[Pkg-zsh-devel] Differences between git repo and official tar balls: Are they still relevant?

[Axel Beckert]

Hi Daniel,

Daniel Shahaf wrote:
> Axel Beckert wrote on Mon, Nov 30, 2015 at 03:04:36 +0100:
> > now that pws started signing upstream tar balls and I implemented
> > verifying these signatures in
> > https://anonscm.debian.org/cgit/collab-maint/zsh.git/commit/?id=e6d23bf3
> > we should have a look if we still need to generate fake upstream tar
> > balls from git.
> ...
> > What do you think?
> 
> I'm curious: wouldn't it be simpler to teach git-orig-source to verify
> the signed git tag (and continue using a 'git archive' of the upstream
> tag) than to maintain a list of all the generated files that need to be
> moved away and back again?  What exactly is gained by using the upstream
> tarball (over using the signed tag)?

Actually being able to use the upstream tarball is the goal. :-)

It's preferred in Debian to use upstream tar balls whereever possible.
Not having to generate a fake .orig.tar.xz would be one gain, also
less work for us when packaging a new upstream release.

IIRC the main reason for not using the zsh upstream tar ball was that

* we don't want to use pre-generated documentation due to different
  paths than we passed to ./configure, but
* missing files to regenerate this documentation with the proper
  paths.

The latter issue seems gone.

Another side-story is that Ubuntu e.g. rapes our zsh package and uses
the upstream tar ball to not have to regenerate the documentation
(because they don't want to maintain the yodl package in main while
they want zsh in main -- long story, many bug reports on Launchpad).

Minimizing the diff to Ubuntu would be another gain.

		Regards, Axel
-- 
 ,''`.  |  Axel Beckert <abe at debian.org>, http://people.debian.org/~abe/
: :' :  |  Debian Developer, ftp.ch.debian.org Admin
`. `'   |  4096R: 2517 B724 C5F6 CA99 5329  6E61 2FF9 CD59 6126 16B5
  `-    |  1024D: F067 EA27 26B9 C3FC 1486  202E C09E 1D89 9593 0EDE