[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 0-day NMUs, DELAYED/n uploads

On Wed, Oct 17, 2007 at 06:03:14PM +0200, Lucas Nussbaum wrote:
> Hi,


> This thread started on -mentors@. The process that should be followed for
> 0-day NMUs is being discussed. (reminder: it's allowed to do 0-day NMUs
> for RC bugs and release goals, for bugs older than 7 days).
> Please continue the discussion on -devel@, as it's of interest to
> everybody.
> On 17/10/07 at 07:51 -0700, Richard Hecker wrote:
> >>> If a person cannot communicate with some email, being a 'lone wolf'
> >>> submitting NMUs will not benefit the project in the long term.  The
> >>> NMU does not replace communication skills.
> >>>     
> >>
> >> Mails were sent to the relevant bugs on the BTS (and thus to the
> >> maintainer(s)). Do you need a personal email, because you filter out BTS
> >> mails? Then we probably have another problem.

Mailing the BTS (with the NMU patch) when you do an NMU is indeed still

> > Yes, we do have a problem. From section 5.11.1 of our
> > developers reference (http://www.debian.org/doc/developers-reference),
> > the admonition is to "contact the developer first, and act later." It
> > appears to me that people do not even understand what the usual
> > rules are for NMUs.
> When dealing with a bug, it is painful to do an action at one point (notifying
> the maintainer), and to have to remember that another action needs to be
> done later (do the upload). The DELAYED queue helps a lot with that by
> making it possible to do both actions at the same time, but still leave
> some time for the maintainer to react.

Indeed that's a possibility.

> The following processes seem appropriate to me:
> - for bugs older than 7 days:
>   - send a mail to the BTS with the NMU diff
>   - at the same time, upload directly to unstable, or to DELAYED/1, 2,
>   or 3, depending on the package (is it a very popular package?), the
>   maintainer (is he active?) and the patch (am I 100% it won't break, or
>   only 99%?)

That's ok, though when in doubt it's also ok to first contact the

> - for bugs newer than 7 days:
>   - send a mail to the BTS with the NMU diff
>   - at the same time, upload to DELAYED/(7-age), or to
>     DELAYED/(7-age+1,2,3), depending on the same conditions as above

This is not so clear cut: If in doubt, please don't do this. This can be
warranted though for some kind of bugs. In general I would propose to
focus on bugs that are older than 7 days. Personally I would start with
bugs older than 10 or 14 days and only look at more recent ones later...

> But it seems that some maintainers disagree. Is there a consensus that
> those processes are OK? If not, what could we do, without making it too
> painful for the NMUer?

I think we can say there is consensus for bugs older than 7 days. There
is also no real problem with the process for bugs newer than that, but I
would advise to focus on the first category.

The most important thing is of course to respect the maintainer and
please try to make sure your NMU is well tested.



Reply to: