Bernahrd, On Sat, 2015-08-15 at 16:59 +0200, Bernhard Voelker wrote:
On 08/14/2015 04:20 PM, Dominique Leuenberger / DimStar wrote:
Staging:A grew to a small monster that does not want to be tamed... but I'm sure the monster is getting tired soon.
I've incidentally followed that growth ... and I'm wondering if "adding packages until no errors" is the fastest strategy?
It is to a certain point - then we generally start splitting stuff up again.
Staging:A started with ~5-8 packages 25 days ago. As there were some errors, other packages with fixes got added - now we're at 47 packages. Wouldn't it have been easier at some point to move some packages out for a while, thus splitting the problem?
If we'd not add the fixes to the same staging directly, the fix would
first have to go through any other staing, be accepted into factory and
the the first staging be completely rebased (stagings are not building
against 'openSUSE:Factory', but contain frozen links to a 'snapshot in
openSUSE:Factory'; so fixes do not auto-propagate).
In some cases it's over-eager devs causing some more delays (like,
superseding a request when a staging is almost ready.. meaning we have
to redo the integration test with the new submission).
:A grew indeed in a strange way, two stagings were merged because one
package contained fixes for both cases.. (A which had glibc and a
couple others and E at the time, which was systemd). Now, systemd has
no longer build fails, but it fails to pass openQA (enrypted disks /
password dialogs with plymouth seem to be a big issue).
Cheers,
--
Dimstar / Dominique Leuenberger