Re: RFS: flint -- C library for number theory

[Julien Puydt <julien.puydt@laposte.net>]

Hi,

Le 09/08/2014 23:39, Tobias Hansen a écrit :
> Am 09.08.2014 11:27, schrieb Julien Puydt:
> > Hi,
> >
> > Le 09/08/2014 00:25, Tobias Hansen a écrit :
> > > Am 09.08.2014 00:20, schrieb Julien Puydt:
> > > > Hi,
> > > >
> > > > Le 07/08/2014 23:17, Tobias Hansen a écrit :
> > > > > Hi Julien,
> > > > >
> > > > > here are my comments:
> > > > >
> > > > > - Check if the patches from the two NMU's are still needed,
> > > > >      if not keep at least the changelog entries and write in the
> > > > >      new entry that they are no longer needed.
> > > >
> > > > Sorry, which NMUs?
> > >
> > > Check the PTS and the package that is actually in Debian. :)
> > > https://packages.qa.debian.org/f/flint.html
> >
> > Both patches are obsolete... I don't know if they're still needed, but
> > the upstream code changed. I would trust upstream : they're quite
> > reactive, so if they were brought to their attention, there is no
> > problem any more.
> >
> > > > > - Fix broken patch header. If there's a web link or something
> > > > > explaining
> > > > > why the test has to be disabled, add it please.
> > > >
> > > > I'll have to ask Felix because:
> > > > 1. it looks like I was the culpr... author,
> > > > 2. furnished it with a proper header claiming forwarding it upstream
> > > > wasn't needed
> > > >
> > > > but I don't remember anything about it!
> > >
> > > It also says the test has to be disabled for ntl 6, but we don't have
> > > that version in Debian (another red spot on our status page).
> >
> > The patch isn't for NTL 6, but for GMP 6, so we do have it. Depending on
> > the version, everything is invertible in the ring Z/1Z or not. Of course
> > that means the tests fail for us... Someone made a better patch upstream
> > which will go in the next version : I have used it.
> >
> > If you want to have another look, I pushed my latest changes.
> >
> > Snark on #debian-science
>
>
> When you updated the patch you committed changes to upstream files.

Of course : the previous patch was applied so I unapplied it, then I 
applied the new one. If you compare the checksums (with sha1sum, md5sum 
or whatever), you'll see that the package .orig.tar.gz is the same as 
the upstream .tar.gz, so I don't think I made a mistake.

> I'm still not satisfied with the changelog. It does not even mention
> that the packaging was totally redone. debian/rules looks nothing like
> the one of the package which is in unstable right now. The changelog
> still says the patch is about ntl >= 6. Also, when the changes of the
> NMUs are not needed, please copy their changelog entries into the
> changelog nevertheless and say in your new entry that the changes are no
> longer needed.

I added those two entries and rewrote the last one.

> I also committed a small change.

Good!

Thanks,

Snark on #debian-science