Bug#806262: lintian: fails to find _all.changes for arch:all-only builds (-A)

[Adam D. Barratt]

On Thu, 2015-11-26 at 02:50 +0100, Cyril Brulebois wrote:
> Adam D. Barratt <adam at adam-barratt.org.uk> (2015-11-26):
> > Control: reassign -1 devscripts
> > 
> > On Thu, 2015-11-26 at 00:45 +0100, Cyril Brulebois wrote:
> > > Running a build with the -A flag leads to a confused lintian as it tried
> > > to find a _amd64.changes file where an _all.changes file was produced.
> > > Full transcript follows:
> > > | (sid-amd64-devel)kibi at wodi:~/debian-installer/packages/base-installer$ debuild -A
> > [...]
> > > |  dpkg-genchanges -A >../base-installer_1.158_all.changes
> > > | dpkg-genchanges: binary-only arch-indep upload (source code and arch-specific packages not included)
> > > |  dpkg-source --after-build base-installer
> > > | dpkg-buildpackage: source-only upload: Debian-native package
> > > | Now running lintian...
> > > | warning: "base-installer_1.158_amd64.changes" cannot be processed.
> > > | warning: It is not a valid lab query and it is not an existing file.
> > > | Finished running lintian.
> > 
> > That's a debuild bug. The majority of your output is from
> > dpkg-buildpackage directly, but debuild is then running lintian against
> > the wrong file.
> 
> -ENEEDMOARCOFFEE

:-) I had to double-check by poking at the source that "Now running
$thing" was specifically debuild rather than dpkg-bp.

debuild already understands -A in order to pass it on to relevant
scripts and set flags internally to tell it to run the binary-indep
target under certain circumstances, so afaict in my not-all-that-awake
state it should be quite simple to have it then reflect that in
the .changes name - at least for the now base case where dpkg-bp is
being run directly.

Regards,

Adam