Bug #16980 for Sys-Syslog: Sys::Syslog blows up rather spectacularly on Solaris

Mon Jan 09 18:52:53 2006 Alan.Burlison [...] sun.com - Ticket created

Date:	Mon, 09 Jan 2006 23:51:53 +0000
From:	Alan Burlison <Alan.Burlison [...] sun.com>
To:	Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Sys::Syslog blows up rather spectacularly on Solaris

I'm in the process of upgrading the Sys::Syslog we ship with Solaris in conjunction with applying the sprintf patch, and Sys::Syslog 0.12 from CPAN goes bang bigtime during 'make test' . If I apply the sprintf patch, do I need to bother with updating Sys::Syslog as well? If so, should I use the 0.6 version from the perl 5.8.7 tarball instead of the broken one on CPAN? -- Alan Burlison -- Running make test PERL_DL_NONLAZY=1 /usr/perl5/5.8.4/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t t/00-load......# Testing Sys::Syslog 0.12, Perl 5.008004, /usr/perl5/5.8.4/bin/perl t/00-load......ok t/constants....Use of uninitialized value in pattern match (m//) at t/constants.t line 30. Use of uninitialized value in pattern match (m//) at t/constants.t line 30. Use of uninitialized value in pattern match (m//) at t/constants.t line 30. Use of uninitialized value in pattern match (m//) at t/constants.t line 30. t/constants....ok 4/78 skipped: irrelevant test in this case t/distchk......skipped all skipped: Test::Distribution required for checking distribution t/pod..........skipped all skipped: Test::Pod 1.14 required for testing POD t/podcover.....skipped all skipped: Test::Pod::Coverage 1.06 required for testing POD coverage t/podspell.....skipped all skipped: Test::Spelling required for testing POD spell t/portfs.......skipped all skipped: Test::Portability::Files required for testing filenames portability t/syslog.......ok 1/119unix dgram connect: Socket operation on non-socket at t/syslog.t line 101 getservbyname failed for tcp at t/syslog.t line 101 Use of uninitialized value in subroutine entry at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # Failed test 'openlog() called with facility 'local0' with option 'ndelay'' # in t/syslog.t at line 103. # got: 'Bad arg length for Socket::pack_sockaddr_in, length is 0, should be 4 at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # ' # expected: '' # Failed test 'openlog() should return true: '0'' # in t/syslog.t at line 104. Use of uninitialized value in subroutine entry at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # Failed test 'syslog() called with level 'info'' # in t/syslog.t at line 108. # got: 'Bad arg length for Socket::pack_sockaddr_in, length is 0, should be 4 at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # ' # expected: '' # Failed test 'syslog() should return true: '0'' # in t/syslog.t at line 109. Use of uninitialized value in subroutine entry at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # Failed test 'syslog() called with level 'info'' # in t/syslog.t at line 113. # got: 'Bad arg length for Socket::pack_sockaddr_in, length is 0, should be 4 at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm line 373. # ' # expected: '' # Failed test 'syslog() should return true: '0'' # in t/syslog.t at line 114. # Looks like you failed 6 tests of 119. t/syslog.......dubious Test returned status 6 (wstat 1536, 0x600) DIED. FAILED tests 81-86 Failed 6/119 tests, 94.96% okay (less 38 skipped tests: 75 okay, 63.03%)Failed Test Stat Wstat Total Fail Failed List of Failed ------------------------------------------------------------------------------- t/syslog.t 6 1536 119 6 5.04% 81-86 5 tests and 42 subtests skipped. Failed 1/8 test scripts, 87.50% okay. 6/198 subtests failed, 96.97% okay. *** Error code 6 make: Fatal error: Command failed for target `test_dynamic' /usr/ccs/bin/make test -- NOT OK

Tue Jan 10 07:41:15 2006 Alan.Burlison [...] sun.com - Ticket #16995: Ticket created

Date:	Tue, 10 Jan 2006 12:40:07 +0000
From:	Alan Burlison <Alan.Burlison [...] sun.com>
To:	Alan Burlison <Alan.Burlison [...] sun.com>
CC:	Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris

Alan Burlison wrote: Show quoted text

> I'm in the process of upgrading the Sys::Syslog we ship with Solaris in > conjunction with applying the sprintf patch, and Sys::Syslog 0.12 from > CPAN goes bang bigtime during 'make test' . If I apply the sprintf > patch, do I need to bother with updating Sys::Syslog as well? If so, > should I use the 0.6 version from the perl 5.8.7 tarball instead of the > broken one on CPAN?

maint also blows up: ext/Sys/Syslog/t/syslog...................unix dgram connect: Socket operation on non-socket at ../ext/Sys/Syslog/t/syslog.t line 101 getservbyname failed for tcp at ../ext/Sys/Syslog/t/syslog.t line 101 # Failed test 'openlog() called with facility 'local0' with option 'ndelay'' # in ../ext/Sys/Syslog/t/syslog.t at line 103. # got: 'Bad arg length for Socket::pack_sockaddr_in, length is 0, should be 4 at ../lib/Socket.pm line 373. # ' # expected: '' Broken Pipe FAILED at test 81 maint includes version 0.11 of Sys::Syslog, at this time the latest CPAN version is 0.12, which is what I reported the original problem against. -- Alan Burlison --

Tue Jan 10 08:13:10 2006 gisle [...] ActiveState.com - Ticket #16996: Ticket created

To:	Alan Burlison <Alan.Burlison [...] sun.com>
CC:	Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris
From:	Gisle Aas <gisle [...] ActiveState.com>
Date:	10 Jan 2006 05:12:19 -0800

Alan Burlison <Alan.Burlison@sun.com> writes: Show quoted text

> maint also blows up:

We have now integrated the current maint branch into ActivePerl-816tobe and it does not fail on any of the daily Solaris builds we do (Solaris {2.6,7,8,9} for sparc and Solaris 10 for x86). We do see this kind of noise though: ... ext/Sys/Syslog/t/00-load..................ok ext/Sys/Syslog/t/constants................skipping test on this platform ext/Sys/Syslog/t/distchk..................skipping test on this platform ext/Sys/Syslog/t/podcover.................skipping test on this platform ext/Sys/Syslog/t/podspell.................skipping test on this platform ext/Sys/Syslog/t/pod......................skipping test on this platform ext/Sys/Syslog/t/portfs...................skipping test on this platform ext/Sys/Syslog/t/syslog...................unix dgram connect: Socket operation on non-socket at ../ext/Sys/Syslog/t/syslog. t line 101 getservbyname failed for tcp at ../ext/Sys/Syslog/t/syslog.t line 101 console is not writable at ../ext/Sys/Syslog/t/syslog.t line 101 ok ext/threads/shared/t/0nothread............ok ext/threads/shared/t/av_refs..............ok ... --Gisle

Tue Jan 10 08:16:01 2006 rgarciasuarez [...] gmail.com - Ticket #16997: Ticket created

Date:	Tue, 10 Jan 2006 14:15:24 +0100
From:	Rafael Garcia-Suarez <rgarciasuarez [...] gmail.com>
To:	Gisle Aas <gisle [...] activestate.com>
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris
CC:	Alan Burlison <Alan.Burlison [...] sun.com>, Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org

On 10 Jan 2006 05:12:19 -0800, Gisle Aas <gisle@activestate.com> wrote: Show quoted text

> Alan Burlison <Alan.Burlison@sun.com> writes: >

> > maint also blows up:

> > We have now integrated the current maint branch into > ActivePerl-816tobe and it does not fail on any of the daily Solaris > builds we do (Solaris {2.6,7,8,9} for sparc and Solaris 10 for x86). > We do see this kind of noise though:

I see it too on Linux. That said, I'll remove the optional tests.

Tue Jan 10 08:25:25 2006 rgarciasuarez [...] gmail.com - Ticket #16998: Ticket created

Date:	Tue, 10 Jan 2006 14:24:44 +0100
From:	Rafael Garcia-Suarez <rgarciasuarez [...] gmail.com>
To:	Gisle Aas <gisle [...] activestate.com>
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris
CC:	Alan Burlison <Alan.Burlison [...] sun.com>, Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org

On 10 Jan 2006 05:12:19 -0800, Gisle Aas <gisle@activestate.com> wrote: Show quoted text

> We do see this kind of noise though: > > ... > ext/Sys/Syslog/t/00-load..................ok > ext/Sys/Syslog/t/constants................skipping test on this platform

On second thought, this one should really not be skipped. Is the macros.all file being built, and how do you run tests ? Is PERL_CORE defined in the environment ?

Tue Jan 10 09:09:33 2006 gisle [...] ActiveState.com - Ticket #17000: Ticket created

To:	Rafael Garcia-Suarez <rgarciasuarez [...] gmail.com>
CC:	Alan Burlison <Alan.Burlison [...] sun.com>, Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris
From:	Gisle Aas <gisle [...] ActiveState.com>
Date:	10 Jan 2006 06:08:31 -0800

Rafael Garcia-Suarez <rgarciasuarez@gmail.com> writes: Show quoted text

> On 10 Jan 2006 05:12:19 -0800, Gisle Aas <gisle@activestate.com> wrote:

> > We do see this kind of noise though: > > > > ... > > ext/Sys/Syslog/t/00-load..................ok > > ext/Sys/Syslog/t/constants................skipping test on this platform

> > On second thought, this one should really not be skipped. Is the > macros.all file being built, and how do you run tests ? Is PERL_CORE > defined in the environment ?

It's just the result of a plain 'make test'. I see the same on Linix, this test gets skipped for 'maint' but runs for 'blead'. Probably because 'maint' version doesn't have this in constants.t: my $macrosall = $ENV{PERL_CORE} ? File::Spec->catfile(qw(.. ext Sys Syslog macros.all)) : 'macros.all'; --Gisle

Tue Jan 10 09:58:35 2006 gisle [...] ActiveState.com - Ticket #17001: Ticket created

To:	Alan Burlison <Alan.Burlison [...] sun.com>
CC:	Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris
From:	Gisle Aas <gisle [...] ActiveState.com>
Date:	10 Jan 2006 06:57:46 -0800

Gisle Aas <gisle@ActiveState.com> writes: Show quoted text

> ext/Sys/Syslog/t/syslog...................unix dgram connect: Socket operation on non-socket at ../ext/Sys/Syslog/t/syslog. > t line 101 > getservbyname failed for tcp at ../ext/Sys/Syslog/t/syslog.t line 101 > console is not writable at ../ext/Sys/Syslog/t/syslog.t line 101 > ok

I think it would be an improvement if extra error messages wasn't printed on STDERR when the module croaks. I suggest something like the following patch: ==== //depot/perl/ext/Sys/Syslog/Syslog.pm#38 - /home/gisle/perl/blead/ext/Sys/Syslog/Syslog.pm ==== Index: perl/ext/Sys/Syslog/Syslog.pm --- perl/ext/Sys/Syslog/Syslog.pm.~1~ Tue Jan 10 15:51:39 2006 +++ perl/ext/Sys/Syslog/Syslog.pm Tue Jan 10 15:51:39 2006 @@ -799,10 +799,11 @@ my($old) = select(SYSLOG); $| = 1; select($old); } else { @fallbackMethods = (); + my $msg = "no connection to syslog available"; foreach my $err (@errs) { - carp $err; + $msg .= "\n\t- $err"; } - croak "no connection to syslog available"; + croak $msg; } } End of Patch. This should make the 'make test' output clean as well. --Gisle

Tue Jan 10 10:14:29 2006 nick [...] ccl4.org - Ticket #17002: Ticket created

Date:	Tue, 10 Jan 2006 15:13:44 +0000
From:	Nicholas Clark <nick [...] ccl4.org>
To:	Alan Burlison <Alan.Burlison [...] sun.com>
CC:	Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris

On Mon, Jan 09, 2006 at 11:51:53PM +0000, Alan Burlison wrote: Show quoted text

> I'm in the process of upgrading the Sys::Syslog we ship with Solaris in > conjunction with applying the sprintf patch, and Sys::Syslog 0.12 from > CPAN goes bang bigtime during 'make test' . If I apply the sprintf > patch, do I need to bother with updating Sys::Syslog as well? If so, > should I use the 0.6 version from the perl 5.8.7 tarball instead of the > broken one on CPAN?

Show quoted text

> # Failed test 'openlog() called with facility 'local0' with option > 'ndelay'' > # in t/syslog.t at line 103. > # got: 'Bad arg length for Socket::pack_sockaddr_in, length is > 0, should be 4 at /usr/perl5/5.8.4/lib/i86pc-solaris-64int/Socket.pm > line 373. > # ' > # expected: ''

