Re: RFS: libonig NMU for a RC bug
Loïc Minier schrieb am Sonntag, den 05. August 2007:
> Hi Alexander,
>
> On Sun, Aug 05, 2007, Alexander Wirt wrote:
> > Uploading such an NMU is NOT okay. If you ever intent to upload such a crap
> > again, use the delayed queue.
>
> Please explain what part of this NMU was "crap". I see no new bugs
> were filed after the upload.
It is crap in a way that it is not like an NMU should be.
Please read
http://www.us.debian.org/doc/manuals/developers-reference/ch-pkgs.en.html#s-nmu-guidelines
For example:
Make sure that the package's bugs that the NMU is meant to address are all
filed in the Debian Bug Tracking System (BTS). If they are not, submit them
immediately.
or:
Upload your package to incoming in DELAYED/7-day (cf. Delayed
uploads, Section 5.6.2), send the final patch to the maintainer via
the BTS, and explain to them that they have 7 days to react if they
want to cancel the NMU.
So I still consider this as crap.
>
> > And please only NMU the bug and not other stuff in the package. If you
> > have problems with them, open a bug. But don't do this again.
>
> While you're technically correct, I do think there's some latitude
> between NMUs and QA uploads:
> - last upload of the package 6 months ago -- despite the new upstream
> releases
> - RC bug since two months, with no comment from any of the two
> maintainers (this includes you) and which had a fix (new upstream)
I'm just the sponsor and wasn't aware of this bug. First time I heard about
the NMU was yesterday from the mentors system. I already ping the maintainer
who was on holidays til this weekend. I would have reacted tomorrow.
> Full disclosure: I explained my exact position on the subject to my
> sponsoree; French IRC conversation follows:
I don't speak french, so without a translation this is useless to me.
> My understanding is that my sponsoree did his best to do what is
> usually requested to NMs: produce a lintian clean package, and AFAICT,
> his changes were correct and fixed lintian warnings.
No, there is a reason why the stated guidelines exist.
Don't EVER do ANY package changes in an NMU if they are not bug. Linitan
warnings are no bugs, some of them are, some not. If its a bug, fill one.
But don't do such NMUs.
He did fundamental changes to the package. This is not something that should
ever be done in an nmu.
> If you did notice any regression after this result, could you please
> document them? I should be responsible for tracking their resolution,
> and as you can see, I took full responsability of the upload by
> subscribing to its PTS feeds and requiring my sponsoree to do so as
> well.
This is not a point of regression, but of following guidelines. They exists
to get used, not to get ignored.
Alex
Reply to: