[NB Not discussing whether the feature should have been enabled by
default, but whether that should be changed now that it _is_ enabled]

On Mon, 2008-04-07 at 20:56 +0200, Frans Pop wrote:
> (Replying to correct cloned BR for second issue.)
> 
> On Monday 07 April 2008, Luk Claes wrote:
> > Frans Pop wrote:
> > > + Set to NMU if packager is neither in Uploaders nor Maintainer
> > >
> > > This has already proven to be incorrect for me today in two cases.
> > > - Team maintained D-I packages are frequently uploaded by a team member
> > >   who's not listed as Uploader, mostly in the case of pre-release
> > >   uploads for translation updates. These uploads are not NMUs.
> >
> > They are as team maintainers who upload should be in Uploaders or change
> > the default if they really don't want to be in Uploaders AFAICS...
> 
> Where is the policy rule that says they _must_ be listed in uploaders?

It's a "should" in policy (5.6.3). However I don't think that's
necessarily an argument against the functionality.

Policy's use of "should" to refer to conventions where non-conformance
is (in general) a (non-wishlist, non-RC) bug doesn't imply that tools
shouldn't default to assuming that those conventions will be followed,
imho. (ugh, too many negatives and too little punctuation).

> The overhead of maintaining that for D-I would be huge for no benefit at 
> all, especially as being listed in uploaders in no longer required for bugs 
> being closed.
> 
> We prefer to have the regular, primary maintainers listed in Uploaders and 
> thus keep relatively usable QA pages for our team members. Other teams has 
> chosen to include everybody and their ponies in Uploaders. Both are valid 
> choices IMO.

They are both valid, of course. I suspect that the d-i team's method is
less common than the alternative method, but would be happy to be proved
wrong.

> > > - I wanted to do a testbuild for  a patch for wireless-tools and ran
> > > dch to increase the version number so I could later run debdiff to get
> > > the patch. As the intention is not to NMU the package, having that set
> > > in the changelog is just silly.
> >
> > No, it's not as you might accidently upload the package. Besides you're
> > changing things in the package as a non-maintainer, so it's a sane
> > default IMHO.
> 
> IMO it is not as it introduces cruft in the changelog I have not asked for.
> The NMU line should only be added when one actually intends to _do_ and NMU.
> I have in the past included changelog changes in a bug I submitted if I 
> considered having a correct changelog entry an important part of the patch. 
> Now I can no longer do that as easily.

Personally, when I receive a patch which includes a changelog stanza
I've never been concerned whether that stanza claims to be for an NMU of
the package or just for a local version. In either case, I'm unlikely to
just apply the changelog change as-is.

There's a clear difference between "here's a patch that I intend to
include in an NMU" and "here's a patch that happens to include a
changelog stanza marking it as an NMU", IMHO.

[...]
> I forsee that the D-I team will see incorrect uploads (NMUs that should be 
> regular uploads) because of this change and therefore request that the old 
> behavior be made default.

Has this actually happened (as yet)?

Cheers,

Adam




-- 
To unsubscribe, send mail to [EMAIL PROTECTED]

Reply via email to