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

Re: NMUing generally

+++ Holger Levsen [2014-07-16 15:17 +0200]:
> Hi Riku,
> On Mittwoch, 16. Juli 2014, Riku Voipio wrote:
> > This a side-effect from Debian's policy that discourages fixing bugs in
> > other maintainers packages via NMUs. Somehow, it is felt better to
> > remove packages from testing than possibly offending the maintainer with
> > an unwarranted NMU.
> Huh, as I see it, this has changed and a.) NMUs to fix RC bugs are generally 
> very much encouraged and b.) NMUs to fix other bugs are also perceived much 
> better nowadays than "in some old days".
> Especially if the NMU is properly done (eg documented in the BTS, uploaded to 
> DELAYED-foo, etc) I've hardly seen any complaints in recent years.

Mostly. I've done a pile recently (40+) (largely autoconf-update bugs)
and only had one instance of outright hostility, a few queries, a few
pre-empting uploads by the maintainer, and a couple of 'thanks, please
upload ASAP'. The majority just went through with no explicit
response. That was almost all packages that have had a bug filed for
between 5 years and a couple of months, where an NMU seems
unarguable. So that does mean we are down to less than 2% obnoxious
maintainers :-)

I'm now getting on to ones where I file a new NMU bug along with the
delayed upload, and that feels a bit 'pushier'. We'll see how that
goes. I still feel like I need to make excuses for not filing a bug,
waiting a couple of weeks and _then_ doing an NMU, which of course is
a great way of adding a pile of fairly pointless delay.

BTW is there a reason why nmudiff doesn't actually do the upload, as
well as filing the bug? It's very easy to forget the upload and thus
confuse the maintainer by saying 'I have uploaded foo to delayed/X'
when in fact you've not.

It could easily enough run dupload or dput at the end...

I do think there is plenty more room for soialising more NMUing. There
are piles of bugs in the archive that mostly just need uploading, and
many packages with a backlog of minor bugs, or packaging that frankly
just needs updating. But a lot of this quickly gets into 'updating'
and 'de-crufting' rather than simply minimally-bug-fixing. I suspect
that would see more pushback if one did much of it.

Principal hats:  Linaro, Emdebian, Wookware, Balloonboard, ARM

Reply to: