On Sun, May 13, 2012 at 08:57:34PM +0200, Daniel Dehennin wrote:
> Guido Günther <a...@sigxcpu.org> writes:
> 
> 
> [...]
> 
> >> -                # FIXME: When creating a new package: find merge base 
> >> between HEAD and origin
> >> -                since = repo.find_version(options.debian_tag, 
> >> cp['Version'])
> >> +                # Take care of newly created debian/changelog
> >> +                if hasattr(cp, 'first') and cp.first:
> >
> > Hmm...where is first being set? I cloned
> >
> >  git://git.baby-gnu.org/git-buildpackage 
> > tags/dad/create-inexistant-changelog/rebasable/on-f495df9-1
> >
> > and couldn't find it at a first glance.
> 
> I did it in gbp.deb.Changelog.create.
> 
> I was thinking about adding a "first=Fales" parameter to
> Changelog.__init__ and make Changelog.create using True.
> 
> But this seems to be a "private" attribute, I don't like the idea of
> exposing it in __init__ signature.

Can you point me to your tree that uses the first attribute?

> 
> Maybe doing "self._first = False" in Changelog.__init__ and
> Changelog.create set it to True is a better idea, no need call hasattr.
> 
> >
> >
> >> +                    pattern = options.upstream_tag % dict(version='*')
> >> +                    try:
> >> +                        upstream = repo.find_tag('HEAD', pattern=pattern)
> >> +                    except GitRepositoryError:
> >> +                        gbp.debug('No upstream tag found')
> >> +                        upstream = options.upstream_branch
> >> +                    if options.upstream_tree == 'branch':
> >> +                        upstream = options.upstream_branch
> >> +                    since = repo.get_merge_base('HEAD', upstream)
> >
> > Need to catch exception if no merge base is found.
> 
> I'm not sure, if we don't find a merge base there is nothing we can do.
> 
> The user may set a wrong upstream-branch name in its gbp.conf, I think
> the most significant thing to do is to report the
> "GitRepositoryError: Failed to get common ancestor: fatal: Not a valid object 
> name XXXX"
O.k. then. Just wanted to make sure you don't want to fall back to other
heuristics in that case.
Cheers,
 -- Guido



--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to