Le 11 juillet 2011 16:20:38 Philip Martin a écrit :
> Philip Martin writes:
> > However this object code is dynamically loaded by Perl at runtime so
> > it's not impossible that the Perl compiler flags need to be applied. It
> > seems to me that Subversion is responsible for ensuring that this is
On Mon, Jul 11, 2011 at 04:20:38PM +0100, Philip Martin wrote:
> Philip Martin writes:
>
> > However this object code is dynamically loaded by Perl at runtime so
> > it's not impossible that the Perl compiler flags need to be applied. It
> > seems to me that Subversion is responsible for ensurin
Philip Martin writes:
> However this object code is dynamically loaded by Perl at runtime so
> it's not impossible that the Perl compiler flags need to be applied. It
> seems to me that Subversion is responsible for ensuring that this is
> done and that we should be fixing this in Subversion.
I
Stéphane Gaudreault writes:
> I think the problem may be caused by the absence of -D_FILE_OFFSET_BITS=64 in
> the subversion cflags. I do not get the failure in the "make test" if I do
> export CFLAGS+=" -D_FILE_OFFSET_BITS=64"
> before building subversion.
I don't know whether we can dismis
Le 7 juillet 2011 09:38:08, Philip Martin a écrit :
> Stéphane Gaudreault writes:
> > Le 7 juillet 2011 08:43:06, Philip Martin a écrit :
> >> If you change the line in configure from
> >>
> >> SWIG_PL_INCLUDES="\$(SWIG_INCLUDES) `$PERL -MExtUtils::Embed -e
> >>
> >> ccopts`"
> >>
> >> to
Stéphane Gaudreault writes:
> Le 7 juillet 2011 08:43:06, Philip Martin a écrit :
>> If you change the line in configure from
>>
>> SWIG_PL_INCLUDES="\$(SWIG_INCLUDES) `$PERL -MExtUtils::Embed -e
>> ccopts`"
>>
>> to
>>
>> SWIG_PL_INCLUDES="\$(SWIG_INCLUDES) `unset CFLAGS; $PERL
>>
Le 7 juillet 2011 08:43:06, Philip Martin a écrit :
> Stéphane Gaudreault writes:
> > # perl -MExtUtils::Embed -e ccopts
> >
> > -D_REENTRANT -D_GNU_SOURCE -fno-strict-aliasing -pipe -fstack-protector
> > -I/usr/local/include -D_LARGEFILE_SOURCE -
> >
> > D_FILE_OFFSET_BITS=64 -I/usr/lib/perl
Stéphane Gaudreault writes:
> # perl -MExtUtils::Embed -e ccopts
> -D_REENTRANT -D_GNU_SOURCE -fno-strict-aliasing -pipe -fstack-protector
> -I/usr/local/include -D_LARGEFILE_SOURCE -
> D_FILE_OFFSET_BITS=64 -I/usr/lib/perl5/core_perl/CORE
>
> It seems that the new perl 5.14 ExtUtils::MakeMake
Le 7 juillet 2011 07:27:05, Philip Martin a écrit :
> Stéphane Gaudreault writes:
> >> What is the typdef line for apr_off_t in apr.h? It should be something
> >> that is 64bit irrespective of _FILE_OFFSET_BITS.
> >
> > typedef off64_t apr_off_t;
>
> So APR/Subversion are using 64-of
Stéphane Gaudreault writes:
>> What is the typdef line for apr_off_t in apr.h? It should be something
>> that is 64bit irrespective of _FILE_OFFSET_BITS.
>
> typedef off64_t apr_off_t;
So APR/Subversion are using 64-offsets irrespective of
_FILE_OFFSET_BITS.
>> Are there any occuren
Le 6 juillet 2011 05:42:15, Philip Martin a écrit :
> Stéphane Gaudreault writes:
> > # apr-1-config --cflags
> >
> > -pthread
> >
> > # apr-1-config --cppflags
> >
> > -DLINUX=2 -D_REENTRANT -D_GNU_SOURCE -D_LARGEFILE64_SOURCE
>
> So that is the standard APR with large-file support. APR d
Stéphane Gaudreault writes:
> # apr-1-config --cflags
> -pthread
>
> # apr-1-config --cppflags
> -DLINUX=2 -D_REENTRANT -D_GNU_SOURCE -D_LARGEFILE64_SOURCE
So that is the standard APR with large-file support. APR doesn't use
-D_FILE_OFFSET_BITS=64 because it defines its own 64bit apr_off_t a
Le 5 juillet 2011 15:51:58, Philip Martin a écrit :
> Stefan Sperling writes:
> > This pretty much confirms the suspicion that this is a packaging
> > problem, as discussed here:
> > http://svn.haxx.se/users/archive-2011-06/0261.shtml
> >
> > Please fix the packaging of Perl and Subversion in Arc
Stefan Sperling writes:
> This pretty much confirms the suspicion that this is a packaging
> problem, as discussed here:
> http://svn.haxx.se/users/archive-2011-06/0261.shtml
>
> Please fix the packaging of Perl and Subversion in Arch Linux.
> Other affected distributions should do the same.
Sub
On Tue, Jul 05, 2011 at 03:27:51PM -0400, Stéphane Gaudreault wrote:
> Compilation is not failing. The bindings fail at runtime on i686, but works
> on
> x86_64.
>
> Here are the perl and subversion cflags on i686 :
>
> subversion: -march=i686 -mtune=generic -O2 -pipe -pthread -
> D_LARGEFILE
Le 5 juillet 2011 04:24:11, Philip Martin a écrit :
> Stéphane Gaudreault writes:
> > Here is a patch that fix compilation of the perl bindings on i686 :
> Is the compilation failing? I think the compilation works, the problem
> is the bindings fail at runtime.
>
> > https://bugs.archlinux.org/t
Stéphane Gaudreault writes:
> Here is a patch that fix compilation of the perl bindings on i686 :
Is the compilation failing? I think the compilation works, the problem
is the bindings fail at runtime.
> https://bugs.archlinux.org/task/24540
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=6
Here is a patch that fix compilation of the perl bindings on i686 :
https://bugs.archlinux.org/task/24540
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=628507
This problem was also discussed on the user mailing list.
Regards,
--
Stéphane Gaudreault
ArchLinux developer
==
18 matches
Mail list logo
