Skip to content
Commits on Sep 5, 1997
  1. [inseparable changes from patch to perl 5.004_04]

    Perl 5 Porters committed with Tim Bunce Sep 5, 1997
    [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 0b166b6)
    
    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 4a8e146)
    
    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 0806a92)
    
    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 41f926b)
    
    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 62b753c)
  2. Missing 'require' in auto-generated .pm by h2xs

    David Kerry committed with Tim Bunce Sep 5, 1997
    private-msgid: 97Aug27.131618edt.11650@janus.tor.securecomputing.com
  3. perldoc diffs: don't search auto - much faster

    Joseph N. Hall committed with Tim Bunce Sep 5, 1997
    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
  4. @nthallen

    fix poor t/op/runlevel.t test

    nthallen committed with Tim Bunce Sep 5, 1997
    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
  5. 5.004m4t1: perlbug: NIS domainname gets into wrong places

    Andreas J. Koenig committed with Tim Bunce Sep 5, 1997
    >>>>> 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
  6. Identify t/*/*.t test failing because of file permissions

    Andreas J. Koenig committed with Tim Bunce Sep 5, 1997
    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
  7. eg/sysvipc/ipcsem bug, new hints/bsdos.sh

    Tony Sanders committed with Tim Bunce Aug 24, 1997
        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
  8. tests fail if localhost/loopback address not defined

    David McLean committed with Tim Bunce Sep 5, 1997
    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
  9. h2ph broken in 5.004_02

    Kurt D. Starsinic committed with Tim Bunce Aug 20, 1997
    > 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
  10. op/glob.t test failure under Win32 with CVS

    Warren Jones committed with Tim Bunce Sep 5, 1997
    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
  11. Perldoc tiny patch to avoid $0

    Ilya Zakharevich committed with Tim Bunce Jul 28, 1997
    Editing $0 may be not-so-portable.
    
    Enjoy,
    
    p5p-msgid: 199709122141.RAA16846@monk.mps.ohio-state.edu
  12. OS/2 Hints

    Ilya Zakharevich committed with Tim Bunce Sep 5, 1997
    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
  13. [dummy merge]

    Tim Bunce committed Sep 5, 1997
    This merge exists so that the p5p version of the patch and the applied
    version are both in the history
  14. New hints file for IBM OS/390 OpenEdition (MVS)

    Tim Bunce committed Sep 5, 1997
    (this is the same change as commit c13c896, but as applied)
  15. New hints file for IBM OS/390 OpenEdition (MVS)

    Peter Prymmer committed with Tim Bunce Sep 5, 1997
    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
  16. [dummy merge]

    Tim Bunce committed Sep 5, 1997
    This merge exists so that the p5p version of the patch and the applied
    version are both in the history
  17. hints/qnx.sh update

    Tim Bunce committed Sep 5, 1997
    (this is the same change as commit 0419447, but as applied)
  18. @nthallen

    hints/qnx.sh update

    nthallen committed with Tim Bunce Sep 5, 1997
    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
  19. Minor OS/2 patch for 4_03

    Ilya Zakharevich committed with Tim Bunce Sep 5, 1997
    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
  20. Dynixptx hints

    Bruce P. Schuck committed with Tim Bunce Sep 5, 1997
    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
  21. Fix perl build on Digital UNIX after JDK installs libnet.so

    Spider Boardman committed with Tim Bunce Sep 5, 1997
    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
  22. lib/timelocal.t fails test 1 for VMS 7.1

    Dan Sugalski committed with Tim Bunce Sep 5, 1997
    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
  23. [dummy merge]

    Tim Bunce committed Sep 5, 1997
    This merge exists so that the p5p version of the patch and the applied
    version are both in the history
  24. Irix 6.2 build problem - so_locations

    Tim Bunce committed Sep 5, 1997
    (this is the same change as commit 62e6370, but as applied)
  25. Irix 6.2 build problem - so_locations

    David Billinghurst committed with Tim Bunce Sep 5, 1997
    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
  26. Tweak to hints/machten.sh: stop t/lib/complex.t from failing

    Dominic Dunlop committed with Tim Bunce Sep 5, 1997
    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. Bug in Win32::GetShortPathName

    Gurusamy Sarathy committed with Tim Bunce Sep 5, 1997
    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
  28. pl2bat.bat -> pl2bat.pl change in win32/pod.mak

    Jan Dubois committed with Tim Bunce Sep 5, 1997
    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
  29. [dummy merge]

    Tim Bunce committed Sep 5, 1997
    This merge exists so that the p5p version of the patch and the applied
    version are both in the history
  30. Win32 archnames

    Tim Bunce committed Sep 5, 1997
    (this is the same change as commit 7e8ba8f, but as applied)
  31. Win32 archnames

    Bill Middleton committed with Tim Bunce Sep 5, 1997
    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
  32. Enhance perly.fixer to help porters.

    Tim Bunce committed Sep 5, 1997
  33. Emacs CPerl update for 5.004_04

    Ilya Zakharevich committed with Tim Bunce Sep 5, 1997
    private-msgid: 199710140835.EAA26825@monk.mps.ohio-state.edu
  34. myconfig / perl -V: remove randbits and add prototype

    Tim Bunce committed Sep 5, 1997
    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
  35. Updated CPAN.pm for 5.004_04

    Andreas J. Koenig committed with Tim Bunce Sep 5, 1997
    private-msgid: sfcpvpv8teo.fsf@anna.in-berlin.de
Something went wrong with that request. Please try again.