Re: Removing more packages from unstable
* Scott Kitterman <debian@kitterman.com> [240820 14:35]:
> On August 20, 2024 12:16:47 PM UTC, Andrey Rakhmatullin <wrar@debian.org> wrote:
> >On Tue, Aug 20, 2024 at 12:12:33PM +0000, Scott Kitterman wrote:
> >> There are people doing this, we could use more, but it does
> >> happen. I've processed lots of these and it's virtually always
> >> fine. In the rare case of a mistake, the cost to rectify the
> >> mistake is a trip through New.
> >>
> >> I don't think we need more process.
> >
> >Oh, I'm sure it's fine both for people filing these and the FTP team, I'm
> >worried about reactions from the maintainers of those packages.
> >
> For those cases, the people who have been doing this will
> sometimes file a bug against the package as a heads up and then as
> for the removal a bit later. Of course I don't ever see the ones
> where the maintainer objects and nothing further comes of it, but
> my impression is it's rare.
I have been trying to do this, with varying levels of success. When
I tried this, my process often was:
- file bug against the package in unstable, with a title of "RM?
<package> <reason>". Have some text, indicating that I'll wait a
month and then reassign to ftp.debian.org.
The text was often ad-hoc and can certainly be improved.
- Wait a month (or often more, because I did not keep track of these
packages systematically)
- Clone or reassign to ftp.debian.org, in the process retitling and
fixing up the metadata.
In more recent attempts, I tried:
- file RM bug against ftp.debian.org, immediately tagged moreinfo
and indicate to wait a month before untagging.
- Wait a month (or often more, because I did not keep track of these
packages systematically)
- Untag
The newer process is to me clearly superior because it's less work,
and IMO has the same chance of the maintainers reacting to it.
Hand-crafting bug metadata (when reassigning) for ftp.debian.org is
easy to mess up, so I'd rather skip that.
> [..]
> For most of the packages that fall into this category, I don't think maintainer reaction is a major issue.
I don't have any numbers of how often I've attempted these, but for
both versions I've heard back from one maintainer each in total.
However I want to point out that often I picked packages that saw no
maintainer uploads for many years (think 5 or more).
I suspect that many maintainers do not actually receive bug mails at
all.
For the usrmerge NMUs (all with 10 day delay and nmudiff mail etc),
a lot of maintainers were surprised to find out about the bug via
the ACCEPT mail.
Chris
Reply to: