[Debian-med-packaging] Bug#833493: Bug#833493: cufflinks: FTBFS with gcc-6

[Andreas Tille]

Hi Matthia,

On Tue, Aug 16, 2016 at 02:10:05PM +0000, Mattia Rizzolo wrote:
> On Tue, Aug 16, 2016 at 03:53:04PM +0200, Andreas Tille wrote:
> > > Btw, how do you re-import repacked tarball with the same version ?
> > 
> >     git tag -d upstream/<version>
> >     gbp import-orig
> >     git tag -d upstream/<version>
> >     git pull
> > 
> > No idea whether this is the best solution but it works.
> 
> umh, why?
> Then the final tag would point to the old import, when another one
> happened in the meantime, and pristine-tar would be buggy (if that one
> worked fine).

My intention was actually to the old import and not changing the tags to
restore the old history completely.  I imported an md5sum identical
tarball just over the one that was not used and is now in a separate
branch.
 
> If the upstream tarball has been repacked it has to have a different
> version (also because otherwise the archive software is going to be
> crazy).

I just intended to withdraw the import of a repackaged tarball and
restore the original situation.

> Why not just calling the tarball cufflinks_2.2.1+ds.orig.tar.xz
> and import that?

Since the tarball is not repackaged.

> Then you'd have a different upstream version, and
> everything should be happy.
> If you're talking about the double repacking (i.e.
> 3c9d6da6781433b1e47c4fbe9310e64062fe0873 vs
> 4fe2b2e595b10b8f0fbf95b0c686037163d376c0), then my suggestion is to call
> the second one +ds1 instead (i.e., appending a '1').

This would be confusing since +ds1 would just be identical to the
upstream tarball.  I just committed a new Debian revision of an existing
source tarball and wanted to fix the Git repository to enable exactly
this.  May be a sequence of rolled back commits could have lead to the
same result but I regarded this method as way more simple.
 
> BTW, "usually" (I don't recall if this is written anywhere clearly),
> repacking for non-dfsg reasons (like this one) +ds is used instead of
> +dfsg.  But I don't believe anybody cares, anyway :)

I agree that I usually use +dfsg also for non-dfsg reasons which is a
bit sloppy.  My excuse for beeing sloppy that it could also be a
"guideline" to not ship unmaintained code copies.

Hope this explanation makes sense

      Andreas.

-- 
http://fam-tille.de