27 years agoCorrect threads_mutex locking in main thread destruction.
Malcolm Beattie [Thu, 16 Oct 1997 16:26:53 +0000]
Correct threads_mutex locking in main thread destruction.
Add per-interp thrsv to hold SV struct thread for main thread.
Move Thread.xs MUTEX_DESTROY from end of threadstart to remove_thread.
Add Thread/list.t test of Thread->list method.
Let Thread::Semaphore methods up and down take an extra argument.

p4raw-id: //depot/perl@140

27 years agoFix up merge with 5.004_04.
Malcolm Beattie [Thu, 16 Oct 1997 14:01:11 +0000]
Fix up merge with 5.004_04.

p4raw-id: //depot/perl@139

27 years agoMerge maint-5.004 branch (5.004_04) with mainline.
Malcolm Beattie [Thu, 16 Oct 1997 11:09:25 +0000]
Merge maint-5.004 branch (5.004_04) with mainline.

p4raw-id: //depot/perl@137

27 years agoFinish thread state machine: fixes global destruction of threads,
Malcolm Beattie [Wed, 15 Oct 1997 16:57:45 +0000]
Finish thread state machine: fixes global destruction of threads,
detaching, joining etc. Alter FAKE_THREADS-specific fields to use
new HAVE_THREAD_INTERN stuff. Updates docs. Various fixes to
Thread.xs.

p4raw-id: //depot/perl@131

27 years agoAdd HAVE_THREAD_INTERN for platform-dependent struct thread additions.
Malcolm Beattie [Wed, 15 Oct 1997 16:55:10 +0000]
Add HAVE_THREAD_INTERN for platform-dependent struct thread additions.
Fix ThrSETSTATE not to lock t->mutex itself.

p4raw-id: //depot/perl@130

27 years agoRemove stale code from pp_entersub which breaks sub ownership locks.
Malcolm Beattie [Wed, 15 Oct 1997 16:53:35 +0000]
Remove stale code from pp_entersub which breaks sub ownership locks.

p4raw-id: //depot/perl@129

27 years agoMaintenance 5.004_04 changes
Tim Bunce [Wed, 15 Oct 1997 15:55:26 +0000]
Maintenance 5.004_04 changes

p4raw-id: //depot/maint-5.004/perl@128

27 years agoAdded fakethr.h.
Malcolm Beattie [Wed, 15 Oct 1997 10:00:18 +0000]
Added fakethr.h.

p4raw-id: //depot/perl@127

27 years agopthread_condattr_init in thread.h for OLD_PTHREADS_API.
Malcolm Beattie [Wed, 15 Oct 1997 09:50:57 +0000]
pthread_condattr_init in thread.h for OLD_PTHREADS_API.

p4raw-id: //depot/perl@126

27 years agoStarted rewriting thread state machine.
Malcolm Beattie [Wed, 15 Oct 1997 09:09:24 +0000]
Started rewriting thread state machine.

p4raw-id: //depot/perl@125

27 years agoRewrite thread destruction system using linked list of threads.
Malcolm Beattie [Fri, 10 Oct 1997 17:22:46 +0000]
Rewrite thread destruction system using linked list of threads.
Still not completely done. Add methods self, equal, flags, list
to Thread.xs. Add Thread_MAGIC_SIGNATURE check to typemap.

p4raw-id: //depot/perl@120

27 years agoFix up locking/synchronisation for pp_entersub.
Malcolm Beattie [Fri, 10 Oct 1997 17:19:55 +0000]
Fix up locking/synchronisation for pp_entersub.

p4raw-id: //depot/perl@119

27 years agoPut back entries in MANIFEST for the four now-returned win32/* files
Malcolm Beattie [Fri, 10 Oct 1997 09:55:32 +0000]
Put back entries in MANIFEST for the four now-returned win32/* files

p4raw-id: //depot/perl@118

27 years agoTook out mystack_foo for good, fixed up interp.sym and win32/makedef.pl
Malcolm Beattie [Fri, 10 Oct 1997 08:12:23 +0000]
Took out mystack_foo for good, fixed up interp.sym and win32/makedef.pl

p4raw-id: //depot/perl@117

27 years agoFix up 5.004_03 merge: remove missing win32 files from MANIFEST,
Malcolm Beattie [Wed, 8 Oct 1997 15:40:46 +0000]
Fix up 5.004_03 merge: remove missing win32 files from MANIFEST,
add missing dTHR; to new function unwind_handler_stack() in mg.c
and bump patchlevel.h to 5.004_52.

p4raw-id: //depot/perl@115

27 years agoMerge maint-5.004 branch (5.004_03) with mainline.
Malcolm Beattie [Wed, 8 Oct 1997 10:19:27 +0000]
Merge maint-5.004 branch (5.004_03) with mainline.
MANIFEST is out of sync.

p4raw-id: //depot/perl@114

27 years agoMove init of global mutexes/cond vars earlier.
Malcolm Beattie [Sun, 5 Oct 1997 17:52:49 +0000]
Move init of global mutexes/cond vars earlier.

p4raw-id: //depot/perl@113

27 years agoRemove last traces of "tokenbuf as temp buffer" and removed it
Malcolm Beattie [Fri, 3 Oct 1997 17:12:33 +0000]
Remove last traces of "tokenbuf as temp buffer" and removed it
from struct thread. Added missing thr->Tfoo defines for statbuf
and timesbuf and removed unused Tbuf field.

p4raw-id: //depot/perl@106

27 years agoBack out sv_bless3 change which made pp_bless zap '~'-magic.
Malcolm Beattie [Fri, 3 Oct 1997 15:23:25 +0000]
Back out sv_bless3 change which made pp_bless zap '~'-magic.

