104comp.lang.perl.announce.
105
aa689395 106=head2 Why such a complicated scheme?
107
108Two reasons, really. At least.
109
7b5757d1
AD
110First, we need some way to identify and release collections of patches
111that are known to have new features that need testing and exploration. The
aa689395 112subversion scheme does that nicely while fitting into the
113C<use 5.004;> mold.
114
115Second, since most of the folks who help maintain perl do so on a
116free-time voluntary basis, perl development does not proceed at a
117precise pace, though it always seems to be moving ahead quickly.
118We needed some way to pass around the "patch pumpkin" to allow
119different people chances to work on different aspects of the
120distribution without getting in each other's way. It wouldn't be
121constructive to have multiple people working on incompatible
122implementations of the same idea. Instead what was needed was
123some kind of "baton" or "token" to pass around so everyone knew
124whose turn was next.
125
126=head2 Why is it called the patch pumpkin?
127
128Chip Salzenberg gets credit for that, with a nod to his cow orker,
129David Croy. We had passed around various names (baton, token, hot
130potato) but none caught on. Then, Chip asked:
131
132[begin quote]
133
134 Who has the patch pumpkin?
135
136To explain: David Croy once told me once that at a previous job,
137there was one tape drive and multiple systems that used it for backups.
138But instead of some high-tech exclusion software, they used a low-tech
139method to prevent multiple simultaneous backups: a stuffed pumpkin.
140No one was allowed to make backups unless they had the "backup pumpkin".
141
142[end quote]
143
144The name has stuck.
145
146=head1 Philosophical Issues in Patching Perl
147
148There are no absolute rules, but there are some general guidelines I
149have tried to follow as I apply patches to the perl sources.
150(This section is still under construction.)
151
152=head2 Solve problems as generally as possible
153
7b5757d1
AD
154Never implement a specific restricted solution to a problem when you
155can solve the same problem in a more general, flexible way.
156
157For example, for dynamic loading to work on some SVR4 systems, we had
158to build a shared libperl.so library. In order to build "FAT" binaries
159on NeXT 4.0 systems, we had to build a special libperl library. Rather
160than continuing to build a contorted nest of special cases, I
161generalized the process of building libperl so that NeXT and SVR4 users
162could still get their work done, but others could build a shared
163libperl if they wanted to as well.
aa689395 164
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
187=head2 Allow for lots of testing
188
189We should never release a main version without testing it as a
190subversion first.
191
7b5757d1 192=head2 Automate generation of derivative files
aa689395 193
194The F<embed.h>, F<keywords.h>, F<opcode.h>, and F<perltoc.pod> files
195are all automatically generated by perl scripts. In general, don't
196patch these directly; patch the data files instead.
197
198F<Configure> and F<config_h.SH> are also automatically generated by
199B<metaconfig>. In general, you should patch the metaconfig units
200instead of patching these files directly. However, minor changes to
201F<Configure> may be made in between major sync-ups with the metaconfig
202units, which tends to be complicated operations.
203
204=head1 How to Make a Distribution
205
206There really ought to be a 'make dist' target, but there isn't.
207The 'dist' suite of tools also contains a number of tools that I haven't
208learned how to use yet. Some of them may make this all a bit easier.
209
210Here are the steps I go through to prepare a patch & distribution.
211
212Lots of it could doubtless be automated but isn't.
213
214=head2 Announce your intentions
215
216First, you should volunteer out loud to take the patch pumpkin. It's
217generally counter-productive to have multiple people working in secret
218on the same thing.
219
220At the same time, announce what you plan to do with the patch pumpkin,
221to allow folks a chance to object or suggest alternatives, or do it for
222you. Naturally, the patch pumpkin holder ought to incorporate various
223bug fixes and documentation improvements that are posted while he or
224she has the pumpkin, but there might also be larger issues at stake.
225
226One of the precepts of the subversion idea is that we shouldn't give
7b5757d1
AD
227the patch pumpkin to anyone unless we have some idea what he or she
228is going to do with it.
aa689395 229
230=head2 refresh pod/perltoc.pod
231
232Presumably, you have done a full C<make> in your working source
233directory. Before you C<make spotless> (if you do), and if you have
234changed any documentation in any module or pod file, change to the
235F<pod> directory and run C<make toc>.
236
237=head2 update patchlevel.h
238
239Don't be shy about using the subversion number, even for a relatively
240modest patch. We've never even come close to using all 99 subversions,
241and it's better to have a distinctive number for your patch. If you
242need feedback on your patch, go ahead and issue it and promise to
243incorporate that feedback quickly (e.g. within 1 week) and send out a
244second patch.
245
246=head2 run metaconfig
247
248If you need to make changes to Configure or config_h.SH, it may be best to
249change the appropriate metaconfig units instead, and regenerate Configure.
250
251 metaconfig -m
252
253will regenerate Configure and config_h.SH. More information on
254obtaining and running metaconfig is in the F<U/README> file that comes
255with Perl's metaconfig units. Perl's metaconfig units should be
256available the same place you found this file. On CPAN, look under my
257directory F<id/ANDYD/> for a file such as F<5.003_07-02.U.tar.gz>.
258That file should be unpacked in your main perl source directory. It
259contains the files needed to run B<metaconfig> to reproduce Perl's
7b5757d1
AD
260Configure script. (Those units are for 5.003_07. There have been
261changes since then; please contact me if you want more recent
262versions, and I will try to point you in the right direction.)
aa689395 263
264Alternatively, do consider if the F<*ish.h> files might be a better
265place for your changes.
266
267=head2 MANIFEST
268
269Make sure the MANIFEST is up-to-date. You can use dist's B<manicheck>
270program for this. You can also use
271
272 perl -MExtUtils::Manifest -e fullcheck
273
274to do half the job. This will make sure everything listed in MANIFEST
275is included in the distribution. dist's B<manicheck> command will
276also list extra files in the directory that are not listed in
277MANIFEST.
278
279The MANIFEST is normally sorted, with one exception. Perl includes
280both a F<Configure> script and a F<configure> script. The
281F<configure> script is a front-end to the main F<Configure>, but
282is there to aid folks who use autoconf-generated F<configure> files
283for other software. The problem is that F<Configure> and F<configure>
284are the same on case-insensitive file systems, so I deliberately put
285F<configure> first in the MANIFEST so that the extraction of
286F<Configure> will overwrite F<configure> and leave you with the
287correct script. (The F<configure> script must also have write
288permission for this to work, so it's the only file in the distribution
289I normally have with write permission.)
290
291If you are using metaconfig to regenerate Configure, then you should note
292that metaconfig actually uses MANIFEST.new, so you want to be sure
293MANIFEST.new is up-to-date too. I haven't found the MANIFEST/MANIFEST.new
294distinction particularly useful, but that's probably because I still haven't
295learned how to use the full suite of tools in the dist distribution.
296
297=head2 Check permissions
298
299All the tests in the t/ directory ought to be executable. The
300main makefile used to do a 'chmod t/*/*.t', but that resulted in
301a self-modifying distribution--something some users would strongly
302prefer to avoid. Probably, the F<t/TEST> script should check for this
303and do the chmod if needed, but it doesn't currently.
304
305In all, the following files should probably be executable:
306
307 Configure
308 configpm
309 configure
310 embed.pl
311 installperl
312 installman
313 keywords.pl
314 lib/splain
315 myconfig
316 opcode.pl
317 perly.fixer
318 t/TEST
319 t/*/*.t
320 *.SH
321 vms/ext/Stdio/test.pl
322 vms/ext/filespec.t
323 vms/fndvers.com
324 x2p/*.SH
325
326Other things ought to be readable, at least :-).
327
328Probably, the permissions for the files could be encoded in MANIFEST
329somehow, but I'm reluctant to change MANIFEST itself because that
330could break old scripts that use MANIFEST.
331
332I seem to recall that some SVR3 systems kept some sort of file that listed
333permissions for system files; something like that might be appropriate.
334
335=head2 Run Configure
336
337This will build a config.sh and config.h. You can skip this if you haven't
338changed Configure or config_h.SH at all.
339
340=head2 Update config_H
341
342The config_H file is provided to help those folks who can't run Configure.
343It is important to keep it up-to-date. If you have changed config_h.SH,
344those changes must be reflected in config_H as well. (The name config_H was
345chosen to distinguish the file from config.h even on case-insensitive file
346systems.) Simply edit the existing config_H file; keep the first few
347explanatory lines and then copy your new config.h below.
348
349It may also be necessary to update vms/config.vms and
350plan9/config.plan9, though you should be quite careful in doing so if
351you are not familiar with those systems. You might want to issue your
352patch with a promise to quickly issue a follow-up that handles those
353directories.
354
355=head2 make run_byacc
356
357If you have byacc-1.8.2 (available from CPAN), and if there have been
358changes to F<perly.y>, you can regenerate the F<perly.c> file. The
359run_byacc makefile target does this by running byacc and then applying
360some patches so that byacc dynamically allocates space, rather than
361having fixed limits. This patch is handled by the F<perly.fixer>
362script. Depending on the nature of the changes to F<perly.y>, you may
363or may not have to hand-edit the patch to apply correctly. If you do,
364you should include the edited patch in the new distribution. If you
365have byacc-1.9, the patch won't apply cleanly. Changes to the printf
366output statements mean the patch won't apply cleanly. Long ago I
367started to fix F<perly.fixer> to detect this, but I never completed the
368task.
369
370Some additional notes from Larry on this:
371
372Don't forget to regenerate perly.c.diff.
373
7b5757d1 374 byacc -d perly.y
aa689395 375 mv y.tab.c perly.c
376 patch perly.c <perly.c.diff
377 # manually apply any failed hunks
378 diff -c2 perly.c.orig perly.c >perly.c.diff
379
380One chunk of lines that often fails begins with
381
382 #line 29 "perly.y"
383
384and ends one line before
385
386 #define YYERRCODE 256
387
388This only happens when you add or remove a token type. I suppose this
389could be automated, but it doesn't happen very often nowadays.
390
391Larry
392
393=head2 make regen_headers
394
395The F<embed.h>, F<keywords.h>, and F<opcode.h> files are all automatically
396generated by perl scripts. Since the user isn't guaranteed to have a
397working perl, we can't require the user to generate them. Hence you have
398to, if you're making a distribution.
399
400I used to include rules like the following in the makefile:
401
402 # The following three header files are generated automatically
403 # The correct versions should be already supplied with the perl kit,
404 # in case you don't have perl or 'sh' available.
405 # The - is to ignore error return codes in case you have the source
406 # installed read-only or you don't have perl yet.
407 keywords.h: keywords.pl
408 @echo "Don't worry if this fails."
409 - perl keywords.pl
410
411
7b5757d1 412However, I got B<lots> of mail consisting of people worrying because the
aa689395 413command failed. I eventually decided that I would save myself time
414and effort by manually running C<make regen_headers> myself rather
415than answering all the questions and complaints about the failing
416command.
417
418=head2 global.sym and interp.sym
419
420Make sure these files are up-to-date. Read the comments in these
421files and in perl_exp.SH to see what to do.
422
423=head2 Binary compatibility
424
425If you do change F<global.sym> or F<interp.sym>, think carefully about
426what you are doing. To the extent reasonable, we'd like to maintain
427souce and binary compatibility with older releases of perl. That way,
428extensions built under one version of perl will continue to work with
429new versions of perl.
430
431Of course, some incompatible changes may well be necessary. I'm just
432suggesting that we not make any such changes without thinking carefully
433about them first. If possible, we should provide
434backwards-compatibility stubs. There's a lot of XS code out there.
435Let's not force people to keep changing it.
436
437=head2 Changes
438
439Be sure to update the F<Changes> file. Try to include both an overall
440summary as well as detailed descriptions of the changes. Your
441audience will include bother developers and users, so describe
442user-visible changes (if any) in terms they will understand, not in
443code like "initialize foo variable in bar function".
444
445There are differing opinions on whether the detailed descriptions
446ought to go in the Changes file or whether they ought to be available
447separately in the patch file (or both). There is no disagreement that
448detailed descriptions ought to be easily available somewhere.
449
450=head2 OS/2-specific updates
451
452In the os2 directory is F<diff.configure>, a set of OS/2-specific
453diffs against B<Configure>. If you make changes to Configure, you may
454want to consider regenerating this diff file to save trouble for the
455OS/2 maintainer.
456
7b5757d1
AD
457You can also consider the OS/2 diffs as reminders of portability
458things that need to be fixed in Configure.
459
aa689395 460=head2 VMS-specific updates
461
462If you have changed F<perly.y>, then you may want to update
463F<vms/perly_{h,c}.vms> by running C<perl vms/vms_yfix.pl>.
464
465The Perl version number appears in several places under F<vms>.
466It is courteous to update these versions. For example, if you are
467making 5.004_42, replace "5.00441" with "5.00442".
468
469=head2 Making the new distribution
470
471Suppose, for example, that you want to make version 5.004_08. Then you can
472do something like the following
473
474 mkdir ../perl5.004_08
475 awk '{print $1}' MANIFEST | cpio -pdm ../perl5.004_08
476 cd ../
477 tar cf perl5.004_08.tar perl5.004_08
478 gzip --best perl5.004_08.tar
479
480=head2 Making a new patch
481
482I find the F<makepatch> utility quite handy for making patches.
483You can obtain it from any CPAN archive under
7b5757d1 484http://www.perl.com/CPAN/authors/Johan_Vromans/ . The only
aa689395 485difference between my version and the standard one is that I have mine
486do a
487
488 # Print a reassuring "End of Patch" note so people won't
489 # wonder if their mailer truncated patches.
490 print "\n\nEnd of Patch.\n";
491
492at the end. That's because I used to get questions from people asking if
493their mail was truncated.
494
495Here's how I generate a new patch. I'll use the hypothetical
4965.004_07 to 5.004_08 patch as an example.
