Robin Barker [Tue, 7 May 2002 16:13:34 +0000]
avoid compiler casting warnings
Message-Id: <
200205071513.QAA05846@tempest.npl.co.uk>
p4raw-id: //depot/perl@16445
Michael A Chase [Mon, 6 May 2002 13:58:48 +0000]
[ID
20020506.009] File::Spec::Unix::canonpath under cygwin
Message-Id: <E174pZk-0008IP-00@smtp6.mindspring.com>
(plus add what POSIX says)
p4raw-id: //depot/perl@16444
John Peacock [Mon, 6 May 2002 10:14:21 +0000]
Re: lib/User/pwent.t on Cygwin
Message-ID: <
3CD68FBD.8020502@rowman.com>
p4raw-id: //depot/perl@16443
Radu Greab [Tue, 7 May 2002 11:37:03 +0000]
[PATCH] Storable (Re: perl@16433)
Date: Tue, 7 May 2002 11:37:03 +0300
Message-ID: <15575.37423.446700.9930@ix.netsoft.ro>
Subject: Re: [PATCH] Storable (Re: perl@16433)
From: Radu Greab <radu@netsoft.ro>
Date: Tue, 7 May 2002 12:49:24 +0300
Message-ID: <15575.41764.744956.7193@ix.netsoft.ro>
Plug the Storable memory leaks.
p4raw-id: //depot/perl@16442
Jarkko Hietaniemi [Tue, 7 May 2002 12:48:30 +0000]
Upgrade to CPAN 1.61.
p4raw-id: //depot/perl@16441
Jarkko Hietaniemi [Tue, 7 May 2002 02:49:47 +0000]
Document the bad behaviour of for(1..5){$_++}
(it works but it should fail)
p4raw-id: //depot/perl@16439
Jarkko Hietaniemi [Tue, 7 May 2002 00:14:12 +0000]
README.ko update from Jungshik Shin.
p4raw-id: //depot/perl@16438
Nicholas Clark [Tue, 7 May 2002 00:21:46 +0000]
Re: perl@16433
Message-ID: <
20020506232146.GF4698@Bagpuss.unfortu.net>
p4raw-id: //depot/perl@16437
Vadim Konovalov [Mon, 6 May 2002 23:54:24 +0000]
two and a half steps further for WinCE cross-build
Message-ID: <
022001c1f537$d2dd4f10$cb5cc3d9@vad>
p4raw-id: //depot/perl@16436
Jarkko Hietaniemi [Mon, 6 May 2002 15:30:29 +0000]
Detypo.
p4raw-id: //depot/perl@16435
Dave Mitchell [Mon, 6 May 2002 17:17:00 +0000]
Re: [proposed PATCH] correctly unlocalise exists on tied/%ENV
Message-ID: <
20020506171700.A256@fdgroup.com>
p4raw-id: //depot/perl@16434
Jarkko Hietaniemi [Mon, 6 May 2002 13:38:56 +0000]
Update Changes.
p4raw-id: //depot/perl@16433
Jarkko Hietaniemi [Mon, 6 May 2002 13:29:22 +0000]
Add a test for #16431, and document Dave's campaign
against localised hash element bugs.
p4raw-id: //depot/perl@16432
Dave Mitchell [Sat, 4 May 2002 00:34:18 +0000]
[proposed PATCH] correctly unlocalise exists on tied/%ENV
Message-ID: <
20020504003418.F22026@fdgroup.com>
p4raw-id: //depot/perl@16431
Jarkko Hietaniemi [Mon, 6 May 2002 12:39:31 +0000]
Regen uconfig.h.
p4raw-id: //depot/perl@16430
Jarkko Hietaniemi [Mon, 6 May 2002 12:18:23 +0000]
Unfortunately AmigaOS has been broken since 11423
and is unlikely to get fixed in time for 5.8.0.
p4raw-id: //depot/perl@16429
Jarkko Hietaniemi [Mon, 6 May 2002 12:12:59 +0000]
Ultimate discouragement for the old tools, from Andreas.
p4raw-id: //depot/perl@16428
Jarkko Hietaniemi [Mon, 6 May 2002 12:01:36 +0000]
Upgrade to Digest::MD5 2.20.