p4raw-id: //depot/perl@104

27 years agoFixed sv_mutex locking for new_SV, del_SV and nice_chunks.
Malcolm Beattie [Fri, 3 Oct 1997 15:17:39 +0000]
Fixed sv_mutex locking for new_SV, del_SV and nice_chunks.

p4raw-id: //depot/perl@103

27 years agoReliable thread signal handling.
Malcolm Beattie [Fri, 3 Oct 1997 11:53:51 +0000]
Reliable thread signal handling.

p4raw-id: //depot/perl@102

27 years agoBumped patchlevel to 51. Updated Todo.5.005.
Malcolm Beattie [Thu, 2 Oct 1997 17:19:44 +0000]
Bumped patchlevel to 51. Updated Todo.5.005.

p4raw-id: //depot/perl@98

27 years agoUpdate README.threads amd Thread/README
Malcolm Beattie [Thu, 2 Oct 1997 17:07:47 +0000]
Update README.threads amd Thread/README

p4raw-id: //depot/perl@97

27 years agoConfigure -Dusethreads hints for dec_osf and solaris_2 and
Malcolm Beattie [Thu, 2 Oct 1997 16:58:47 +0000]
Configure -Dusethreads hints for dec_osf and solaris_2 and
fix sv_bless3 prototype.

p4raw-id: //depot/perl@96

27 years agoAdd Todo.5.005 to MANIFEST and submit remade embed.h.
Malcolm Beattie [Thu, 2 Oct 1997 13:44:46 +0000]
Add Todo.5.005 to MANIFEST and submit remade embed.h.

p4raw-id: //depot/perl@93

27 years agoAdd Todo.5.005
Malcolm Beattie [Thu, 2 Oct 1997 13:27:10 +0000]
Add Todo.5.005

p4raw-id: //depot/perl@92

27 years agoStart of Configure support for -Dusethreads plus associated
Malcolm Beattie [Wed, 1 Oct 1997 17:04:12 +0000]
Start of Configure support for -Dusethreads plus associated
Linux hints.

p4raw-id: //depot/perl@88

27 years agoMove runops_foo prototypes from proto.h to early in perl.h.
Malcolm Beattie [Wed, 1 Oct 1997 17:03:34 +0000]
Move runops_foo prototypes from proto.h to early in perl.h.

p4raw-id: //depot/perl@87

27 years agoMerge maint-5.004 branch (5.004_01) with mainline.
Malcolm Beattie [Tue, 30 Sep 1997 15:11:07 +0000]
Merge maint-5.004 branch (5.004_01) with mainline.

p4raw-id: //depot/perl@84

27 years agoMaintenance 5.004_03 changes (addendum)
Tim Bunce [Tue, 30 Sep 1997 14:27:09 +0000]
Maintenance 5.004_03 changes (addendum)

p4raw-id: //depot/maint-5.004/perl@83

27 years agoMaintenance 5.004_03 changes
Tim Bunce [Tue, 30 Sep 1997 14:11:29 +0000]
Maintenance 5.004_03 changes

p4raw-id: //depot/maint-5.004/perl@82

27 years agoMaintenance 5.004_02 changes
Tim Bunce [Tue, 30 Sep 1997 13:17:27 +0000]
Maintenance 5.004_02 changes

p4raw-id: //depot/maint-5.004/perl@81

27 years agoRe-introduce the changes from change 68 (runops becomes a
Malcolm Beattie [Mon, 29 Sep 1997 16:57:23 +0000]
Re-introduce the changes from change 68 (runops becomes a
function pointer and sv_bless3 for '~'-magic) which got lost
during the preparation for the maint-merge.

p4raw-id: //depot/perl@78

27 years agoStart merge with maint-5.004 branch by creating an ancestral
Malcolm Beattie [Mon, 29 Sep 1997 16:44:16 +0000]
Start merge with maint-5.004 branch by creating an ancestral
branch point via a fake resolution with the maint-merge branch.
See Perforce Tech Note 9 for details.

p4raw-id: //depot/perl@77

27 years agostruct thread now stored in an SV and uses '~'-magic for access.
Malcolm Beattie [Mon, 22 Sep 1997 16:02:37 +0000]
struct thread now stored in an SV and uses '~'-magic for access.

p4raw-id: //depot/perl@69

27 years agorunops becomes a funtion pointer and sv_bless3 created
Malcolm Beattie [Mon, 22 Sep 1997 16:01:48 +0000]
runops becomes a funtion pointer and sv_bless3 created
to avoid pointer forgery with '~'-magic.

p4raw-id: //depot/perl@68

27 years agoMore fprintf -> PerlIO_printf changes.
Malcolm Beattie [Mon, 22 Sep 1997 15:45:56 +0000]
More fprintf -> PerlIO_printf changes.

p4raw-id: //depot/perl@67

27 years agoMinor multi-threading patches for VMS.
Malcolm Beattie [Mon, 22 Sep 1997 15:10:40 +0000]
Minor multi-threading patches for VMS.

p4raw-id: //depot/perl@66

27 years agoDebugging output for lock handling.
Malcolm Beattie [Wed, 10 Sep 1997 16:39:41 +0000]
Debugging output for lock handling.

p4raw-id: //depot/perl@64

27 years agoSolaris fixes: delete pad and padname from thread.h and remove
Malcolm Beattie [Wed, 10 Sep 1997 13:56:50 +0000]
Solaris fixes: delete pad and padname from thread.h and remove
MUTEX_* stuff when malloc.c gets copied to x2p/malloc.c.

p4raw-id: //depot/perl@61

27 years agoUpdate README.threads to mention -DL.
Malcolm Beattie [Tue, 9 Sep 1997 16:57:41 +0000]
Update README.threads to mention -DL.

p4raw-id: //depot/perl@60

27 years agoUpdate README.threads
Malcolm Beattie [Tue, 9 Sep 1997 16:33:45 +0000]
Update README.threads

p4raw-id: //depot/perl@58

27 years agoRewrite synchronisation of subs/methods and add attrs
Malcolm Beattie [Tue, 9 Sep 1997 15:04:26 +0000]
Rewrite synchronisation of subs/methods and add attrs
extension for specifying 'locked' and 'method' attributes.

p4raw-id: //depot/perl@56

27 years ago[inseparable changes from patch to perl 5.004_04]
Perl 5 Porters [Fri, 5 Sep 1997 00:00:00 +0000]
[inseparable changes from patch to perl 5.004_04]

[editor's note: this one imported like a charm!]

 TESTS  -

Subject: Improve pragma/locale test 102 - and don't fail, just warn
From: Jarkko Hietaniemi <jhi@anna.in-berlin.de>
Files: t/pragma/locale.t

Subject: Invalid test output in t/op/taint.t in trial 1
From: Dan Sugalski <sugalsd@lbcc.cc.or.us>
Files: t/op/taint.t

    t/op/taint.t prints out invalid ok messages for tests it skips.
    Rather than printing "ok 136" it prints "136 ok".

    p5p-msgid: 3.0.3.32.19970919160918.00857a50@stargate.lbcc.cc.or.us

 UTILITIES  -

Subject: Perldoc tiny patch to avoid $0
From: Ilya Zakharevich <ilya@math.ohio-state.edu>
Files: utils/perldoc.PL
Msg-ID: 199709122141.RAA16846@monk.mps.ohio-state.edu

    (applied based on p5p patch as commit 0b166b6635cf199f072db516b2a523ee659394d5)

Subject: h2ph broken in 5.004_02
From: David Mazieres <dm@reeducation-labor.lcs.mit.edu>
Files: utils/h2ph.PL
Msg-ID: 199708201700.KAA02621@www.chapin.edu

    (applied based on p5p patch as commit 4a8e146e38ec2045f1f817a7cb578e1b1f80f39f)

Subject: add key_t caddr_t to h2ph
From: Tony Sanders <sanders@bsdi.com>
Files: eg/sysvipc/ipcsem utils/h2ph.PL
Msg-ID: 199708272301.RAA12803@austin.bsdi.com

    (applied based on p5p patch as commit 0806a92ffc3a74ca70aa81051cdf2a306cd0a8af)

Subject: perldoc search ., lib and blib/* if -f 'Makefile.PL'
From: Tim Bunce <Tim.Bunce@ig.co.uk>
Files: utils/perldoc.PL

    Subject: perldoc finds wrong pod2man

    (from perldoc source)

        # We must look both in @INC for library modules and in PATH
        # for executables, like h2xs or perldoc itself.

    Unfortunately, searching PATH for installed perl executables like
    pod2man is INCORRECT.  perldoc should start by searching the
    directory it was executed from, which might not be in the PATH
    at all.

    Credited: Joseph "Moof-in'" Hall <joseph@cscaper.com>

    p5p-msgid: 199708251732.KAA19299@gadget.cscaper.com

Subject: 5.004m4t1: perlbug: NIS domainname gets into wrong places
From: Andreas J. Koenig <koenig@anna.mind.de>
Files: utils/perlbug.PL
Msg-ID: sfcg1qy38as.fsf@anna.in-berlin.de

    (applied based on p5p patch as commit 41f926b844140b7f7eaa9302113e45df3a9f9ff4)

Subject: add better local patch info to perlbug
From: Tim Bunce <Tim.Bunce@ig.co.uk>
Files: utils/perlbug.PL

Subject: perldoc - suggest modules if requested module not found
From: Anthony David <adavid@netinfo.com.au>
Files: utils/perldoc.PL

    private-msgid: 3439CD83.6969@netinfo.com.au

Subject: perldoc mail::foo tries to read binary /usr/ucb/mail
From: Tim Bunce <Tim.Bunce@ig.co.uk>
Files: utils/perldoc.PL

    Subject: perldoc weirdness

    perldoc mail::imap yields:

    {joseph}:79% perldoc mail::foo
    can't open /usr/ucb/mail: Permission denied at ./pod2man line 362.

    Credited: Joseph "Moof-in'" Hall <joseph@cscaper.com>

    p5p-msgid: 199710082014.NAA00808@gadget.cscaper.com

Subject: perldoc -f setpwent (for example) returns no descriptive text
From: Tim Bunce <Tim.Bunce@ig.co.uk>
Files: utils/perldoc.PL

Subject: perldoc diffs: don't search auto - much faster
From: "Joseph N. Hall" <joseph@5sigma.com>
Files: utils/perldoc.PL
Msg-ID: MailDrop1.2d7dPPC.971012211957@screechy.cscaper.com

    (applied based on p5p patch as commit 62b753c6ae4ab9bf22fbb6ec7ceac820bcef8fe4)

27 years ago[inseperable differences to perl 5.004_03]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[inseperable differences to perl 5.004_03]

[editor's note: the following patches could not be applied from the
list messages.  There are also various unattributed changes.]