497
498 # unpack perl5.004_07/
499 gzip -d -c perl5.004_07.tar.gz | tar -xof -
500 # unpack perl5.004_08/
501 gzip -d -c perl5.004_08.tar.gz | tar -xof -
502 makepatch perl5.004_07 perl5.004_08 > perl5.004_08.pat
503
504Makepatch will automatically generate appropriate B<rm> commands to remove
505deleted files. Unfortunately, it will not correctly set permissions
506for newly created files, so you may have to do so manually. For example,
507patch 5.003_04 created a new test F<t/op/gv.t> which needs to be executable,
508so at the top of the patch, I inserted the following lines:
509
510 # Make a new test
511 touch t/op/gv.t
512 chmod +x t/opt/gv.t
513
514Now, of course, my patch is now wrong because makepatch didn't know I
515was going to do that command, and it patched against /dev/null.
516
517So, what I do is sort out all such shell commands that need to be in the
518patch (including possible mv-ing of files, if needed) and put that in the
519shell commands at the top of the patch. Next, I delete all the patch parts
520of perl5.004_08.pat, leaving just the shell commands. Then, I do the
521following:
522
7b5757d1
AD
523 cd perl5.004_07
524 sh ../perl5.004_08.pat
aa689395 525 cd ..
7b5757d1 526 makepatch perl5.004_07 perl5.004_08 >> perl5.004_08.pat
aa689395 527
528(Note the append to preserve my shell commands.)
529Now, my patch will line up with what the end users are going to do.
530
531=head2 Testing your patch
532
533It seems obvious, but be sure to test your patch. That is, verify that
534it produces exactly the same thing as your full distribution.
535
7b5757d1
AD
536 rm -rf perl5.004_07
537 gzip -d -c perl5.004_07.tar.gz | tar -xf -
538 cd perl5.004_07
539 sh ../perl5.004_08.pat
540 patch -p1 -N < ../perl5.004_08.pat
aa689395 541 cd ..
7b5757d1 542 gdiff -r perl5.004_07 perl5.004_08
aa689395 543
544where B<gdiff> is GNU diff. Other diff's may also do recursive checking.
545
546=head2 More testing
547
548Again, it's obvious, but you should test your new version as widely as you
549can. You can be sure you'll hear about it quickly if your version doesn't
550work on both ANSI and pre-ANSI compilers, and on common systems such as
551SunOS 4.1.[34], Solaris, and Linux.
552
553If your changes include conditional code, try to test the different
554branches as thoroughly as you can. For example, if your system
555supports dynamic loading, you can also test static loading with
556
557 sh Configure -Uusedl
558
559You can also hand-tweak your config.h to try out different #ifdef
560branches.
561
562=head1 Common Gotcha's
563
564=over 4
565
566=item #elif
567
568The '#elif' preprocessor directive is not understood on all systems.
569Specifically, I know that Pyramids don't understand it. Thus instead of the
570simple
571
572 #if defined(I_FOO)
573 # include <foo.h>
574 #elif defined(I_BAR)
575 # include <bar.h>
576 #else
577 # include <fubar.h>
578 #endif
579
580You have to do the more Byzantine
581
582 #if defined(I_FOO)
583 # include <foo.h>
584 #else
585 # if defined(I_BAR)
586 # include <bar.h>
587 # else
588 # include <fubar.h>
589 # endif
590 #endif
591
592Incidentally, whitespace between the leading '#' and the preprocessor
593command is not guaranteed, but is very portable and you may use it freely.
594I think it makes things a bit more readable, especially once things get
595rather deeply nested. I also think that things should almost never get
596too deeply nested, so it ought to be a moot point :-)
597
598=item Probably Prefer POSIX
599
600It's often the case that you'll need to choose whether to do
601something the BSD-ish way or the POSIX-ish way. It's usually not
602a big problem when the two systems use different names for similar
603functions, such as memcmp() and bcmp(). The perl.h header file
604handles these by appropriate #defines, selecting the POSIX mem*()
605functions if available, but falling back on the b*() functions, if
606need be.
607
608More serious is the case where some brilliant person decided to
609use the same function name but give it a different meaning or
610calling sequence :-). getpgrp() and setpgrp() come to mind.
611These are a real problem on systems that aim for conformance to
612one standard (e.g. POSIX), but still try to support the other way
613of doing things (e.g. BSD). My general advice (still not really
614implemented in the source) is to do something like the following.
615Suppose there are two alternative versions, fooPOSIX() and
616fooBSD().
617
618 #ifdef HAS_FOOPOSIX
619 /* use fooPOSIX(); */
620 #else
621 # ifdef HAS_FOOBSD
622 /* try to emulate fooPOSIX() with fooBSD();
623 perhaps with the following: */
624 # define fooPOSIX fooBSD
625 # else
626 # /* Uh, oh. We have to supply our own. */
627 # define fooPOSIX Perl_fooPOSIX
628 # endif
629 #endif
630
631=item Think positively
632
633If you need to add an #ifdef test, it is usually easier to follow if you
634think positively, e.g.
