Re: Howto tackle RC-bugs (and perhaps remove packages)?
On Wed, Jul 23, 2003 at 09:35:37PM +0200, Andreas Barth wrote:
> The RC-bugs (and the packages they belong to) can IMHO be qualified
> into different categories (together with the IMHO appropriate action):
[...]
> 2. RC-bug has an easy-to-use patch for some time, but there was no
> upload in the last time. Ask the developer why, and after two weeks do
> an NMU with the patch.
After two weeks of what, no activity? You'll want to check echelon and
vacation status (only available to DDs) to make sure they're not just off
sunning themselves, and are normally quite responsive.
> 3. Packages with RC-bugs that have also a lot of other bugs and no
> upload in last time, means they are unmaintained. In this case raise a
> bug against the package in asking the maintainer (and possibly
> sponsors) whether package should be orphaned and retitle bug to orphan
> after four weeks (and send copy of orphaning to d-d) unless action is
> taken.
I'd be careful with that criteria - there are a lot of packages (like, say,
dpkg) that have RC bugs, a lot of other bugs, and no uploads since 28/4/03,
but sending an O: to WNPP is likely to raise a few eyebrows...
> 4. (For some time) Orphaned packages with more and/or difficult bugs,
> including an RC-bug. Developers should be asked on d-d whether anyone
> would like to adopt, and if no-one stands up remove package after four
> weeks (that would mean to retitle and reassign to ftp.debian.org).
Arbitrary rules like this tends to catch things which shouldn't be caught.
Was it menu, I think, that was orphaned for months with big bad bugs?
Removal of that would be rather unpleasant (although the argument that no
matter how critical it is, if nobody wants it it must be useless is valid as
far as that goes). Eventually it got taken over by an NM (IIRC, again), but
a 4 week removal would have killed it.
> Is this procedure ok in general? Can I as a non-DD trying an NMU (with
> help of a sponsor of course), and who is ready to sponsor me? Is
> anyone else doing a general tackling of RC-bugs (because I want to
> avoid duplicate work)?
The MIA checks you allude to in (3) are being done by Martin Michlmayr over
time (floods of O's to WNPP). Careful consideration of packages in (4)
should be done, but making up hard rules about it will not work. I'd be
careful about (2), for the reasons I mentioned above.
My personal preference (just getting into Q&A) would be for carefully tested
patches to go into the BTS, followed by an e-mail here bringing attention to
the bug. Giving a complete package, ready to upload, doesn't fill me with
joy, since checking that will probably be more pain. Others may see it
differently, though.
- Matt
Reply to: