Re: Please consider migrating kernel-package to testing ...

[Sven Luther <sven.luther@wanadoo.fr>]

On Sun, Nov 26, 2006 at 11:28:59PM -0800, Steve Langasek wrote:
> On Thu, Nov 23, 2006 at 08:16:25PM +0100, Sven Luther wrote:
> 
> > Please consider hinting kernel-package to move into testing. The current
> > version in testing is still victim of 391643, which made the built kernel
> > packages uninstallable with : 
> 
> >   Missing Required paramater 'Old' at
> >   /var/lib/dpkg/info/linux-image-2.6.19-rc5-efika1.postinst line 393.
> 
> > This was fixed in 10.063, but there may be other changes which went into
> > 10.064 and 10.065, but as is the etch kernel-package is unusable, which would
> > warrant a grave RC bug against it.
> 
> Hint added.

Thanks,

> > I guess you missed this because of the funny versioned thingy ? What would be
> > the best way to handle such things, add a "found in 10.062" in the bug report
> > and re-open it ? 
> 
> If you're referring to bug #391643, this bug is listed at severity:
> important.  Reopening the bug report is absolutely wrong here, that would
> lose the information that it was fixed in 10.063.

Yeah, the problem here, is that it was cloned all over without real
uunderstanding of the matter by Frans, and that Manoj fixed it without telling
the other sub-bugs, and then took a really agressive stance to me when i
failed to notice, and cloned the bug and bumped the severity, so the
information got lost.

I believe strongly that post-etch, we need to follow on with jonas proposal,
and externalize the kernel post/pre inst/rm scripts into a separate package,
and have only minimal support in the kernel packages. Will make the
responsability much easier, but then Manoj cooperating more with the kernel
team would also help. But then i am the devil personified for mentioning that
Manoj's behaviour may not be the most ideal stuff in this, so ...

Friendly,

Sven Luther