635
636 #ifdef HAS_NEATO_FEATURE
637 /* use neato feature */
638 #else
639 /* use some fallback mechanism */
640 #endif
641
642rather than the more impenetrable
643
644 #ifndef MISSING_NEATO_FEATURE
645 /* Not missing it, so we must have it, so use it */
646 #else
647 /* Are missing it, so fall back on something else. */
648 #endif
649
650Of course for this toy example, there's not much difference. But when
651the #ifdef's start spanning a couple of screen fulls, and the #else's
652are marked something like
653
654 #else /* !MISSING_NEATO_FEATURE */
655
656I find it easy to get lost.
657
658=item Providing Missing Functions -- Problem
659
660Not all systems have all the neat functions you might want or need, so
661you might decide to be helpful and provide an emulation. This is
662sound in theory and very kind of you, but please be careful about what
663you name the function. Let me use the C<pause()> function as an
664illustration.
665
666Perl5.003 has the following in F<perl.h>
667
668 #ifndef HAS_PAUSE
669 #define pause() sleep((32767<<16)+32767)
670 #endif
671
672Configure sets HAS_PAUSE if the system has the pause() function, so
673this #define only kicks in if the pause() function is missing.
674Nice idea, right?
675
676Unfortunately, some systems apparently have a prototype for pause()
677in F<unistd.h>, but don't actually have the function in the library.
678(Or maybe they do have it in a library we're not using.)
679
680Thus, the compiler sees something like
681
682 extern int pause(void);
683 /* . . . */
684 #define pause() sleep((32767<<16)+32767)
685
686and dies with an error message. (Some compilers don't mind this;
687others apparently do.)
688
689To work around this, 5.003_03 and later have the following in perl.h:
690
691 /* Some unistd.h's give a prototype for pause() even though
692 HAS_PAUSE ends up undefined. This causes the #define
693 below to be rejected by the compiler. Sigh.
694 */
695 #ifdef HAS_PAUSE
696 # define Pause pause
697 #else
698 # define Pause() sleep((32767<<16)+32767)
699 #endif
700
701This works.
702
703The curious reader may wonder why I didn't do the following in
704F<util.c> instead:
705
706 #ifndef HAS_PAUSE
707 void pause()
708 {
709 sleep((32767<<16)+32767);
710 }
711 #endif
712
713That is, since the function is missing, just provide it.
714Then things would probably be been alright, it would seem.
715
716Well, almost. It could be made to work. The problem arises from the
717conflicting needs of dynamic loading and namespace protection.
718
719For dynamic loading to work on AIX (and VMS) we need to provide a list
720of symbols to be exported. This is done by the script F<perl_exp.SH>,
721which reads F<global.sym> and F<interp.sym>. Thus, the C<pause>
722symbol would have to be added to F<global.sym> So far, so good.
723
724On the other hand, one of the goals of Perl5 is to make it easy to
725either extend or embed perl and link it with other libraries. This
726means we have to be careful to keep the visible namespace "clean".
727That is, we don't want perl's global variables to conflict with
728those in the other application library. Although this work is still
729in progress, the way it is currently done is via the F<embed.h> file.
730This file is built from the F<global.sym> and F<interp.sym> files,
731since those files already list the globally visible symbols. If we
732had added C<pause> to global.sym, then F<embed.h> would contain the
733line
734
735 #define pause Perl_pause
736
737and calls to C<pause> in the perl sources would now point to
738C<Perl_pause>. Now, when B<ld> is run to build the F<perl> executable,
739it will go looking for C<perl_pause>, which probably won't exist in any
740of the standard libraries. Thus the build of perl will fail.
741
742Those systems where C<HAS_PAUSE> is not defined would be ok, however,
743since they would get a C<Perl_pause> function in util.c. The rest of
744the world would be in trouble.
745
746And yes, this scenario has happened. On SCO, the function C<chsize>
747is available. (I think it's in F<-lx>, the Xenix compatibility
748library.) Since the perl4 days (and possibly before), Perl has
749included a C<chsize> function that gets called something akin to
750
751 #ifndef HAS_CHSIZE
752 I32 chsize(fd, length)
753 /* . . . */
754 #endif
755
756When 5.003 added
757
758 #define chsize Perl_chsize
759
760to F<embed.h>, the compile started failing on SCO systems.
