Package categorization and distribution construction

[Toshio Kuratomi]

On Wed, Jan 18, 2012 at 8:04 AM, Bill Nottingham <notting at redhat.com> wrote:
> Tom Callaway (tcallawa at redhat.com) said:
>> On 01/18/2012 09:30 AM, Bill Nottingham wrote:
>> > Later it was brought up that it may just be simpler to create a second file
>> > of metadata, similar to the comps file, that just contains lists of
>> > categorized packages. (i.e., #3 above.)
>> >
>> > Opinions?
>>
>> One of the ideas that we've been seriously considering implementing is a
>> Fedora specific extension to DOAP, where the DOAP files would be checked
>> into git. It should be straightforward to add a Fedora extension field
>> for "tags", allowing the maintainer to add tags, along with other useful
>> metadata about the package that we don't know from other sources.
>>
>> Then, tagger/packages can simply read in the DOAP files and use those
>> tags in addition to the user-generated ones.
>
> The concern that was raised from infrastructure is that scouring package git
> for DOAP (or any other metadata files) might be prohibitive in cost to do
> regularly. PackageDB would be simpler, of course (maybe have git hooks that
> populate PackageDB on commit?)
>
We discussed pkgdb but one of the concerns was that this is
information about binary packages and the part of pkgdb that we were
thinking of keeping (as opposed to the portion that we want to hand
off to tagger/packages) deals with the srpm/git repo level view.

For similar reasons, the git repository isn't a natural fit for the
information.  It could be the interface to adding the information but
if we wanted to do any querying of the information it would make more
sense to have it available in something that deals with the built rpm
level rather than source rpm/git repo.

-Toshio