[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bypassing mentors http/ftp area durng sponsorship (was: Re: Introducing spurious revisions ... )



On Wednesday 21 January 2009 00:46:22 Martin Meredith wrote:
> On Tue, Jan 20, 2009 at 11:09:17PM +0100, Adeodato Simó wrote:
> > * Martin Meredith [Tue, 20 Jan 2009 22:05:11 +0000]:
> >
> > Hey Martin,
> >
> > > On Mon, Jan 19, 2009 at 02:29:10PM +0100, Adeodato Simó wrote:
> > > >   * history: I haven't seen this argument fly be in (this and other
> > > >     instances of) this discussion, but the answer is that the proper
> > > >     place to record that you had an extra and fatal space in your
> > > > rules file is your VCS. (And FWIW, I think much of this discussion
> > > > would go away if sponsorship was VCS-based rather than dsc+diff.gz
> > > > based.)
> > >
> > > That actually sounds like an interesting project. Something I might
> > > start hacking on.
> >
> > Uhm, what needs hacking, exactly? The sponsoree just says "Repo for the
> > package is at $URL", no? Or do you mean some kind of integration with
> > mentors.debian.net?
>
> Tools to make things easier to grab stuff from $repo :) etc etc... and a
> good workflow documentation... maybe some other stuff :D

Well, the tools are already there I believe, it is the possible difference in 
the workflows which might cause some dissensions. Yes, there might be 
spurious revisions during sponsorship, but there is also a sponsoree overhead 
which should be taken into account. For instance, I almost always hesitate to 
ask sponsoree to upload large tarballs to mentors (my inet connectivity might 
be untra cheap, but I coundn't be certain of theirs easily). Thus if 
sponsoree uses a VCS (as suggested by Dato and also several times in the 
past), it would be nice to try to do it his/her way if that could save them 
some hassle, and let them spend their resourses on actually improving the 
package.

A possible example, though the package is real and you can test these 
yourself:

Grab the package from sid and put it in sid/ for firther comparisons
$ apt-get source gxemul

Grab sponsoree's work
$ debcheckout gxemul gxemul-0.4.7

cp orig.tar.gz from sid/ 
* yes, I know of pristine-tar, but there is no way for it to preserve the 
timestamps inside the tarball, i.e. hashsums will not much with these on the 
upstream site, unless I'm awfully mistaken) 
* if that is a new upstream version, unless the sponsoree has provided 
get-orig-source target, you grab that from upstream site.

cd gxemul-0.4.7
git-branch master
git-checkout master
(later git-buildpackage will insist on working on master)

* review the repo changelog and files; fix stuff if any, mail diffs to the 
sponsoree, git-pull, discuss, etc, until done. This could go via -mentors 
mailing list in order to acquire some more peer review.
* of course the, spurious revisions are gone now (possibly forgotten -v 
also;-), and sponsoree uploads their possibly *small* changes only once; no 
tarballs uploaded by them at all, it is the sponsor's work to get the tarball 
anyway.

git-buildpackage [options] (yes, there is also a --git-pristine-tar option;-)

Compare with debdiff or whatever the result with the package from sid/ 
directory. Install, test, whatever ... Upload, and let the sponsoree knows 
about that (though they would receive a mail from the Debian queue daemon 
anyway) so they could tag their work.

P.S. Yes, this workflow surely could be better; I don't insist on it being 
streamlined to perfection. And, yes, I also know of gitpkg and topgit ;-)

-- 
pub 4096R/0E4BD0AB 2003-03-18 <people.fccf.net/danchev/key pgp.mit.edu>

Attachment: signature.asc
Description: This is a digitally signed message part.


Reply to: