[pkg-GD-devel] Bug#446290: Bug#446290: *.orig files from upstream interact badly with dh_clean

[Ian Jackson]

Jonas Smedegaard writes ("Re: [pkg-GD-devel] Bug#446290: *.orig files from upstream interact  badly with dh_clean"):
> You file this bugreport against the version in stable Debian, the latest
> version in unstable does not have the reported problem. If I am not
> confusing you with someone else, you very well know that for stable
> software the only allowed changes are security-related bugfixes.

Indeed so.

As it happens this source package will have been copied out of sid by
Ubuntu into Ubuntu's current development release about 5 months ago.
Ubuntu is nearing the end of its release cycle and will merge changes
from sid again in about another month.  In the case of this particular
package there were no changes made by Ubuntu.

If I find, as part of my paid work on Ubuntu, a bug in the Ubuntu
package which I consider is very likely to affect the Debian package
as well, and where the same fix should be made to both branches, I
typically report the bug to Debian.  This seems no different to the
situation where I discover a bug in Debian stable on one of my
personal systems.

> But in this case, the 2 *.orig files seems to be simple left-overs of
> earlier versions of files. Yes, they are mentioned in MANIFEST but seems
> to do no harm at all if missing. I suspect noone will ever benefit from
> keeping this bugreport open in the system.

I found that the change in the patch was necessary to stop the build
failing, complaining about a missing file.  See the build log in this
(automatically filed) bug report against Ubuntu:
 https://launchpad.net/bugs/144495

Note the line at the bottom where make looks for
Graph/axestype.pm.orig.  I didn't inspect the Makefiles in detail but
I assumed that the error was because the .pm is generated somehow from
the .pm.orig (upstream having chosen a poor naming scheme).

It is of course possible that upstream have (sensibly) changed the
name so that the bug doesn't appear in sid any more, in which case
it would be right to mark the bug as fixed in the relevant version.

Or perhaps I am mistaken, of course :-).

Regards,
Ian.