On Wed, 16 Oct 2024 at 19:03:11 +0200, Thorsten Alteholz wrote:
> On 15.10.24 20:26, Simon McVittie wrote:
>
> Thanks for doing this removal, but it looks as though all binary packages
> from the gdm3 source were removed as a batch, whether they were NBS or
> not:
>
> Hmm, strange I thought I had adapted the parameters accordingly.
>From some playing around with `dak rm -R -n` on coccia and comparing
its output to your dak-generated mail that closed the bug, I think that
possibly you might have run
dak rm -a armel -B gdm3
(which means: remove every binary package that came from src:gdm3, but
because of the -B, don't remove src:gdm3 itself), whereas I think what
I was asking for is:
dak rm -a armel -b gdm3
(which means: gdm3 is a binary package name, not a source package name)
Does that sound plausible?
Would it be helpful if I tried to construct and quote a suitable `dak rm`
command-line for future removal requests?
If that arrangement works well, it might even be possible for someone
to enhance reportbug so that it asks some questions and then generates
a dak command-line (for the submitter to try with -n on coccia, and for
the ftp team to use without -n to act on the request), the same way that
unblock bugs against release.debian.org generate a ready-to-apply hint.
Looking at `dak rm --help` on coccia, I think perhaps what I should
*actually* have been asking you to do was
dak rm -a armel --outdated gdm3
which I hadn't previously realised was even possible. I'll try to remember
that if we need to drop support for an architecture from a package or
family of packages in future.
smcv
