Bug#661799: FTBFS

[Niko Tyni]

tag 661799 patch
thanks

On Sun, Mar 11, 2012 at 08:41:17PM +0100, gregor herrmann wrote:
> On Sun, 11 Mar 2012 11:19:11 +0100, Salvatore Bonaccorso wrote:
> 
> > > I can reproduce the error, both with the version from the archive and
> > > with the package in git (funnily the first error is different in
> > > both, and also differs from your log).
> > I can confirm now too. It succeeded for me every time in a unstable
> > chroot (sbuild) within a squeeze host. But I can reproduce the build
> > failure under 'plain' unstable.
> 
> Now that's funny - the "host" shouldn't have an influece ...
> 
> And now I built it (from git) on a fast amd64 machine (so far I tried
> only old i386 boxen) -- and now it builds - ?!

I can reproduce it easily by putting some load on the build host
during the test suite. Just a "find /" in another window triggers
it for me.

The problem is in parsing the output of the 'cvs' command, which
sometimes comes in chunks of incomplete lines.

The IPC::Run based loop in Cvs::Command::Base almost handles those,
but breaks when there's a linefeed in the expected regexp, for instance
qr/File: (.*)\s+Status: (.*)\n$/ in Cvs::Command::Status.

Patch attached, this fixes it for me.

BTW, I wonder why is the test suite run under fakeroot (in the 'install'
target).
-- 
Niko Tyni   ntyni at debian.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Properly-handle-incomplete-lines-when-looking-for-a-.patch
Type: text/x-diff
Size: 1066 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-perl-maintainers/attachments/20120327/1be9392a/attachment-0001.patch>