Bug#609228: libpadre-plugin-spellcheck-perl: Failing tests; Tries to create .padre directory for building user

[Salvatore Bonaccorso]

Hi Damyan!

On Tue, Jan 11, 2011 at 07:40:19AM +0200, Damyan Ivanov wrote:
> -=| Salvatore Bonaccorso, Mon, Jan 10, 2011 at 09:20:32AM +0100 |=-
> > On Sat, Jan 08, 2011 at 02:27:44PM +0200, Damyan Ivanov wrote:
> > > -=| Salvatore Bonaccorso, Fri, Jan 07, 2011 at 04:48:12PM +0100 |=-
> > > > Source: libpadre-plugin-spellcheck-perl
> > > > Version: 1.1.2-1
> > > > Severity: serious
> > > > Justification: FTBFS
> > > 
> > > Is this with sbuilder settings as on official buildds? Otherwise I'd 
> > > say it is  of severity:importatnt. (worth fixing in anu case)
> > 
> > I talked about that to the release team. We have for all of them with
> > arch:all anyway the squeeze-ignore tag, but they should be fixed at
> > least later on. When Lucas Nussbaum builds the archive, he generally
> > notes too such problems with unwritable homes IIRC. Official buildds
> > seems to have something like:
> > 
> > ---(mount-defaults.buildd)----------------------------------------------
> > # mount.defaults: static file system information for chroots.
> > # Note that the mount point will be prefixed by the chroot path
> > # (CHROOT_PATH)
> > #
> > # <file system> <mount point>   <type>  <options>       <dump>  <pass>
> > proc            /proc           proc    defaults        0       0
> > /dev/pts        /dev/pts        none    rw,bind         0       0
> > tmpfs           /dev/shm        tmpfs   defaults        0       0
> > /home/buildd/build-trees                /build          none    rw,bind         0       0
> > ------------------------------------------------------------------------
> 
> I am still not convinced that this is a problem with the package or 
> a problem that should be solved per package.
> 
> Home directories are a fundamental UNIX thing, similar to mktemp call. 
> For what I guess are safety reasons, sbuild decided not to heep a home 
> directory across builds. This solves the problem of build 
> interference, but as seen it creates problems for somw packages.
> 
> I agree that a fix in the package is simple and quick, but I believe 
> this is better fixed globally, e.g. at sbuild level. How difficult is 
> it to set up a temporary home directory before starting the build and 
> wipe it after (and save many maintainers the job)? Would there be some 
> negative effects?

Well the buildd do not have home for the building user, only the above
mounts shown. One further reason apart the safety reasons, to have
this way is to avoid that building user may use already present
configuration influenting the build or testing (as an example). The
solution you propose to have the buildd's set first a temporary home
directory may work indeed, but has to be implemented there.
Fortunately most of our packages are arch:all packages, so the builds
are workaroundable (thus the Release Team decided to allow them to be
squeeze-ignore, but even with that seen as RC bugs, see reply from
Julien from Release Team yesterday).

Bests
Salvatore
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-perl-maintainers/attachments/20110112/48f3c258/attachment.pgp>