  ------  BUILD PROCESS  ------

  Title:  "Configure can stop without fully explaining itself"
   From:  Jim Anderson <jander@ml.com>
 Msg-ID:  <199708111328.JAA28976@nsd15.ny-swaps-develop.ml.com>,
          <199708111952.PAA29346@nsd15.ny-swaps-develop.ml.com>
  Files:  Configure

  ------  CORE LANGUAGE  ------

  Title:  "GNU style perl --version (or any other --foo) ignored"
   From:  "M.J.T. Guy" <mjtg@cus.cam.ac.uk>, Kenneth Albanowski
          <kjahds@kjahds.com>, Stephen McCamant <alias@mcs.com>
 Msg-ID:  <E0wx8MO-0007BS-00@ursa.cus.cam.ac.uk>,
          <Pine.LNX.3.93.970813122557.9443C-100000@kjahds.com>,
          <m0wy8nl-000EYgC@alias-2.pr.mcs.net>
  Files:  pod/perldiag.pod perl.c

  ------  DOCUMENTATION  ------

  Title:  "perlop pod inconsistent in presentation of regexp options"
   From:  "M.J.T. Guy" <mjtg@cus.cam.ac.uk>, Hans Mulder <hansm@icgned.nl>,
          jmr@whirlwind.fmr.com
 Msg-ID:  <199708061404.KAA06717@whirlwind.fmr.com>,
          <199708081505.LAA09810@whirlwind.fmr.com>,
          <1997Aug7.160530.2196011@hmivax.humgen.upenn.edu>,
          <E0wwnqc-00057s-00@ursa.cus.cam.ac.uk>,
          <E0wwswg-00017x-00@ursa.cus.cam.ac.uk>
  Files:  pod/perlop.pod

  [ 2 messages had applied changes,
    d1a7f0f436d72614358862f92db9613296be2744 and
    b7e30b65e77616e7336a6cda54d9c3d5935d0cfc ]

  ------  PORTABILITY - WIN32  ------

  Title:  "[PATCH] Win95-proofing pl2bat"
   From:  Gurusamy Sarathy <gsar@engin.umich.edu>
 Msg-ID:  <199708121733.NAA14888@aatma.engin.umich.edu>
  Files:  win32/bin/*.bat

  [ d444a43172237b6bdd9f0a52017be3b0d792aa5c didn't apply the deletion
    patches, so the deletes in this commit are a carried error ]

  ------  PORTABILITY - OTHER  ------

  Title:  "5.004_02 Configure - worrying but normal errors displayed to user"
   From:  Paul Marquess <pmarquess@bfsec.bt.co.uk>, pmarquess@bfsec.bt.co.uk
          (Paul Marquess)
 Msg-ID:  <01BCA3DE.E257BFC0.pmarquess@bfsec.bt.co.uk>,
          <9708102159.AA11726@claudius.bfsec.bt.co.uk>
  Files:  Configure os2/diff.configure

  [ one patch found, see 61167c6fd6d55c5f975404dcb56c3d0a87cd2c21 ]

  Title:  "Minor glitch with Perl 5.004_01 on SunOS 4.1.3 (groupstype)"
   From:  thad@thadlabs.com (Thad Floryan)
 Msg-ID:  <9708111415.AA03808@thadlabs.com>
  Files:  hints/sunos_4_1.sh

27 years agoMissing 'require' in auto-generated .pm by h2xs
David Kerry [Fri, 5 Sep 1997 00:00:00 +0000]
Missing 'require' in auto-generated .pm by h2xs

private-msgid: 97Aug27.131618edt.11650@janus.tor.securecomputing.com

27 years agoperldoc diffs: don't search auto - much faster
Joseph N. Hall [Fri, 5 Sep 1997 00:00:00 +0000]
perldoc diffs: don't search auto - much faster

Give this a try.  It causes unsuccessful searches to run about
3 times as fast on my machine.  Not exactly a speed demon but
at least you don't have time to get up and make coffee.

I tried adding caching to stop the repeated readdir in the
case-insensitive matching subroutine, but a) it only lopped off
another 20% and b) the patch made a kind of a Frankenstein thing
out of the subroutine, so c) I will try to write a better one
later when I have time.

p5p-msgid: MailDrop1.2d7dPPC.971012211957@screechy.cscaper.com

27 years agofix poor t/op/runlevel.t test
Norton Allen [Fri, 5 Sep 1997 00:00:00 +0000]
fix poor t/op/runlevel.t test

This patch fixes the Can't "goto" outside block test. Before, the
test only hit the desired error if the system sort did a compare
where $a was 0, which is not guaranteed (and does not occur on at
least one system.) Now both arguments are tested.

Credited: Hugo van der Sanden <hv@crypt.compulink.co.uk>
Credited: Gurusamy Sarathy <gsar@engin.umich.edu>

p5p-msgid: 199709261458.KAA28611@dolores.harvard.edu

27 years ago5.004m4t1: perlbug: NIS domainname gets into wrong places
Andreas J. Koenig [Fri, 5 Sep 1997 00:00:00 +0000]
5.004m4t1: perlbug: NIS domainname gets into wrong places

>>>>> Jarkko Hietaniemi writes:

jhi> [patch mode]

jhi> I suggest the following patch: appending the ".`domainname`" only iff
jhi> it contains dots and when last element has only [A-Za-z] and is no
jhi> longer than 3 characters.  What were the latest plans about adding new
jhi> top level domains?  Will these heuristics still be true?

No. And I recall we have been flamed heavily in the past for using
`domainname` at all and I'm surprised that we still do.

I'd suggest such a patch instead:

p5p-msgid: sfcg1qy38as.fsf@anna.in-berlin.de

27 years agoIdentify t/*/*.t test failing because of file permissions
Andreas J. Koenig [Fri, 5 Sep 1997 00:00:00 +0000]
Identify t/*/*.t test failing because of file permissions

tim> I've added
tim>  system("find t -name '*.t' -print | xargs chmod +x");
tim> to my Porting/makerel script. So the problem should not reappear.

That doesn't help for the patch. The perl5.004m4t1.patch.gz contained

    touch t/lib/dosglob.t

but no

    chmod +x t/lib/dosglob.t

Short term I'd recommend Ilya's one-line patch to t/TEST because it at
least saves us debugging time. Stupid me lost over an hour not
understanding the reason for the failure. Here it is:

p5p-msgid: sfcraah0xvy.fsf@anna.in-berlin.de

27 years agoeg/sysvipc/ipcsem bug, new hints/bsdos.sh
Tony Sanders [Sat, 23 Aug 1997 22:17:29 +0000]
eg/sysvipc/ipcsem bug, new hints/bsdos.sh

    There appears to be a bug in the example ipcsem with the number of
    arguments to the SYSV IPC semop() function -- the POD's agree that
    it only takes two arguments now.

    The version of hints/bsdos.sh below has been cleaned up and
    updated for the next BSD/OS release.

p5p-msgid: 199708272301.RAA12803@austin.bsdi.com

27 years agotests fail if localhost/loopback address not defined
David McLean [Fri, 5 Sep 1997 00:00:00 +0000]
tests fail if localhost/loopback address not defined

Subject: HPUX Perl problem

Larry:

I did a sh Configure -des after downloading perl5.004_03.
This was done on workstation running HPUX 10.20.

All looked normal.

Then I did a make and everything also looked normal.

However the test resulted in:

lib/io_sock.......Bad file number at ./lib/io_sock.t line 55.
FAILED at test 2
lib/io_udp........Can't call method "sockname" without a package or
object reference at ./lib/io_udp.t line 35.
FAILED at test 2
lib/odbm..........Bad free() ignored at ./lib/odbm.t line 63.
Bad free() ignored at (eval 4) line 2.

p5p-msgid: 34048947.2944@icc.gsfc.nasa.gov

27 years agoh2ph broken in 5.004_02
Kurt D. Starsinic [Tue, 19 Aug 1997 21:43:32 +0000]
h2ph broken in 5.004_02

> The h2ph in perl5.00402 generates perl code that the interpeter
> chokes on.  For example, running h2ph on /usr/include/sys/errno.h
> (under openbsd) transforms this:
>
>  #define EPERM           1               /* Operation not permitted */
>
> into this:
>
>  unless(defined(&EPERM) {
>  sub EPERM () {1;}
>  }
>
> which in turn causes this:
>
>  syntax error at - line 2, near "sub EPERM ()"
>
> I'm not exactly sure why h2ph is worried about avoiding redefinition . . .

    It's not a redefinition problem; errno.ph is missing a right paren.  Here
is a patch, tested against 5.004_02:

Credited: David Mazieres <dm@reeducation-labor.lcs.mit.edu>

p5p-msgid: 199708201700.KAA02621@www.chapin.edu

27 years agoop/glob.t test failure under Win32 with CVS
Warren Jones [Fri, 5 Sep 1997 00:00:00 +0000]
op/glob.t test failure under Win32 with CVS

This is a bug report for perl from wjones@tc.fluke.com,
generated with the help of perlbug 1.19 running under perl 5.00403.

p5p-msgid: 97Aug26.091048pdt.35761-1@gateway.fluke.com

27 years agoPerldoc tiny patch to avoid $0
Ilya Zakharevich [Mon, 28 Jul 1997 04:23:32 +0000]
Perldoc tiny patch to avoid $0

Editing $0 may be not-so-portable.

Enjoy,

p5p-msgid: 199709122141.RAA16846@monk.mps.ohio-state.edu

27 years agoOS/2 Hints
Ilya Zakharevich [Fri, 5 Sep 1997 00:00:00 +0000]
OS/2 Hints

People build OS/2 Perl on weirder and weirder configured systems, so I
needed to add a full-featured gcc environment parser... :-(

Enjoy,

p5p-msgid: 199710130631.CAA25426@monk.mps.ohio-state.edu

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoNew hints file for IBM OS/390 OpenEdition (MVS)
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
New hints file for IBM OS/390 OpenEdition (MVS)

(this is the same change as commit c13c896351005f8551466fa8cd5d7293b5332fa0, but as applied)

27 years agoNew hints file for IBM OS/390 OpenEdition (MVS)
Peter Prymmer [Fri, 5 Sep 1997 00:00:00 +0000]
New hints file for IBM OS/390 OpenEdition (MVS)

I'd like to thank the folks mentioned for their help.
This file by itself does not do the build - but every
little bit helps.

Peter Prymmer

p5p-msgid: 9709240106.AA26484@forte.com

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agohints/qnx.sh update
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
hints/qnx.sh update

(this is the same change as commit 0419447e3dca6b659723638f9b0899e6c3ee37ca, but as applied)

27 years agohints/qnx.sh update
Norton Allen [Fri, 5 Sep 1997 00:00:00 +0000]
hints/qnx.sh update

This patch cleans up the QNX hints file, bringing it up to date
with 5.004_04. -Norton Allen

p5p-msgid: 199709261508.LAA07889@dolores.harvard.edu

27 years agoMinor OS/2 patch for 4_03
Ilya Zakharevich [Fri, 5 Sep 1997 00:00:00 +0000]
Minor OS/2 patch for 4_03

This makes "$^X -w script" skip the shell though $^X contains backslashes:

Only os2/os2.c is touched.

Enjoy,

p5p-msgid: 199710032224.SAA15345@monk.mps.ohio-state.edu

27 years agoDynixptx hints
Bruce P. Schuck [Fri, 5 Sep 1997 00:00:00 +0000]
Dynixptx hints

Before I say anything else, let me say that the O'Reilly Perl Conference
was super!

In compiling Perl5.004_1 on our Sequent boxes running DYNIX/ptx 4.0 V4.2.1
we found that we had to add the following ccflags and cppflags to get the
sockets io to work.

 cppflags='-Wc,+abi-socket -I/usr/local/include'
 ccflags='-Wc,+abi-socket -I/usr/local/include'

This also could have been set by setting an environment variable
_SEQUENT_CC_=-Wc,+abi-socket
before compiling.

I have attached an snippet of the socket(3N) man pages as well.  This also
must be noted when compiling and adding other modules as well.

p5p-msgid: Pine.PTX.3.95.971002104651.12112G-200000@lancelot.aps.org

27 years agoFix perl build on Digital UNIX after JDK installs libnet.so
Spider Boardman [Fri, 5 Sep 1997 00:00:00 +0000]
Fix perl build on Digital UNIX after JDK installs libnet.so

Subject: [PATCH] Perl & JAVA co-existence

It turns out that the JDK on Digital UNIX installs a libnet.so
which gets in the way when building Perl.  Back during 5.003_28,
I didn't do anything to remove libnet because there wasn't one,
but now there can be a libnet, and it's wrong.  So, here's a
patch to the dec_osf hints file to skip libnet as well as some of
the others we're already ignoring.

p5p-msgid: 199709191826.OAA18040@Orb.Nashua.NH.US

27 years agolib/timelocal.t fails test 1 for VMS 7.1
Dan Sugalski [Fri, 5 Sep 1997 00:00:00 +0000]
lib/timelocal.t fails test 1 for VMS 7.1

The VMS Perl port has local routines to handle time(), localtime(), and
gmtime(), because the implementations of these on VMS are broken for VMS
v6.2 and below. This implemetation gets strange for most of Jan 1, 1970
(From midnight to 4 PM), and behaves oddly for daylight savings time, but
only on OpenVMS 7.0 and 7.1.

The following patch disables the local versions of time(), localtime(), and
gmtime(), and uses the system provided versions, for OpenVMS 7.0 and up.
5.004_03 passes all tests with this patch. Tested on OpenVMS Alpha 7.1 and
6.2, and OpenVMS Vax 7.1, 6.2, and 5.5-2.

p5p-msgid: 3.0.3.32.19970908112449.0087bc90@stargate.lbcc.cc.or.us

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoIrix 6.2 build problem - so_locations
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Irix 6.2 build problem - so_locations

(this is the same change as commit 62e637042b7d9758a9b3364fbddc1d6eecaf53c8, but as applied)

27 years agoIrix 6.2 build problem - so_locations
David Billinghurst [Fri, 5 Sep 1997 00:00:00 +0000]
Irix 6.2 build problem - so_locations

Subject: Perl 5.004_03 - Irix 6.2 build problem - probably fixed

I have just built and install perl 5.004_03 on and SGI Power Challenge
under Irix 6.2 with the SGI Mongoose 7.1 compiler.

Problem

Did the standard "sh Configure -des; make".  Had a problem building
shared libraries, with compaints about file ./so_locations.

Solution:

  [ patch ]

Caveat:

You should get an SGI guru to check this.

p5p-msgid: D54B1932FFB4CF11B5C80000F8018BD2907E31@CRCMAIL

27 years agoTweak to hints/machten.sh: stop t/lib/complex.t from failing
Dominic Dunlop [Fri, 5 Sep 1997 00:00:00 +0000]
Tweak to hints/machten.sh: stop t/lib/complex.t from failing

Tim, this patch, posted to you and copied to p5p on 23rd September, and
posted to you again on 10th October, seems to have slipped through your
net.  Please please please put it in before 5.004_04.  (Copied again to p5p
this time in the hope of achieving better reception through multipath
diversity.)

Thanks

Credited: Dominic Dunlop <domo@slipper.ip.lu>

p5p-msgid: v03110700b06a30bdfc42@[194.51.248.80]

27 years agoBug in Win32::GetShortPathName
Gurusamy Sarathy [Fri, 5 Sep 1997 00:00:00 +0000]
Bug in Win32::GetShortPathName

On Thu, 09 Oct 1997 08:30:55 PDT, "Greg Chapman" wrote:
>I'm using Perl 5.004_02 under WIndows 95 (the latest binary distribution
>available on CPAN).  The Win32::GetShortPathName function does not properly
>truncate the returned string when the short file name is shorter than the
>long name.  Specifically, the returned string looks like:
><ShortFileName><NULL char><characters from the end of the long name (the
>original buffer)>.

Thanks for that report, and here's a patch.

p5p-msgid: 199710092229.SAA21556@aatma.engin.umich.edu

27 years agopl2bat.bat -> pl2bat.pl change in win32/pod.mak
Jan Dubois [Fri, 5 Sep 1997 00:00:00 +0000]
pl2bat.bat -> pl2bat.pl change in win32/pod.mak

On Thu, 4 Sep 1997 23:45:42 +0100, you wrote:

>These are the changes I'm considering for _04:

[snip]

>Tim.

Maybe I'm missing something, but I think the following mini-patch (which
I send while everyone was at the conference) is essential for
*installing* on Win32 (with MSC 4.2 on NT4.0). I'm just very suprised
that noboday else seems to encounter this problem. Perhaps everyone is
using the binary distribution?

Maybe Gurusamy could comment if I'm right?

p5p-msgid: 3411ee6f.9143607@smtp-gw01.ny.us.ibm.net

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoWin32 archnames
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Win32 archnames

(this is the same change as commit 7e8ba8fa5f78d7fa23683d0f33555018e530f3e0, but as applied)

27 years agoWin32 archnames
Bill Middleton [Fri, 5 Sep 1997 00:00:00 +0000]
Win32 archnames

Subject: [PATCH] Re: Perl for Win32

> It's pretty fundamental that $archname be different for binary INcompatible
> systems. If it's not then that's a bug.

Here's a proposed patch for the config.h templates which
are distributed with the kit

Credited: Gurusamy Sarathy <gsar@engin.umich.edu>
Credited: Peter Prymmer <pvhp@forte.com>
Credited: Tim Bunce <Tim.Bunce@ig.co.uk>

p5p-msgid: Pine.GSO.3.95.970905123145.12361B-100000@ducks

27 years agoEnhance perly.fixer to help porters.
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Enhance perly.fixer to help porters.

27 years agoEmacs CPerl update for 5.004_04
Ilya Zakharevich [Fri, 5 Sep 1997 00:00:00 +0000]
Emacs CPerl update for 5.004_04

private-msgid: 199710140835.EAA26825@monk.mps.ohio-state.edu

27 years agomyconfig / perl -V: remove randbits and add prototype
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
myconfig / perl -V: remove randbits and add prototype

In the myconfig / perl -V output, like the following example:

Summary of my perl5 (5.0 patchlevel 4 subversion 1) configuration:
  Platform:
    osname=solaris, osvers=2.5, archname=sun4-solaris
    uname='sunos lion 5.5 generic_103093-03 sun4m sparc sunw,sparcstation-10 '
    hint=recommended, useposix=true, d_sigaction=define
    bincompat3=y useperlio= d_sfio=
  Compiler:
    cc='gcc', optimize='-O6', gccversion=2.7.2.2
    cppflags='-I/usr/local/include -I/opt/gnu/include'
    ccflags ='-I/usr/local/include -I/opt/gnu/include'
    stdchar='unsigned char', d_stdstdio=define, usevfork=false
    voidflags=15, castflags=0, d_casti32=define, d_castneg=define
    intsize=4, alignbytes=8, usemymalloc=y, randbits=15
  Linker and Libraries:
    ld='gcc', ldflags =' -L/usr/local/lib -L/opt/gnu/lib'
    libpth=/usr/local/lib /opt/gnu/lib /lib /usr/lib /usr/ccs/lib
    libs=-lsocket -lnsl -ldb -ldl -lm -lc -lcrypt
    libc=/lib/libc.so, so=so
    useshrplib=false, libperl=libperl.a
  Dynamic Linking:
    dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=, ccdlflags=' '
    cccdlflags='-fpic', lddlflags='-G -L/usr/local/lib -L/opt/gnu/lib'

I'd like to change $randbits, in the Compiler section, for $prototype.

Why remove randbits: randbits was added before we had such good and
informative tests in the test suite. The tests now pinpoint the problem
and explain the solution, including recommending the right value for randbits.

Why add prototype: In the medium term there may be advantages to moving
towards using ANSI C prototypes. At the moment there's no way of making
an impact assessment. This would be a first step. Especially when combined
with automated perlbug and perlbug -ok mail handling and archiving.

While I'm on the subject, libpth and libs values should be in quotes.

Any objections?

(Please don't turn this into a "my compiler doesn't support prototypes" rant.)

p5p-msgid: 199709290857.JAA07706@toad.ig.co.uk

27 years agoUpdated CPAN.pm for 5.004_04
Andreas J. Koenig [Fri, 5 Sep 1997 00:00:00 +0000]
Updated CPAN.pm for 5.004_04

private-msgid: sfcpvpv8teo.fsf@anna.in-berlin.de

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoTrap Time::Local infinite loop
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Trap Time::Local infinite loop

(this is the same change as commit 160771401404fad17bea1cbac474c73580214f79, but as applied)

27 years agoTrap Time::Local infinite loop
Hugo van der Sanden [Fri, 5 Sep 1997 00:00:00 +0000]
Trap Time::Local infinite loop

Here's a new patch for Time::Local that dies if it detects that it is
looping:

work% perl -wle 'use Time::Local; print timegm(8, 14, 3, 19, 11, 997)'
Couldn't handle date (8, 14, 3, 19, 11, 997) at -e line 1
work%

Under Linux here, it handles all dates in the years 1970 thru 2038
(70-99, 00-38) and dies for all other years. I don't try to address
the fact that strange values are returned for dates in '38 from
February onwards: the 31-bit epoch ends at 03:14:07 19th Jan 2038.

p5p-msgid: 199710030030.BAA17372@crypt.compulink.co.uk

27 years agoBanishing eval from getopt.pl and Getopt/Std.pm
John L. Allen [Fri, 5 Sep 1997 00:00:00 +0000]
Banishing eval from getopt.pl and Getopt/Std.pm

This is a minimal patch *adding no new features*.
I submitted a prior patch that reworded the pods and allowed -- to
terminate args, among other things, but it never made it
in.  Perhaps this will.

p5p-msgid: Pine.SOL.3.91.970920154720.3683A@gateway

27 years agoxsubpp fix to allow #ifdef's around entire XSubs
John Tobey [Fri, 5 Sep 1997 00:00:00 +0000]
xsubpp fix to allow #ifdef's around entire XSubs

This rather trivial patch fixes a feature broken in 5.004_02's
xsubpp.  Namely, #line directives are interfering with the ability to
put #ifdef around entire XSubs.

(Well I'm glad I caught my own bug. ;-)  Hope it didn't break too many
of the CPAN modules!!)

p5p-msgid: 199709070034.AAA16457@remote119

27 years agoDocument that File::Find doesn't follow symlinks
Greg Ward [Fri, 5 Sep 1997 00:00:00 +0000]
Document that File::Find doesn't follow symlinks

Subject: following symlinks with File::Find

The behaviour of File::Find with respect to symbolic links is inconsistent
between Perl 5.002/5.003 and 5.004_02.  Since 5.004_02's behaviour (don't
follow symlinks) is more consistent with `find' itself, this is probably
more of a documentation bug -- i.e., the File::Find man page should
mention that 1) it doesn't follow symlinks, and 2) previous versions
did, so watch out!

p5p-msgid: 199708191853.OAA07111@bottom.bic.mni.mcgill.ca

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoSyslog.pm and missing _PATH_LOG
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Syslog.pm and missing _PATH_LOG

(this is the same change as commit d6a06c6b3ce238e0862e09fdca0c17e9d40e9b2c, but as applied)

27 years agoSyslog.pm and missing _PATH_LOG
Ulrich Pfeifer [Fri, 5 Sep 1997 00:00:00 +0000]
Syslog.pm and missing _PATH_LOG

I just ran into a problem with Syslog.pm. I tried to install some
modules which did call Sys::Syslog::setlogsock('unix'). The following
connect fails because my syslog.ph does not define _PATH_LOG (This is
SunOS 5.5.1).