p4raw-id: //depot/perl@16427
Dan Kogai [Mon, 6 May 2002 19:36:15 +0000]
[Encode] 1.70 released
Message-Id: <
17AB2CE6-60DD-11D6-9982-00039301D480@dan.co.jp>
p4raw-id: //depot/perl@16426
Gisle Aas [Sun, 5 May 2002 22:50:23 +0000]
PerlIO::Via reference
Message-Id: <lrd6w925b4.fsf_-_@caliper.activestate.com>
p4raw-id: //depot/perl@16425
Brendan O'Dea [Mon, 6 May 2002 14:04:48 +0000]
Term/Complete.pm: quote metachars
Message-Id: <
20020506040448.GA8241@londo.odea.dropbear.id.au>
p4raw-id: //depot/perl@16424
Gurusamy Sarathy [Mon, 6 May 2002 05:25:59 +0000]
p4genpatch: avoid stderr noise during adds
p4raw-id: //depot/perl@16423
Paul Green [Sun, 5 May 2002 23:11:00 +0000]
Fixup VOS builds of miniperl
Message-Id: <
200205060311.XAA12996@mailhub1.stratus.com>
p4raw-id: //depot/perl@16422
Radu Greab [Mon, 6 May 2002 04:23:10 +0000]
Re: [ID
20020504.006] $s=`command` fails if $/=\integer
Message-ID: <15573.56062.264379.981864@ix.netsoft.ro>
p4raw-id: //depot/perl@16421
Jarkko Hietaniemi [Sun, 5 May 2002 22:35:58 +0000]
Further comment tweak, to agree with perlxs.pod.
p4raw-id: //depot/perl@16420
Paul Green [Sun, 5 May 2002 17:15:00 +0000]
Update README.vos and release vos build macros
Message-Id: <
200205052114.RAA26329@mailhub2.stratus.com>
To: perl5-porters@perl.org
p4raw-id: //depot/perl@16419
Jarkko Hietaniemi [Sun, 5 May 2002 22:09:27 +0000]
Retract MM_NW5.pm part of #16371, at the request
of Michael Schwern.
p4raw-id: //depot/perl@16418
Jarkko Hietaniemi [Sun, 5 May 2002 17:57:04 +0000]
Comment tweaks.
p4raw-id: //depot/perl@16417
Gurusamy Sarathy [Sun, 5 May 2002 17:00:33 +0000]
a few more p4genpatch fixes:
don't assume previous revision exists, let p4 compute it from
the one-less-change number instead
the time part in timestamps were always 00:00:00; fix by
fetching metadata via "p4 fsync" rather than "p4 filelog"
quote all file names used in external commands to make them
them work when there are spaces in those file names
looks production worthy now :)
p4raw-id: //depot/perl@16416
Jarkko Hietaniemi [Sun, 5 May 2002 15:51:01 +0000]
The #16401 was far from enough, and mostly in a wrong direction.
The right thing to do would be to convert Storable into using
the new MY_CXT scheme (no static data, and memory management
using SVs).
p4raw-id: //depot/perl@16415
Gurusamy Sarathy [Sun, 5 May 2002 08:41:22 +0000]
p4genpatch diffs against depot files from //depot/perl when called
with -b //depot/maint-5.6/perl
p4raw-id: //depot/perl@16414
Gurusamy Sarathy [Sun, 5 May 2002 07:50:53 +0000]
p4genpatch: current directory may not be writable, so don't
create temp files there
p4raw-id: //depot/perl@16413
Gurusamy Sarathy [Sun, 5 May 2002 07:34:33 +0000]
p4genpatch now works on windows
p4raw-id: //depot/perl@16412
Gurusamy Sarathy [Sun, 5 May 2002 07:16:37 +0000]
p4genpatch: tweak from Andreas to strip leading portion of
path specified in -b (so patch can be applied with "patch -p1")
p4raw-id: //depot/perl@16411
Peter Prymmer [Fri, 3 May 2002 14:52:12 +0000]
add niscat (aka NIS 3, aka NIS +) to t/op/pwent
Message-ID: <O
F365AB827.72878468-ON85256BAE.0066F730@55.25.11>
p4raw-id: //depot/perl@16410
Jarkko Hietaniemi [Sun, 5 May 2002 01:53:24 +0000]
Upgrade to Net::Ping 2.17.
p4raw-id: //depot/perl@16409
Jarkko Hietaniemi [Sun, 5 May 2002 01:38:39 +0000]
Upgrade to Getopt::Long 2.31.