761
762The "fix" is to give the function a different name. The one
763implemented in 5.003_05 isn't optimal, but here's what was done:
764
765 #ifdef HAS_CHSIZE
766 # ifdef my_chsize /* Probably #defined to Perl_my_chsize in embed.h */
767 # undef my_chsize
768 # endif
769 # define my_chsize chsize
770 #endif
771
772My explanatory comment in patch 5.003_05 said:
773
774 Undef and then re-define my_chsize from Perl_my_chsize to
775 just plain chsize if this system HAS_CHSIZE. This probably only
776 applies to SCO. This shows the perils of having internal
777 functions with the same name as external library functions :-).
778
779Now, we can safely put C<my_chsize> in F<global.sym>, export it, and
780hide it with F<embed.h>.
781
782To be consistent with what I did for C<pause>, I probably should have
783called the new function C<Chsize>, rather than C<my_chsize>.
784However, the perl sources are quite inconsistent on this (Consider
785New, Mymalloc, and Myremalloc, to name just a few.)
786
787There is a problem with this fix, however, in that C<Perl_chsize>
788was available as a F<libperl.a> library function in 5.003, but it
789isn't available any more (as of 5.003_07). This means that we've
790broken binary compatibility. This is not good.
791
792=item Providing missing functions -- some ideas
793
794We currently don't have a standard way of handling such missing
795function names. Right now, I'm effectively thinking aloud about a
796solution. Some day, I'll try to formally propose a solution.
797
798Part of the problem is that we want to have some functions listed as
799exported but not have their names mangled by embed.h or possibly
800conflict with names in standard system headers. We actually already
801have such a list at the end of F<perl_exp.SH> (though that list is
802out-of-date):
803
804 # extra globals not included above.
805 cat <<END >> perl.exp
806 perl_init_ext
807 perl_init_fold
808 perl_init_i18nl14n
809 perl_alloc
810 perl_construct
811 perl_destruct
812 perl_free
813 perl_parse
814 perl_run
815 perl_get_sv
816 perl_get_av
817 perl_get_hv
818 perl_get_cv
819 perl_call_argv
820 perl_call_pv
821 perl_call_method
822 perl_call_sv
823 perl_requirepv
824 safecalloc
825 safemalloc
826 saferealloc
827 safefree
828
829This still needs much thought, but I'm inclined to think that one
830possible solution is to prefix all such functions with C<perl_> in the
831source and list them along with the other C<perl_*> functions in
832F<perl_exp.SH>.
833
834Thus, for C<chsize>, we'd do something like the following:
835
836 /* in perl.h */
837 #ifdef HAS_CHSIZE
838 # define perl_chsize chsize
839 #endif
840
841then in some file (e.g. F<util.c> or F<doio.c>) do
842
843 #ifndef HAS_CHSIZE
844 I32 perl_chsize(fd, length)
845 /* implement the function here . . . */
846 #endif
847
848Alternatively, we could just always use C<chsize> everywhere and move
849C<chsize> from F<global.sym> to the end of F<perl_exp.SH>. That would
850probably be fine as long as our C<chsize> function agreed with all the
851C<chsize> function prototypes in the various systems we'll be using.
852As long as the prototypes in actual use don't vary that much, this is
853probably a good alternative. (As a counter-example, note how Configure
854and perl have to go through hoops to find and use get Malloc_t and
855Free_t for C<malloc> and C<free>.)
856
857At the moment, this latter option is what I tend to prefer.
858
859=item All the world's a VAX
860
861Sorry, showing my age:-). Still, all the world is not BSD 4.[34],
862SVR4, or POSIX. Be aware that SVR3-derived systems are still quite
863common (do you have any idea how many systems run SCO?) If you don't
864have a bunch of v7 manuals handy, the metaconfig units (by default
865installed in F</usr/local/lib/dist/U>) are a good resource to look at
866for portability.
867
868=back
869
870=head1 Miscellaneous Topics
871
872=head2 Autoconf
873
874Why does perl use a metaconfig-generated Configure script instead of an
875autoconf-generated configure script?
876
877Metaconfig and autoconf are two tools with very similar purposes.
878Metaconfig is actually the older of the two, and was originally written
879by Larry Wall, while autoconf is probably now used in a wider variety of
880packages. The autoconf info file discusses the history of autoconf and
881how it came to be. The curious reader is referred there for further
882information.
883
884Overall, both tools are quite good, I think, and the choice of which one
885to use could be argued either way. In March, 1994, when I was just
886starting to work on Configure support for Perl5, I considered both
887autoconf and metaconfig, and eventually decided to use metaconfig for the
888following reasons:
889
890=over 4
891
892=item Compatibility with Perl4
893
894Perl4 used metaconfig, so many of the #ifdef's were already set up for
895metaconfig. Of course metaconfig had evolved some since Perl4's days,
896but not so much that it posed any serious problems.