The easiest way to handle this would be to make setlogsock('unix') a
NOOP if _PATH_LOG is not defined. I am not advocating to ignore it
silently. But can we make setlogsock return a value indicating that
the operation did or did not work? It seems a little cleaner that to
force users to check for _PATH_LOG themselves?

p5p-msgid: p5iuw1cris.fsf@knowway.de.uu.net

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoIO::Socket autoflush by default, assume tcp and PeerAddr
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
IO::Socket autoflush by default, assume tcp and PeerAddr

(this is the same change as commit d103aadb464fdc9458aeaedf1165336dcbe99ff0, but as applied)

27 years agoIO::Socket autoflush by default, assume tcp and PeerAddr
Gisle Aas [Thu, 28 Aug 1997 02:36:50 +0000]
IO::Socket autoflush by default, assume tcp and PeerAddr

Subject: IO::Socket suggestion

The following patch to IO::Socket have this effect:

   - put IO::Socket objects in autoflush mode initially
   - assume Proto => "tcp" if no Proto is given (IO::Socket::INET)
   - a single argument to IO::Socket::INET->new is assumed to be
     a PeerAddr specification.

Comments?

It allows the following (rather long) one-liner to work:

 $ perl -I. -MIO::Socket -e '$s=IO::Socket::INET->new("www.perl.com:80");$s->print("HEAD / HTTP/1.0\n\n"); print<$s>'

Credited: Andy Dougherty <doughera@newton.phys.lafayette.edu>
Credited: M.J.T. Guy <mjtg@cus.cam.ac.uk>

p5p-msgid: hvi07zvo9.fsf@bergen.sn.no

27 years agoGetopt::Long 2.11
Johan Vromans [Fri, 5 Sep 1997 00:00:00 +0000]
Getopt::Long 2.11

private-msgid: m0xBcdR-000RArC@plume.nl.compuware.com

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agoSys::Hostname fails under Solaris 2.5 when setuid
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
Sys::Hostname fails under Solaris 2.5 when setuid

(this is the same change as commit b6d5cd8ca8d16f83d5c4c7a0bc602634e3efb321, but as applied)

27 years agoSys::Hostname fails under Solaris 2.5 when setuid
Patrick Hayes [Wed, 20 Aug 1997 01:05:13 +0000]
Sys::Hostname fails under Solaris 2.5 when setuid

encountered a problem when using the Sys::Hostname under Solaris 2.5.1 in
setuid scripts. The problem is due to Solaris' having removed gethostname from
the list of syscalls. gethostname is now a libc compatability function using
the 'sysinfo' syscall. As the environment I'm using this in requires setuid
execution, the other methods (calling `hostname or `uname -n`) are
inapplicable.

I also noticed that Sys::Hostname requires syscall.ph whereas I find only
sys/syscall.ph. The perl headers were created with the standard:
             cd /usr/include; h2ph * sys/*

I've added in the following code to make Hostname.pm work on Solaris. Would it
be possible to integrate this into future releases?

p5p-msgid: 199708201240.OAA04243@goblin.renault.fr

27 years ago[x] FileHandle.pm fails if Exporter has not been loaded previously
Jan Dubois [Fri, 5 Sep 1997 00:00:00 +0000]
[x] FileHandle.pm fails if Exporter has not been loaded previously

I just found out that FileHandle.pm doesn't work if I don't "use" an exporter module
before it:

% perl -MFileHandle -e "print $^O"
Bareword "Exporter::export" not allowed while "strict subs" in use at I:\Perl\lib/FileHandle.pm line 80.
String found where operator expected at I:\Perl\lib/FileHandle.pm line 80, near "Exporter::export 'Fcntl'"
(Do you need to predeclare Exporter::export?)
syntax error at I:\Perl\lib/FileHandle.pm line 80, near "Exporter::export 'Fcntl'"
BEGIN failed--compilation aborted.

% perl -MExporter -MFileHandle -e "print $^O"
MSWin32

This is with Perl5.004_04 trial 3 on WinNT., but it doesn't look like a new problem.

-Jan

p5p-msgid: 3445e05b.17874041@smtp2.ibm.net -> not

27 years agoxsubpp: document advanced dynamic typemap usage
Rujith S. de Silva [Fri, 5 Sep 1997 00:00:00 +0000]
xsubpp: document advanced dynamic typemap usage

27 years ago[dummy merge]
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
[dummy merge]

This merge exists so that the p5p version of the patch and the applied
version are both in the history

27 years agominor fix for perltrap.pod
Tim Bunce [Fri, 5 Sep 1997 00:00:00 +0000]
minor fix for perltrap.pod

(this is the same change as commit 9bf4ea843f7631689691908b2ccde8f571b18cde, but as applied)

27 years agominor fix for perltrap.pod
Abigail [Fri, 5 Sep 1997 00:00:00 +0000]
minor fix for perltrap.pod

perltrap says 'Variables begin with "$" or "@" in Perl.' (twice)
without addressing the possibility of "%". This patch fixes it.

Of course, variables can start with "*" (typeglobs) or no special
symbol (filehandles, functions) too. Should that be mentioned as
well?

p5p-msgid: 199709170500.BAA14805@fnx.com