Re: Bug#243129: NMU in DELAYED/3-DAY
Steve Langasek <email@example.com> writes:
> Demanding that developers make gratuitous uploads to correct for broken
> NMUs is not reasonable. It takes up the maintainer's time, it consumes
> autobuilder resources, it may cause delays for packages trying to get
> into testing (rarely an issue if the upload fixes a legitimate RC bug),
> and it lets people think that buggy or incorrect NMUs are ok -- they're
> NOT. When you do an NMU, you have a responsibility to *get* *it*
Well, I'd say that's true of any package upload, NMU or not.
If an NMU upload is broken, the uploader is obliged to delete the upload
while it's in the delayed queue at the maintainer's request. That
doesn't significantly waste the maintainer's time, and it certainly has
no effect on autobuilder resources.
If the bug really is RC, the maintainer has to prepare a new upload
anyway, so that doesn't really make the it gratuitous nor a waste of the
maintainer's time. The only waste of time and resources I see here is
the public flaming going on.
> An NMU done for a newly opened bug is to be discouraged for other
> reasons, namely, that it's more productive on the whole to NMU for the
> oldest open RC bugs that we know the maintainer *hasn't* taken care of,
> than to NMU for fresh bugs which can probably be addressed more
> efficiently if the maintainer handles them directly.
Sure, but must it be done publicly, unless the NMU'er is a repeat
offender who refuses to get a clue? Publicly flaming someone for an NMU
only serves to discourage more NMU's in the future. I'm certainly wary
of making an NMU, since at best you'll be ignored and at worst you'll be
flamed into oblivion.
NMU's are a Good Thing and should be *encouraged*. Anyone making an NMU
is only trying to reduce the RC count and help Debian release faster.
If an NMU is botched, simply point out the mistake in private mail and
politely ask that more care is used in the future.
You win again, gravity!