p4raw-id: //depot/perl@16408
Vadim Konovalov [Sat, 4 May 2002 03:49:11 +0000]
Re: Cross-compiling as of WinCE
Message-ID: <
00ca01c1f2fd$20a35fb0$cb5cc3d9@vad>
p4raw-id: //depot/perl@16407
Jarkko Hietaniemi [Sun, 5 May 2002 00:23:16 +0000]
The loclibpth is always set by Configure.
p4raw-id: //depot/perl@16406
SADAHIRO Tomoyuki [Sun, 5 May 2002 10:07:57 +0000]
Re: [Unicode::Collate] UCA Version number
Message-Id: <
20020505095942.3BEF.BQW10602@nifty.com>
p4raw-id: //depot/perl@16405
Dave Mitchell [Fri, 3 May 2002 21:17:00 +0000]
delete inaccurate comment
Message-ID: <
20020503211700.C22026@fdgroup.com>
p4raw-id: //depot/perl@16404
Jarkko Hietaniemi [Sat, 4 May 2002 19:02:56 +0000]
Solaris: show what 'that' is.
p4raw-id: //depot/perl@16403
Rafael Garcia-Suarez [Sat, 4 May 2002 18:25:25 +0000]
Syntax fix.
p4raw-id: //depot/perl@16402
Jarkko Hietaniemi [Sat, 4 May 2002 18:25:01 +0000]
(retracted by #16415) Storable: try to free what we allocated.
p4raw-id: //depot/perl@16401
Jarkko Hietaniemi [Sat, 4 May 2002 17:15:10 +0000]
UNICOS: division is really sloppy.
p4raw-id: //depot/perl@16400
Paul Green [Sat, 4 May 2002 14:01:00 +0000]
fix t/io/fs.t for VOS
Message-Id: <
200205041801.OAA13071@mailhub1.stratus.com>
p4raw-id: //depot/perl@16399
SADAHIRO Tomoyuki [Sat, 4 May 2002 16:07:07 +0000]
[Unicode::Collate] UCA Version number
Message-Id: <
20020504160547.D7E0.BQW10602@nifty.com>
p4raw-id: //depot/perl@16398
Michael G. Schwern [Sat, 4 May 2002 13:17:44 +0000]
"\"\""
Message-ID: <
20020504171743.GD640@blackrider>
p4raw-id: //depot/perl@16397
Jarkko Hietaniemi [Sat, 4 May 2002 16:32:45 +0000]
Add the EBCDIC test data to downgrade.t.
p4raw-id: //depot/perl@16396
Jarkko Hietaniemi [Sat, 4 May 2002 16:25:37 +0000]
EBCDIC tweak.
p4raw-id: //depot/perl@16395
Jarkko Hietaniemi [Sat, 4 May 2002 16:09:15 +0000]
Allow p4genpatch to use diff programs other than 'diff'.
p4raw-id: //depot/perl@16394
Jarkko Hietaniemi [Sat, 4 May 2002 16:03:56 +0000]
Add p4genpatch from Andreas; obsoletes both
p4d2p and p4desc.
p4raw-id: //depot/perl@16393
Jarkko Hietaniemi [Sat, 4 May 2002 15:58:53 +0000]
Upgrade to Encode 1.69.
p4raw-id: //depot/perl@16392
Dave Mitchell [Sat, 4 May 2002 16:09:26 +0000]
- remove binary compatibility claims
Message-ID: <
20020504160926.A24990@fdgroup.com>
p4raw-id: //depot/perl@16391
Nicholas Clark [Sat, 4 May 2002 12:38:58 +0000]
Re: [PATCH] Re: Data::Dumper tests with -Mutf8
Message-ID: <
20020504113857.GC317@Bagpuss.unfortu.net>
p4raw-id: //depot/perl@16390
Nicholas Clark [Sat, 4 May 2002 00:30:43 +0000]
Re: Data::Dumper tests with -Mutf8
Message-ID: <
20020503233042.GF294@Bagpuss.unfortu.net>
p4raw-id: //depot/perl@16389
Jarkko Hietaniemi [Sat, 4 May 2002 15:28:56 +0000]
Doc tweak from Elizabeth Mattijsen.
p4raw-id: //depot/perl@16388
Tatsuhiko Miyagawa [Sat, 4 May 2002 11:45:18 +0000]
channames.pm doc tweaks
Message-ID: <86it64d41t.wl@mail.edge.co.jp>
p4raw-id: //depot/perl@16387
Dave Mitchell [Fri, 3 May 2002 23:51:10 +0000]
Re: [PATCH scope.c] Re: local($tied->{foo}) leaks
Message-ID: <
20020503235110.E22026@fdgroup.com>
p4raw-id: //depot/perl@16386
Nicholas Clark [Fri, 3 May 2002 21:49:04 +0000]
Re: perl@16307
Message-ID: <
20020503204903.GC294@Bagpuss.unfortu.net>
canonical #3 was failing under -Mutf8.
p4raw-id: //depot/perl@16384
Dave Mitchell [Fri, 3 May 2002 21:12:10 +0000]
Re: local($tied->{foo}) leaks
Message-ID: <
20020503211210.B22026@fdgroup.com>
p4raw-id: //depot/perl@16383
Gerrit P. Haase [Fri, 3 May 2002 20:30:25 +0000]
perl -i clobbers files on Cygwin
From: "Gerrit P. Haase" <gerrit@familiehaase.de>
Message-ID: <
190109228332.20020503203025@familiehaase.de>
p4raw-id: //depot/perl@16382
Paul Green [Fri, 3 May 2002 14:51:00 +0000]
t/TestInit.pm - clean up comments
Message-ID: <
200205031850.OAA09104@mailhub2.stratus.com>
p4raw-id: //depot/perl@16381
Dan Kogai [Sat, 4 May 2002 01:46:35 +0000]
enc2xs template -- searches script more sensibly
Message-Id: <
544C1568-5EB5-11D6-8F1D-00039301D480@dan.co.jp>
p4raw-id: //depot/perl@16380
Jarkko Hietaniemi [Fri, 3 May 2002 15:43:47 +0000]
Forgotten test arrangement.
p4raw-id: //depot/perl@16379
Jarkko Hietaniemi [Fri, 3 May 2002 14:54:19 +0000]
Comedy of typing errors.
p4raw-id: //depot/perl@16378
Jarkko Hietaniemi [Fri, 3 May 2002 14:45:26 +0000]
Let's try the "no Net:: if no Socket" again.
p4raw-id: //depot/perl@16377
Jarkko Hietaniemi [Fri, 3 May 2002 14:11:42 +0000]
Update Changes.
p4raw-id: //depot/perl@16376
Jarkko Hietaniemi [Fri, 3 May 2002 14:09:38 +0000]
Oops.
p4raw-id: //depot/perl@16375
Jarkko Hietaniemi [Fri, 3 May 2002 13:58:08 +0000]
EBCDIC tweak.
p4raw-id: //depot/perl@16374
Jarkko Hietaniemi [Fri, 3 May 2002 13:51:11 +0000]
Update the EBCDIC MD5 checksums, and automate the process.
p4raw-id: //depot/perl@16373
Jarkko Hietaniemi [Fri, 3 May 2002 13:01:23 +0000]
The EBCDIC remapping of the low 256 bites again.
p4raw-id: //depot/perl@16372
Jarkko Hietaniemi [Fri, 3 May 2002 12:49:58 +0000]
NetWare updates from C Aditya.
p4raw-id: //depot/perl@16371
Jarkko Hietaniemi [Fri, 3 May 2002 12:36:54 +0000]
DJGPP updates from Laszlo Molnar.
p4raw-id: //depot/perl@16370
Jarkko Hietaniemi [Fri, 3 May 2002 12:34:01 +0000]
Upgrade to Encode 1.68.
p4raw-id: //depot/perl@16369
Jarkko Hietaniemi [Fri, 3 May 2002 12:31:52 +0000]
Integrate perlio:
[ 16367]
Several of non-default builds now seem to work reasonably well
English.t seems to fail on an errno test, and socketpair blathers
about something.
Basic fix is to stop PERL_IMPLICIT_SYS turning on USE_PERLIO by the
back door, and instead have perlsdio.h vector stdio via iperlsys.h
function tables (latter was done in earlier change).
Update comments in Makefile.mk
[ 16366]
Finish off 16350 for non-PERLIO build on linux,
non PERL_IMPLICIT_SYS parts of iperlsys.h had junk
for some slots which now perlsdio.h is targeting.
[ 16352]
setbuf / setvbuf are not PerlIO_ concepts
[ 16351]
perl_clone is a threads thing
[ 16350]
*** EXPERIMENTAL ***
Have perlsdio.h use the iperlsys.h aliases and see
if that helps non-PERLIO IMP_SYS on Win32.
(Miniperl okay on linux).
[ 16349]
Win32 builds and mostly works for non-USE_PERLIO non-USE_IMP_SYS case.
- move body of fdupopen() from perlhost.h to win32.h as win32_fdupopen()
- use it in perlio.c
[ 16347]
More layer syms
[ 16346]
Use PerlSIO_fdupopen() if not using PerlIO
[ 16345]
Do not build if not using layers
p4raw-link: @16367 on //depot/perlio:
2cbbe5a150b07f176547702112505e3d30ea082f
p4raw-link: @16366 on //depot/perlio:
079b94bc698b0934beaf07d168872e7ff53ba831
p4raw-link: @16352 on //depot/perlio:
22a46b6e190a186151dafce1a1c165a2532b0f1d
p4raw-link: @16351 on //depot/perlio:
0295eab83c15497fe4260ac4facad25c962a4293
p4raw-link: @16350 on //depot/perlio:
362d0d4483498690d9657a76e53755c85da3b796
p4raw-link: @16349 on //depot/perlio:
30753f56c8bbfb209552e904d9d76202413ebf24
p4raw-link: @16347 on //depot/perlio:
1e8a8f599308436d6e164543b54e022d0b1a9177
p4raw-link: @16346 on //depot/perlio:
0553478ef48e094f904ff993d3c57b7325174272
p4raw-link: @16345 on //depot/perlio:
57143e73be61ad4ed95e8d5fdc799945c48c8339
p4raw-id: //depot/perl@16368
Nick Ing-Simmons [Fri, 3 May 2002 07:25:13 +0000]
Several of non-default builds now seem to work reasonably well
English.t seems to fail on an errno test, and socketpair blathers
about something.
Basic fix is to stop PERL_IMPLICIT_SYS turning on USE_PERLIO by the
back door, and instead have perlsdio.h vector stdio via iperlsys.h
function tables (latter was done in earlier change).
Update comments in Makefile.mk
p4raw-id: //depot/perlio@16367
Nick Ing-Simmons [Fri, 3 May 2002 07:07:36 +0000]
Finish off 16350 for non-PERLIO build on linux,
non PERL_IMPLICIT_SYS parts of iperlsys.h had junk
for some slots which now perlsdio.h is targeting.
p4raw-id: //depot/perlio@16366
Jarkko Hietaniemi [Fri, 3 May 2002 03:52:08 +0000]
If no Socket should not try to test Net::
p4raw-id: //depot/perl@16365
Jarkko Hietaniemi [Fri, 3 May 2002 02:22:56 +0000]
Integrate macperl #16360 and #16363;
Make op/alarm.t pass
README.macos update
p4raw-id: //depot/perl@16364
p4raw-integrated: from //depot/macperl@16358 'copy in' README.macos
t/op/alarm.t (@16123..)
Jarkko Hietaniemi [Thu, 2 May 2002 22:35:31 +0000]
Forgot the most important part of #16357.
p4raw-id: //depot/perl@16358
SADAHIRO Tomoyuki [Fri, 3 May 2002 03:19:29 +0000]
Unicode::Collate 0.11 Released
Message-Id: <
20020503030904.35B3.BQW10602@nifty.com>
p4raw-id: //depot/perl@16357
Andy Dougherty [Thu, 2 May 2002 12:01:50 +0000]
Re: Perl 5.8-to-be NetBSD testing?
Message-ID: <Pine.SOL.4.10.
10205021156390.325-100000@maxwell.phys.lafayette.edu>
p4raw-id: //depot/perl@16356
Jarkko Hietaniemi [Thu, 2 May 2002 22:26:38 +0000]
Integrate #16353 from macperl;
"fix" for utf8_heavy.pl, lexical UTF8 var crashed
in test 92 of run/fresh_perl.t on MacOS
(as pudge rightfully points out, this is voodoo
programming at it best, the real bug is somewhere else,
now we just happened to shake the chicken the right way)
p4raw-id: //depot/perl@16355
p4raw-integrated: from //depot/macperl@16354 'merge in'
lib/utf8_heavy.pl (@16123..)