On IRC we nailed this down to this line failing my $that = sockaddr_in($syslog, inet_aton($host)); because $host didn't resolve, which made us wonder what all the hoop jumping with Sys::Hostname was for. So I got a bit radical, and applied the following which just uses INADDR_LOOPBACK instead. Nicholas Clark Change 26769 by nicholas@nicholas-saigo on 2006/01/10 15:09:07 Why should Syslog jump through hoops to look up the hostname so it can immediately convert it to an IP address, when all it really wants is a connection to the loopback device? Affected files ... ... //depot/perl/ext/Sys/Syslog/Syslog.pm#39 edit Differences ... ==== //depot/perl/ext/Sys/Syslog/Syslog.pm#39 (text) ==== @@ -4,7 +4,7 @@ require 5.006; require Exporter; -our $VERSION = '0.12'; +our $VERSION = '0.13'; our @ISA = qw(Exporter); our %EXPORT_TAGS = ( @@ -42,7 +42,7 @@ my $current_proto = undef; my $failed = undef; my $fail_time = undef; -our ($connected, @fallbackMethods, $syslog_send, $host); +our ($connected, @fallbackMethods, $syslog_send); use Socket ':all'; use POSIX qw(strftime setlocale LC_TIME); @@ -808,11 +808,6 @@ sub connect_tcp { my ($errs) = @_; - unless ($host) { - require Sys::Hostname; - my($host_uniq) = Sys::Hostname::hostname(); - ($host) = $host_uniq =~ /([A-Za-z0-9_.-]+)/; # allow FQDN (inc _) - } my $tcp = getprotobyname('tcp'); if (!defined $tcp) { push(@{$errs}, "getprotobyname failed for tcp"); @@ -826,11 +821,7 @@ } my $this = sockaddr_in($syslog, INADDR_ANY); - my $that = sockaddr_in($syslog, inet_aton($host)); - if (!$that) { - push(@{$errs}, "can't lookup $host"); - return 0; - } + my $that = sockaddr_in($syslog, INADDR_LOOPBACK); if (!socket(SYSLOG,AF_INET,SOCK_STREAM,$tcp)) { push(@{$errs}, "tcp socket: $!"); return 0; @@ -847,11 +838,6 @@ sub connect_udp { my ($errs) = @_; - unless ($host) { - require Sys::Hostname; - my($host_uniq) = Sys::Hostname::hostname(); - ($host) = $host_uniq =~ /([A-Za-z0-9_.-]+)/; # allow FQDN (inc _) - } my $udp = getprotobyname('udp'); if (!defined $udp) { push(@{$errs}, "getprotobyname failed for udp"); @@ -863,11 +849,7 @@ return 0; } my $this = sockaddr_in($syslog, INADDR_ANY); - my $that = sockaddr_in($syslog, inet_aton($host)); - if (!$that) { - push(@{$errs}, "can't lookup $host"); - return 0; - } + my $that = sockaddr_in($syslog, INADDR_LOOPBACK); if (!socket(SYSLOG,AF_INET,SOCK_DGRAM,$udp)) { push(@{$errs}, "udp socket: $!"); return 0;

Tue Jan 10 10:52:48 2006 Alan.Burlison [...] sun.com - Ticket #17004: Ticket created

Date:	Tue, 10 Jan 2006 15:51:42 +0000
From:	Alan Burlison <Alan.Burlison [...] sun.com>
To:	Gisle Aas <gisle [...] ActiveState.com>
CC:	Rafael Garcia-Suarez <rgarciasuarez [...] gmail.com>, Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris

Gisle Aas wrote: Show quoted text

> It's just the result of a plain 'make test'. I see the same on Linix, > this test gets skipped for 'maint' but runs for 'blead'. Probably > because 'maint' version doesn't have this in constants.t: > > my $macrosall = $ENV{PERL_CORE} ? File::Spec->catfile(qw(.. ext Sys Syslog macros.all)) > : 'macros.all';

I've fixed this by checking the macro actually exists before fetching it's value - now I get this: t/constants....ok 4/78 skipped: irrelevant test in this case instead of a load of uninitialised variable warnings. I've also fixed a seperate bug where Sys::Syslog assumes that the path returned by _PATH_LOG() is a AF_UNIX socket and then tries to open/connect to it. -- Alan Burlison --

