Comment # 8
from Paul Howarth
(In reply to comment #6)
> (In reply to comment #5)
> > (In reply to comment #4)
> > > - Shipping the *.asc sig seems pretty meaningless to me.
> >
> > I just see it being used as an extra check that the tarball hasn't changed,
> > (if we keep the signature in git) given that upstream's releases aren't
> > versioned.
> I guess you are aware, Fedora's git is automatically adding m5sums to
> "sources"?
True, but I guess the signature is a bit more obvious.
> This would apply if you intend to push the tarball to git's lookaside cache.
>
> However, this would raise the next point, I don't know the answer to: Does
> Fedora's git lookaside cache allow unversioned tarballs, rsp. can the
> lookaside cache handle this?
Yes: the URL to a tarball in the lookaside cache is:
http://pkgs.fedoraproject.org/repo/pkgs/<PACKAGE>/<TARBALL>/<MD5SUM>/<TARBALL>
So tarballs with the same name for the same package can co-exist as long as
their md5sums are different, which is a pretty safe bet.
(In reply to comment #7)
> (In reply to comment #5)
>
> > > Only issue I see now is that it should have a "Requires: pari-gp"
> > > for proper resolution of dependencies.
> >
> > Can't do that as it would lead to a circular build dependency for pari
> > itself, and it's also conceivable that other software could use the same
> > data without requiring pari-gp. I did add a "Conflicts: pari-gp < 2.2.11" as
> > that's the oldest version that can use this data, though perhaps that should
> > be "pari" rather than "pari-gp"?
>
> Do you mean these packages will be build requires of pari now? I
> see they could be useful in a very complete %check.
Yes, that's the intention - see
https://bugzilla.redhat.com/show_bug.cgi?id=821191#c10
> I think the conflict does not matter much, as pari-gp is a pari
> subpackage.
Don't know what you mean by that. I was wondering whether it was pari-gp that
used the data packages, or the underlying library "pari" - the conflict should
refer to the part of pari that actually uses the data.
> About data reuse, actually, sagemath ships elldata in a different
> format (not sure if complete and optimized for size, but a lot smaller),
> as well as it also creates a cremona_mini.db sqlite3 db during build.
But that's derived from this package though, isn't it?
