On 11 Oct 2017 4:48 pm, "Pierre-Yves Chibon" <pingou(a)pingoured.fr> wrote:
On Wed, Oct 11, 2017 at 04:34:52PM +0100, James Hogarth wrote:
On 11 October 2017 at 16:23, Gerald B. Cox <gbcox(a)bzb.us>
wrote:
On Wed, Oct 11, 2017 at 7:23 AM, Till Hofmann
<thofmann(a)fedoraproject.org> wrote:
The very first sentence of the page you linked above:
The updates-testing repository, also referred to as Test Updates,
contains updates scheduled to be released for Branched
pre-releases
(after the Bodhi enabling point) and stable releases of
Fedora
The point is that your update is not intended to ever make it to
the
stable update, i.e., it is not "scheduled to be
released" for
anything. If I understood correctly, you want people to test the
beta
version and then eventually submit the final release (i.e.,
not
this
update) to stable. I don't think that's how the
updates-testing
repository is supposed to be used. Instead, it should only contain
updates that will eventually make it into stable.
Regards,
Till
+1 - Exactly...
Thought I'd quickly test this being built in a COPR ... but where
exactly
are you building from?
https://src.fedoraproject.org/rpms/firefox/blob/f27/f/firefox.spec
That shows vesion 56, not 57, and same with rawhide.
I didn't think we could build from a non-fedora-branch git branch for a
bodhi update ... that feels ... wrong ...Â
Is this an intended effect of the "arbitrary branches" for
modularity?Â
This really feels like it breaks the history/audit trail fro what
ends
up
in our repos.
This doesn't even show the branch it came
from:Â
https://koji.fedoraproject.org/koji/buildinfo?buildID=981886
But it gives the commit which is:
https://src.fedoraproject.org/rpms/firefox/c/fd700ad0ae450c4705017e05db7af7
09f7ea90f0?branch=stransky-firefox-57
itself part of the stransky-firefox-57 branch:
https://src.fedoraproject.org/rpms/firefox/commits/stransky-firefox-57
This behavior is nothing new and is the reason why releng does not allow to
delete branches in dist-git.
Pierre
_______________________________________________
devel mailing list -- devel(a)lists.fedoraproject.org
To unsubscribe send an email to devel-leave(a)lists.fedoraproject.org
Yes I saw the commit but that is my very point.
I was pretty sure that only scratch builds could be carried out from non
release branches but you get something into a compose you needed to merge
to master or a release branch.
Otherwise things like releng or proven packagers doing rebuilds for library
bumps or similar issues start to go very wrong.
They'd go to do a patch or release bump in the branch and boom...
versioning screwed.
As I said I suspect this is a side affect of the modularity arbitrary
branching stuff, similar to how as an accident it was possible to do docker
container builds from the rpms namespace at first.
It'd be good to get confirmation from releng or FESCo on the intended
behaviour as this feels wrong...