*** Syslog.pm.orig Tue Jan 10 15:32:20 2006 --- Syslog.pm Tue Jan 10 15:33:35 2006 *************** *** 894,899 **** --- 894,903 ---- push(@{$errs}, "_PATH_LOG not available in syslog.h"); return 0; } + if (! -S $syslog_path) { + push(@{$errs}, "$syslog_path is not a socket"); + return 0; + } my $that = sockaddr_un($syslog_path); if (!$that) { push(@{$errs}, "can't locate $syslog_path"); *** t/constants.t.orig Tue Jan 10 14:48:34 2006 --- t/constants.t Tue Jan 10 15:10:24 2006 *************** *** 27,33 **** $name = $1; my $v = eval "${callpack}::$name()"; ! if($v =~ /^\d+$/) { is( $@, '', "calling the constant $name as a function" ); like( $v, '/^\d+$/', "checking that $name is a number ($v)" ); --- 27,33 ---- $name = $1; my $v = eval "${callpack}::$name()"; ! if(defined($v) && $v =~ /^\d+$/) { is( $@, '', "calling the constant $name as a function" ); like( $v, '/^\d+$/', "checking that $name is a number ($v)" );

Tue Jan 10 12:00:54 2006 nick [...] ccl4.org - Ticket #17007: Ticket created

Date:	Tue, 10 Jan 2006 17:00:11 +0000
From:	Nicholas Clark <nick [...] ccl4.org>
To:	Alan Burlison <Alan.Burlison [...] sun.com>
CC:	Gisle Aas <gisle [...] ActiveState.com>, Rafael Garcia-Suarez <rgarciasuarez [...] gmail.com>, Perl 5 Porters <perl5-porters [...] perl.org>, maddingue [...] free.fr, bug-sys-syslog [...] rt.cpan.org
Subject:	Re: Sys::Syslog blows up rather spectacularly on Solaris

On Tue, Jan 10, 2006 at 03:51:42PM +0000, Alan Burlison wrote: Show quoted text

> Gisle Aas wrote: >

> >It's just the result of a plain 'make test'. I see the same on Linix, > >this test gets skipped for 'maint' but runs for 'blead'. Probably > >because 'maint' version doesn't have this in constants.t: > > > > my $macrosall = $ENV{PERL_CORE} ? File::Spec->catfile(qw(.. ext Sys > > Syslog macros.all)) > > : 'macros.all';

> > I've fixed this by checking the macro actually exists before fetching > it's value - now I get this: > > t/constants....ok > 4/78 skipped: irrelevant test in this case > > instead of a load of uninitialised variable warnings. > > I've also fixed a seperate bug where Sys::Syslog assumes that the path > returned by _PATH_LOG() is a AF_UNIX socket and then tries to > open/connect to it.

Thanks, applied (change 26773) Nicholas Clark

Tue Jan 10 12:18:49 2006 SAPER [...] cpan.org - Ticket #16995: Ticket 16995 MergedInto ticket 16980.

Tue Jan 10 12:55:50 2006 SAPER [...] cpan.org - Status changed from 'new' to 'open'

Tue Jan 10 12:55:56 2006 SAPER [...] cpan.org - Severity Normal added

Tue Jan 10 13:24:55 2006 maddingue [...] free.fr - Correspondence added

Date:	Tue, 10 Jan 2006 19:24:07 +0100
From:	Sébastien Aperghis-Tramoni <maddingue [...] free.fr>
To:	bug-Sys-Syslog [...] rt.cpan.org
Subject:	Re: [cpan #16980] Re: Sys::Syslog blows up rather spectacularly on Solaris
RT-Send-Cc:

Alan Burlison wrote: Show quoted text

> > I'm in the process of upgrading the Sys::Syslog we ship with Solaris in > > conjunction with applying the sprintf patch, and Sys::Syslog 0.12 from > > CPAN goes bang bigtime during 'make test' . If I apply the sprintf > > patch, do I need to bother with updating Sys::Syslog as well? If so, > > should I use the 0.6 version from the perl 5.8.7 tarball instead of the > > broken one on CPAN?

> > maint also blows up: > > ext/Sys/Syslog/t/syslog...................unix dgram connect: Socket > operation on non-socket at ../ext/Sys/Syslog/t/syslog.t line 101 > getservbyname failed for tcp at ../ext/Sys/Syslog/t/syslog.t line 101 > # Failed test 'openlog() called with facility 'local0' with option > 'ndelay'' > # in ../ext/Sys/Syslog/t/syslog.t at line 103. > # got: 'Bad arg length for Socket::pack_sockaddr_in, length is > 0, should be 4 at ../lib/Socket.pm line 373. > # ' > # expected: '' > Broken Pipe > FAILED at test 81 > > maint includes version 0.11 of Sys::Syslog, at this time the latest CPAN > version is 0.12, which is what I reported the original problem against.

