>> Generated on a per-repository basis so that consumers get the data
>> they are interested in. Here I mean one for ftp.d.o/debian and one for
>> each of the archive.d.o/* repositories.
> I'm confused about this. We have several additional archives ATM, for
> instance security and lts.

lts is not a seperate archive. Only suite. Same as backports is (no
longer) a seperate archive.

> Does the above mean that the maintainer of an infrastructure piece
> that spans multiple archives will have to separately retrieve the
> metadata from each of them?

Yes. Note: *Archives*. Not suites.

> There is certainly value in keeping the metadata close to the
> corresponding data, but Debian as a project should IMHO also centralize
> somewhere all information about all its official archives. Are you
> ruling out a scheme that would provide such a single source of
> information?

I think the data should get autogenerated as much as possible, and the
few bits that can't should get put right at the place from where each
other autogenerated comes from. Which is dak.
That demands a per-archive file.
Per-archive also allows any other number of archives to easily produce one.

It would be nice if the files end up in a way that you can just
'cat * >foo' them to one datafile, yes.

-- 
bye, Joerg
When will I learn? The answer to life's problems aren't at the bottom of
a bottle, they're on TV!

Attachment: signature.asc
Description: PGP signature

Reply via email to