Tobias Gerschner wrote:

>> Wouldn't be hard to fix, is on my todo++ list, but there's been zero
>> interest in using rpmrepo so far, hence its a low priority.
>> 
>> You know where http://launchpad.net/rpm blueprints are.

[...]
> OTOH switching to a format which is mainly driven by a community that does 
> not seem keen to collaborate is not very pleasing either.
> 
> Has anyone any links/ information about the reasons as to why repo-md has the 
> format that it has today?

Think it went from the first extracted rpm headers (.hdr)
to XML metadata (xml.gz) to SQL metadata (.sqlite.bz2)...

Your best guess is http://createrepo.baseurl.org/ and
http://lists.baseurl.org/mailman/listinfo/rpm-metadata

> Is there any interest to review this format and potentially come up with a 
> new blueprint? While I am personally very interested in sorting this I can 
> only see this work with input and support from multiple drivers which would 
> then collaborate on this new format and actually use it. I don't just want to 
> do it my way, but have an accepted, comfortable solution that is well 
> engineered and has a good future with rpm5.

I'm pretty certain that whatever createrepo/repodata does
for a new format, it will Require yum and Conflict rpm5 ?

http://lists.baseurl.org/pipermail/rpm-metadata/2010-August/001209.html
http://lists.baseurl.org/pipermail/rpm-metadata/2010-August/001233.html

> Any thoughts? Is there an option/ solution I am missing?

There is the mixture of rpm headers and xml metadata
that urpmi is using, or using a rpmdb directly perhaps ?

But changing the "pkgid" (to not include the payload)
and adding partial downloads sounds like improvements...

--anders

______________________________________________________________________
RPM Package Manager                                    http://rpm5.org
User Communication List                             [email protected]

Reply via email to