On 01/30/2014 07:14 PM, R P Herrold wrote:
On Thu, 30 Jan 2014, Tal Nisan wrote:
> I think it will help stabilizing the new features in a short circle, sometimes
> there are small "fine tuning" fixes that don't have bugs opened for
> the flow is easier if you can just backport them without opening a bug.
> On 01/30/2014 04:42 PM, Itamar Heim wrote:
>> I'm inclined to drop the "bug-url is required" for engine stable
> until we GA 3.4, as a lot of the patches are for pre-GA features.
I took Alon's comment as a -1, and do not understand the flow
Cloning a bug is a click and a bit of web trimming. Adding it
as a dependency on its parent is typing a bug number in the
proper field. It is very hard, unless sone follows both
mailing lists, and commit mailing lists, to figure out some
changes and if this is a backport (and not: "stabilizing the
new features" as Tal mentions)
The thrashing in stabilization, and very fast gerrit approvals
on the cpopen patch (it happened in hours, months ago) show
the need to be explaining to a Bugzilla, what is happening.
As near as I can tell the package succession is still not
'right' for upgrades [It seems to need a manual exclude in a
yum repo config in one place, OR a manual Provides, OR an
epoched EVR (!! NOT RECOMMENDED !!)]
cpopen was in vdsm, this is about engine repo.
backporting to stable branch has it overhead already. we forked early to
make sure we are stabilizing rather than getting more forward-looking
changes merged into master.
however, as alon said, in that regard we forked 'early'.
this incurs an overhead of backports already.
looking at the patches i see, which are for stabilization, not all of
them have a bug tracking them, and i don't see a reason forcing this for
pre-rc as the main benefit of forcing bugs is tracking changes for
release notes (that's to force creation of bugs post issues fixed, not
using bugs to track issues).
so if there is a bug, i expect it in bug-url for easier tracking, but i
don't see the value in forcing it pre-rc.
-- Russ herrold