Housekeeping (Was Re: FE Package Status of Sep 4, 2006)

[Ville Skyttä]

On Mon, 2006-09-04 at 21:20 +0200, Thorsten Leemhuis wrote:

> 28 Packages not present in the development repo
> -> most of them probably should be marked as dead.package in cvs. If
> they get re-submitted later they should go though a full review again.

I think taking care of the subset of these that are really dead on/after
Sep 18 is good enough.  See below and
http://fedoraproject.org/wiki/Extras/Schedule/FC6MassRebuild

> 6 orphaned packages available in extras devel
> -> okay, they were orphaned just a few days ago iirc. But I'd sleep much
> better if I knew that theres a group of people that watching this list
> and makes sure that stuff like this gets removed when there is no
> maintainer found after three weeks (but I'm quite sure scop will handle
> it for his old packages in this case)

Sure.  My impression is that we'll handle all orphaned packages *and*
ones whose needs.rebuild status is not taken care of before Sep 18 by
bluntly removing them from the repo, then dealing with the fallout.  I'm
prepared to take care of the removal for all affected packages, not only
ones I've orphaned recently.

BTW, I know mschwendt has been doing general orphan cleanup in devel
every now and then already now, see the green status bar at
http://fedoraproject.org/wiki/Extras/OrphanedPackages#head-7d07f9833c5986e1236196f059b9f13cc6e0a332

Common sense can be used and taking care of needs.rebuild can mean eg.
that additional comments can be added to that file stating why the
package is not rebuilt yet but shouldn't be orphaned and removed.

Here's one related example:
https://www.redhat.com/archives/fedora-extras-commits/2006-August/msg02956.html

Now, IMO removing needs.rebuild this way in this case is ok *but only
because* the package is already missing from the devel repo.  If it'd be
still there, this would not be a good way to handle things - better
would be to add comments to needs.rebuild.

> we also should make
> sure that all of our packages are listed in owners.list

I think this could be fairly easily enforceable in the buildsys.  No
owners.list entry -> no builds for that package.