Lack of update information

[Daniel P. Berrange]

On Mon, Jan 26, 2009 at 01:17:40PM -0500, Bill Nottingham wrote:
> Richard W.M. Jones (rjones at redhat.com) said: 
> > > *All* of that is in the RPM header. It's not pulling anything out of the
> > > file payload.
> > 
> > So?  createrepo would need to parse a bit more of the RPM.
> 
> 1) Having createrepo have to parse the entire cpio metadata to
> look for a particular file would be insanely slow
> 2) I just picked a random package I maintain. The ChangeLog is
> 1.4MB uncompressed, 326k compressed. Even if that's larger than
> typical, imagine scaling that out across 7000 source packages,
> and then the size of the resulting XML to download...

Perhaps this idea of including changelogs in the metdata is addressing
the issue with the wrong hammer. When getting a set of 50 updates, very
few users are likely to want or need the changelogs to all of them
prior to install. More likely, they'll just want to occassionaly read
a changelog for an ad-hoc package to find out some info. 

This suggests that rather than including the full changelog for each
package in the createrepo metadata, just include a URL linking to the
changlog on a static webserver. It'd be easy enough to extract changelogs 
for each newly built package somewhere easily linkable (and mirrorable if
neccessary). So on a rare occassion when a tool processing repodata does
want the changelog it can quickly download it on demand.

Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|