http://fedoraproject.org/wiki/TomCallaway/SecondaryArchitectures

[Jesse Keating]

On Wednesday 11 July 2007 07:42:02 David Woodhouse wrote:
> Yes, I understand that there are disadvantages with that approach, just
> as there are disadvantages with the new approach.
>
> Nevertheless, if we're going to make it a frequent occurrence that
> packages can go missing on secondary architectures, then we are going to
> have to cope with that _somehow_. For the package in question to just
> disappear isn't likely to be workable.
>
> That's true even if we take the sensible option and require explicit
> approval from the package owner before shipping a partially-failed
> build.

Actually I think you get what you want by default here.  If the build fails on 
the secondary arch, that build isn't /imported/ on the secondary arch, and 
thus when querying the secondary arch collection on the secondary arch koji, 
you'll get the last successful build.  Maybe not, I'm not entirely sure how 
the 'latest package' routines for repo creation are going to be handled 
between downstream and upstream kojis.  There is certainly value in both 
approaches, and I think you're right, if the secondary arch koji just doesn't 
import the build that failed, it just has the previously built one, then it 
might be easier for that secondary arch to keep up, even if they have to do 
some rebuilding to really "catch up".

-- 
Jesse Keating
Release Engineer: Fedora
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-devel-list/attachments/20070711/8f50606b/attachment.sig>