897
898=item Metaconfig worked for me
899
900My system at the time was Interactive 2.2, a SVR3.2/386 derivative that
901also had some POSIX support. Metaconfig-generated Configure scripts
902worked fine for me on that system. On the other hand, autoconf-generated
903scripts usually didn't. (They did come quite close, though, in some
904cases.) At the time, I actually fetched a large number of GNU packages
905and checked. Not a single one configured and compiled correctly
906out-of-the-box with the system's cc compiler.
907
908=item Configure can be interactive
909
910With both autoconf and metaconfig, if the script works, everything is
911fine. However, one of my main problems with autoconf-generated scripts
912was that if it guessed wrong about something, it could be B<very> hard to
913go back and fix it. For example, autoconf always insisted on passing the
914-Xp flag to cc (to turn on POSIX behavior), even when that wasn't what I
915wanted or needed for that package. There was no way short of editing the
916configure script to turn this off. You couldn't just edit the resulting
917Makefile at the end because the -Xp flag influenced a number of other
918configure tests.
919
920Metaconfig's Configure scripts, on the other hand, can be interactive.
921Thus if Configure is guessing things incorrectly, you can go back and fix
922them. This isn't as important now as it was when we were actively
923developing Configure support for new features such as dynamic loading,
924but it's still useful occasionally.
925
926=item GPL
927
928At the time, autoconf-generated scripts were covered under the GNU Public
929License, and hence weren't suitable for inclusion with Perl, which has a
930different licensing policy. (Autoconf's licensing has since changed.)
931
932=item Modularity
933
934Metaconfig builds up Configure from a collection of discrete pieces
935called "units". You can override the standard behavior by supplying your
936own unit. With autoconf, you have to patch the standard files instead.
937I find the metaconfig "unit" method easier to work with. Others
938may find metaconfig's units clumsy to work with.
939
940=back
941
942=head2 @INC search order
943
944By default, the list of perl library directories in @INC is the
945following:
946
947 $archlib
948 $privlib
949 $sitearch
950 $sitelib
951
952Specifically, on my Solaris/x86 system, I run
953B<sh Configure -Dprefix=/opt/perl> and I have the following
954directories:
955
956 /opt/perl/lib/i86pc-solaris/5.00307
957 /opt/perl/lib
958 /opt/perl/lib/site_perl/i86pc-solaris
959 /opt/perl/lib/site_perl
960
961That is, perl's directories come first, followed by the site-specific
962directories.
963
964The site libraries come second to support the usage of extensions
965across perl versions. Read the relevant section in F<INSTALL> for
966more information. If we ever make $sitearch version-specific, this
967topic could be revisited.
968
969=head2 Why isn't there a directory to override Perl's library?
970
971Mainly because no one's gotten around to making one. Note that
972"making one" involves changing perl.c, Configure, config_h.SH (and
973associated files, see above), and I<documenting> it all in the
974INSTALL file.
975
976Apparently, most folks who want to override one of the standard library
977files simply do it by overwriting the standard library files.
978
979=head2 APPLLIB
980
981In the perl.c sources, you'll find an undocumented APPLLIB_EXP
982variable, sort of like PRIVLIB_EXP and ARCHLIB_EXP (which are
983documented in config_h.SH). Here's what APPLLIB_EXP is for, from
984a mail message from Larry:
985
986 The main intent of APPLLIB_EXP is for folks who want to send out a
987 version of Perl embedded in their product. They would set the symbol
988 to be the name of the library containing the files needed to run or to
989 support their particular application. This works at the "override"
990 level to make sure they get their own versions of any library code that
991 they absolutely must have configuration control over.
992
993 As such, I don't see any conflict with a sysadmin using it for a
994 override-ish sort of thing, when installing a generic Perl. It should
995 probably have been named something to do with overriding though. Since
996 it's undocumented we could still change it... :-)
997
998Given that it's already there, you can use it to override
999distribution modules. If you do
1000
1001 sh Configure -Dccflags='-DAPPLLIB_EXP=/my/override'
1002
1003then perl.c will put /my/override ahead of ARCHLIB and PRIVLIB.
1004
1005=head1 Upload Your Work to CPAN
1006
1007You can upload your work to CPAN if you have a CPAN id. Check out
1008http://www.perl.com/CPAN/modules/04pause.html for information on
1009_PAUSE_, the Perl Author's Upload Server.
1010
1011I typically upload both the patch file, e.g. F<perl5.004_08.pat.gz>
1012and the full tar file, e.g. F<perl5.004_08.tar.gz>.
1013
1014If you want your patch to appear in the F<src/5.0/unsupported>
1015directory on CPAN, send e-mail to the CPAN master librarian. (Check
7b5757d1 1016out http://www.perl.com/CPAN/CPAN.html ).
aa689395 1017
1018=head1 Help Save the World
1019
1020You should definitely announce your patch on the perl5-porters list.