Dan Kogai [Thu, 2 May 2002 00:44:30 +0000]
Re: Encode, charnames and utf8heavy
Message-Id: <
539D985A-5D1A-11D6-BB19-00039301D480@dan.co.jp>
(plus a respective perlunicode tweak)
p4raw-id: //depot/perl@16354
Nick Ing-Simmons [Thu, 2 May 2002 17:48:31 +0000]
setbuf / setvbuf are not PerlIO_ concepts
p4raw-id: //depot/perlio@16352
Nick Ing-Simmons [Thu, 2 May 2002 17:12:21 +0000]
perl_clone is a threads thing
p4raw-id: //depot/perlio@16351
Nick Ing-Simmons [Thu, 2 May 2002 17:09:43 +0000]
*** EXPERIMENTAL ***
Have perlsdio.h use the iperlsys.h aliases and see
if that helps non-PERLIO IMP_SYS on Win32.
(Miniperl okay on linux).
p4raw-id: //depot/perlio@16350
Nick Ing-Simmons [Thu, 2 May 2002 16:10:15 +0000]
Win32 builds and mostly works for non-USE_PERLIO non-USE_IMP_SYS case.
- move body of fdupopen() from perlhost.h to win32.h as win32_fdupopen()
- use it in perlio.c
p4raw-id: //depot/perlio@16349
Gurusamy Sarathy [Thu, 2 May 2002 15:44:48 +0000]
manual integrate of change#16332 from maint-5.6 branch
p4raw-link: @16332 on //depot/maint-5.6/perl:
9bf7742e23b67e3d7c671615795c570c51951513
p4raw-id: //depot/perl@16348
Nick Ing-Simmons [Thu, 2 May 2002 15:19:59 +0000]
More layer syms
p4raw-id: //depot/perlio@16347
Nick Ing-Simmons [Thu, 2 May 2002 15:19:20 +0000]
Use PerlSIO_fdupopen() if not using PerlIO
p4raw-id: //depot/perlio@16346
Nick Ing-Simmons [Thu, 2 May 2002 15:18:12 +0000]
Do not build if not using layers
p4raw-id: //depot/perlio@16345
Jarkko Hietaniemi [Thu, 2 May 2002 14:02:51 +0000]
Upgrade to Encode 1.67.
p4raw-id: //depot/perl@16344
Jarkko Hietaniemi [Thu, 2 May 2002 13:14:39 +0000]
Integrate perlio:
[ 16341]
Remove show-stoppers to non-PERLIO Win32 builds
(still not a very good show).
p4raw-link: @16341 on //depot/perlio:
b6d604f4723f29adddfe9974781662588dd25ed7
p4raw-id: //depot/perl@16343
Hugo van der Sanden [Thu, 2 May 2002 04:22:02 +0000]
Re: [ID
20020425.009] perlio & -Dm dumps core
Message-Id: <
200205020322.g423M2p19726@crypt.compulink.co.uk>
p4raw-id: //depot/perl@16342
Nick Ing-Simmons [Thu, 2 May 2002 12:57:23 +0000]
Remove show-stoppers to non-PERLIO Win32 builds
(still not a very good show).
p4raw-id: //depot/perlio@16341
Nicholas Clark [Wed, 1 May 2002 22:58:42 +0000]
1/3rd of utf8
Message-ID: <
20020501215842.GB303@Bagpuss.unfortu.net>
p4raw-id: //depot/perl@16340
Jarkko Hietaniemi [Thu, 2 May 2002 12:41:18 +0000]
Integrate #16334 from maint-5.6;
don't pad $0 with spaces when setting it, use nulls instead (the
spaces show up as a very long line in a ps listing)
p4raw-id: //depot/perl@16339
p4raw-integrated: from //depot/maint-5.6/perl@16338 'merge in' mg.c
(@16332..)
Jarkko Hietaniemi [Thu, 2 May 2002 12:37:22 +0000]
Integrate perlio:
[ 16335]
Fix the "PerlIO require leak". Snag was that clean_objs
autoloaded DESTROY (needing IO) after known layers had
been freed. Postpone layer list free to PerlIO_cleanup,
rather than PerlIO_destruct. Tweak sequence in perl_destruct
so that sv_undef is immortal till layer list is done with it.
p4raw-link: @16335 on //depot/perlio:
9f4bd2225c152364d8589911c4c769edae165d9b
p4raw-id: //depot/perl@16338