[Fedora-packaging] Epoch policy

[Tom 'spot' Callaway]

On Sat, 2005-02-26 at 18:24 +0100, Michael Schwendt wrote:
>On Sat, 26 Feb 2005 10:19:15 -0600, Tom 'spot' Callaway wrote:
>
>> On Sat, 2005-02-26 at 11:24 -0500, seth vidal wrote:
>> 
>> >1. Look on the fedora-packaging list for the discussion
>> >2. my guess is: 
>> >   a. if the fedora.us package had a non-zero epoch it needs to be
>> >maintained - just so users have an upgrade path
>> >   b. if the fedora.us package had an Epoch: 0 drop it and remove
>> >%{epoch} from anyplace you have it in ver strings.
>> 
>> I agree with this. Anyone else have thoughts?
>
>Dropping "Epoch: 0" breaks rpm -F updates. This is in bugzilla
>somewhere.

Fixed it. Hopefully, this will make it into FC4. For FC3 and earlier,
we'll have to document it like this:

In Fedora Core 3 and earlier, there was a bug in rpm that caused the
"-F" or "freshen" case to fail if you attempted to upgrade from a
package that had "Epoch: 0" to a package that had no Epoch: value. Thus,
for Extras packages in the Fedora Core 3 branch (or earlier), if the
package has any Epoch: value defined (even 0), then all updates in that
branch must also have an Epoch: value defined.

In the Extras Fedora Core 4 branch, you should not define Epoch: 0, even
if earlier revisions of the package did. If earlier revisions of the
package had a non-zero Epoch, you should keep Epoch, so that users have
an upgrade path.

New packages (packages where there is no previous package to upgrade
from) should not use Epoch.

Does that seem reasonable (pending the fix being included in FC4's rpm)?

~spot
---
Tom "spot" Callaway: Red Hat Sales Engineer || GPG Fingerprint: 93054260
Fedora Extras Steering Committee Member (RPM Standards and Practices)
Aurora Linux Project Leader: http://auroralinux.org
Lemurs, llamas, and sparcs, oh my!