1021You should also consider announcing your patch on
1022comp.lang.perl.announce, though you should make it quite clear that a
1023subversion is not a production release, and be prepared to deal with
1024people who will not read your disclaimer.
1025
1026=head1 Todo
1027
1028Here, in no particular order, are some Configure and build-related
1029items that merit consideration. This list isn't exhaustive, it's just
1030what I came up with off the top of my head.
1031
1032=head2 Good ideas waiting for round tuits
1033
1034=over 4
1035
1036=item installprefix
1037
1038I think we ought to support
1039
1040 Configure -Dinstallprefix=/blah/blah
1041
1042Currently, we support B<-Dprefix=/blah/blah>, but the changing the install
1043location has to be handled by something like the F<config.over> trick
1044described in F<INSTALL>. AFS users also are treated specially.
1045We should probably duplicate the metaconfig prefix stuff for an
1046install prefix.
1047
1048=item Configure -Dsrcdir=/blah/blah
1049
1050We should be able to emulate B<configure --srcdir>. Tom Tromey
1051tromey@creche.cygnus.com has submitted some patches to
1052the dist-users mailing list along these lines. Eventually, they ought
1053to get folded back into the main distribution.
1054
1055=item Hint file fixes
1056
1057Various hint files work around Configure problems. We ought to fix
1058Configure so that most of them aren't needed.
1059
1060=item Hint file information
1061
1062Some of the hint file information (particularly dynamic loading stuff)
1063ought to be fed back into the main metaconfig distribution.
1064
1065=back
1066
1067=head2 Probably good ideas waiting for round tuits
1068
1069=over 4
1070
1071=item GNU configure --options
1072
1073I've received sensible suggestions for --exec_prefix and other
1074GNU configure --options. It's not always obvious exactly what is
1075intended, but this merits investigation.
1076
1077=item make clean
1078
1079Currently, B<make clean> isn't all that useful, though
1080B<make realclean> and B<make distclean> are. This needs a bit of
1081thought and documentation before it gets cleaned up.
1082
1083=item Try gcc if cc fails
1084
1085Currently, we just give up.
1086
1087=item bypassing safe*alloc wrappers
1088
1089On some systems, it may be safe to call the system malloc directly
1090without going through the util.c safe* layers. (Such systems would
1091accept free(0), for example.) This might be a time-saver for systems
1092that already have a good malloc. (Recent Linux libc's apparently have
1093a nice malloc that is well-tuned for the system.)
1094
1095=back
1096
1097=head2 Vague possibilities
1098
1099=over 4
1100
1101=item Win95, WinNT, and Win32 support
1102
1103We need to get something into the distribution for 32-bit Windows.
1104I'm tired of all the private e-mail questions I get, and I'm saddened
1105that so many folks keep trying to reinvent the same wheel.
1106
1107=item MacPerl
1108
1109Get some of the Macintosh stuff folded back into the main
1110distribution.
1111
1112=item gconvert replacement
1113
1114Maybe include a replacement function that doesn't lose data in rare
1115cases of coercion between string and numerical values.
1116
1117=item long long
1118
1119Can we support C<long long> on systems where C<long long> is larger
1120than what we've been using for C<IV>? What if you can't C<sprintf>
1121a C<long long>?
1122
1123=item Improve makedepend
1124
1125The current makedepend process is clunky and annoyingly slow, but it
1126works for most folks. Alas, it assumes that there is a filename
1127$firstmakefile that the B<make> command will try to use before it uses
1128F<Makefile>. Such may not be the case for all B<make> commands,
1129particularly those on non-Unix systems.
1130
1131Probably some variant of the BSD F<.depend> file will be useful.
1132We ought to check how other packages do this, if they do it at all.
1133We could probably pre-generate the dependencies (with the exception of
1134malloc.o, which could probably be determined at F<Makefile.SH>
1135extraction time.
1136
1137=item GNU Makefile standard targets
1138
1139GNU software generally has standardized Makefile targets. Unless we
1140have good reason to do otherwise, I see no reason not to support them.
1141
1142=item File locking
1143
1144Somehow, straighten out, document, and implement lockf(), flock(),
1145and/or fcntl() file locking. It's a mess.
1146
1147=back
1148
1149=head1 AUTHOR
1150
1151Andy Dougherty <doughera@lafcol.lafayette.edu>.
1152
1153Additions by Chip Salzenberg <chip@atlantic.net>.
1154
1155All opinions expressed herein are those of the authorZ<>(s).
1156
1157=head1 LAST MODIFIED
1158
7b5757d1 1159$Id: pumpkin.pod,v 1.9 1997/02/24 20:37:43 doughera Released $