Re: bug versioning and ignored NMUs
* Andreas Barth (email@example.com) [050812 18:46]:
> I noticed that ignoring an NMU can make bugs to "disappear", e.g. bugs
> #321669 and #295270 - as they were reported against NMUed versions, they
> just vanished for unstable.
> Well, what do against this (apart from cluebatting maintainers :) is of
> course a good question. One could just say that a bug reported against a
> version is always open in any newer version as long as it is not fixed
> at all. This is however problematic in two ways:
> * sometimes, bugs are really stable-only, e.g. a broken security upgrade
> * even if the assumption above is true, what happens with the bug if it
> is fixed with some point release. That shouldn't have any effect on
> the bug count of the unstable version.
after some reconsideration, I personally think it would be fair to make
a bug valid in any version if it is found in an earlier version (as
defined by dpkg --versions-compare), and is not resolved in the history
of the package (as defined by its changelog entries), or if it is not
marked as notfound for any version in the history. This would make the
failure on the other side - bugs will appear to exists even if they do
not. But this means, the error is on the safe side.