I saw a CPAN Testers report about this at http://www.nntp.perl.org/group/perl.cpan.testers/279265 and IIRC, Abe Timmerman pasted me a similar result on a IRIX system, but with the precision that its network was badly configured. As I was unable to reproduce this problem on any of the systems I can test on, I ignored it. -- Sébastien Aperghis-Tramoni Close the world, txEn eht nepO.

Tue Jan 10 13:30:52 2006 maddingue [...] free.fr - Correspondence added

Date:	Tue, 10 Jan 2006 19:30:08 +0100
From:	Sébastien Aperghis-Tramoni <maddingue [...] free.fr>
To:	bug-Sys-Syslog [...] rt.cpan.org
CC:	Perl 5 Porters <perl5-porters [...] perl.org>
Subject:	Re: [cpan #16980] Re: Sys::Syslog blows up rather spectacularly on Solaris
RT-Send-Cc:

Gisle Aas <gisle@ActiveState.com> wrote: Show quoted text

> I think it would be an improvement if extra error messages wasn't > printed on STDERR when the module croaks. I suggest something like > the following patch:

[...] Show quoted text

> This should make the 'make test' output clean as well.

Was this patch applied ? As these messages seem to bother people, I'll probably apply it for the CPAN version. -- Sébastien Aperghis-Tramoni Close the world, txEn eht nepO.

Tue Jan 10 13:32:05 2006 gisle [...] ActiveState.com - Correspondence added

To:	Sébastien Aperghis-Tramoni <maddingue [...] free.fr>
CC:	bug-Sys-Syslog [...] rt.cpan.org, Perl 5 Porters <perl5-porters [...] perl.org>
Subject:	Re: [cpan #16980] Re: Sys::Syslog blows up rather spectacularly on Solaris
From:	Gisle Aas <gisle [...] ActiveState.com>
Date:	10 Jan 2006 10:31:16 -0800
RT-Send-Cc:

Sébastien Aperghis-Tramoni <maddingue@free.fr> writes: Show quoted text

> Gisle Aas <gisle@ActiveState.com> wrote: >

> > I think it would be an improvement if extra error messages wasn't > > printed on STDERR when the module croaks. I suggest something like > > the following patch:

> [...]

> > This should make the 'make test' output clean as well.

> > Was this patch applied ?

No. Show quoted text

> As these messages seem to bother people, I'll probably apply it for > the CPAN version.

I'm all for it :) --Gisle

Tue Jan 10 19:46:56 2006 maddingue [...] free.fr - Correspondence added

Date:	Wed, 11 Jan 2006 01:46:15 +0100
From:	Sébastien Aperghis-Tramoni <maddingue [...] free.fr>
To:	bug-Sys-Syslog [...] rt.cpan.org
CC:	Perl 5 Porters <perl5-porters [...] perl.org>
Subject:	Re: [cpan #16980] Re: Sys::Syslog blows up rather spectacularly on Solaris
RT-Send-Cc:

Show quoted text

> > As these messages seem to bother people, I'll probably apply it for > > the CPAN version.

> > I'm all for it :)

I applied it then optimized it with a join(). -- Sébastien Aperghis-Tramoni Close the world, txEn eht nepO.

Tue Jan 10 19:51:56 2006 maddingue [...] free.fr - Correspondence added

Date:	Wed, 11 Jan 2006 01:51:15 +0100
From:	Sébastien Aperghis-Tramoni <maddingue [...] free.fr>
To:	bug-Sys-Syslog [...] rt.cpan.org
CC:	Perl 5 Porters <perl5-porters [...] perl.org>
Subject:	Re: [cpan #16980] Re: Sys::Syslog blows up rather spectacularly on Solaris
RT-Send-Cc:

Nicholas Clark wrote: Show quoted text

> Thanks, applied (change 26773)

Merged CPAN version with changes #26768, #26769, #26772 and #26773 Will be released as 0.13 -- Sébastien Aperghis-Tramoni Close the world, txEn eht nepO.

Wed Jan 11 10:12:05 2006 SAPER [...] cpan.org - Taken

Wed Jan 11 10:12:33 2006 SAPER [...] cpan.org - Fixed in 0.13 added

Wed Jan 11 10:12:33 2006 SAPER [...] cpan.org - Status changed from 'open' to 'resolved'

Mon May 22 19:49:26 2006 SAPER [...] cpan.org - Reference by ticket #19418 added

Thu Jun 01 19:27:40 2006 SAPER [...] cpan.org - Reference by ticket #19622 added