Re: Bits from the Stable Release Managers
On Fri, May 29, 2015 at 11:28:38AM +0200, Cyril Brulebois wrote:
> Wouter Verhelst <wouter@debian.org> (2015-05-27):
> > On Mon, May 25, 2015 at 08:47:43PM +0200, Julien Cristau wrote:
> > > On Mon, May 25, 2015 at 19:42:48 +0100, Simon McVittie wrote:
> > >
> > > > On 25/05/15 18:24, Lisandro Damián Nicanor Pérez Meyer wrote:
> > > > > On Sunday 24 May 2015 21:27:47 Adam D. Barratt wrote:
> > > > > [snip]
> > > > >> Due to the way that the archive manages uploads to proposed-updates, if
> > > > >> you upload a .changes file which only includes source and
> > > > >> architecture:all packages, please ensure that you rename it to something
> > > > >> other than "_$arch.changes".
> > > >
> > > > Does this only apply to source packages that produce at least one binary
> > > > package on $arch, or does it also apply to source packages that solely
> > > > produce architecture-independent binaries?
> > > >
> > > No, the issue is when the maintainer-uploaded .changes file name clashes
> > > with the one from a binary (buildd) upload, so if there aren't any
> > > architecture-dependent binaries, then this doesn't matter.
> >
> > Maybe dak should refuse uploads of a _$arch.changes file that does not
> > refer to any _$arch.deb files with an appropriate error message? Then
> > this isn't an issue.
>
> You're missing _$arch.changes bundling _$arch.udeb.
No, I forgot about udebs in general :-)
The point was, if a .changes file doesn't contain any arch-specific files
(debs, udebs, or tar.gz in the case of dpkg), it shouldn't be called
_$arch.changes (because it isn't).
--
It is easy to love a country that is famous for chocolate and beer
-- Barack Obama, speaking in Brussels, Belgium, 2014-03-26
Reply to: