Is a 16.04 alpha 2 needed?

Steve Langasek steve.langasek at ubuntu.com
Wed Jan 27 08:07:23 UTC 2016


Thanks for the discussion, folks!

On Tue, Jan 26, 2016 at 07:32:09AM +0000, flocculant at gmx.co.uk wrote:
> >On Mon, Jan 25, 2016 at 09:42:07PM -0800, Walter Lapchynski wrote:
> >>I think it's worthwhile doing alphas, especially for an LTS version.
> >Could you elaborate what value you see that you're getting out of these
> >alphas, that couldn't be achieved more effectively by e.g. automated install
> >testing against the daily images?
> Just a quick note on this (Xubuntu aren't going to participate in A2) -
> the auto install testing of images only happens for Ubuntu afaik.  There
> was some effort during last cycle to get it going for flavours again which
> appears to have stalled.

Ok, do you know where this has stalled?

It's true that Canonical is only providing the infrastructure for automated
testing of Ubuntu images.  But it's my understanding that the infrastructure
should be replicable by others who want to also run these tests for other
flavors.


On Tue, Jan 26, 2016 at 12:14:42PM +0100, Philip Muskovac wrote:
> I still believe the Alphas provide some value, because esp. for us flavors
> that don't have lots of testers that regularly check the images, it's a
> nice way to bundle some resources and also communicate between flavors to
> not break each other for a couple days, instead of just "uh hey, can
> people please test the daily images over the next few days, some might
> work as intended".

What kind of breakage are you actually concerned about, here?  The model I'm
advocating, and that has been adopted by Ubuntu, is that breakage at the
package level is blocked by proposed-migration holding back packages with
test failures; and breakage at the image level is blocked by automated
install tests holding back images that don't pass the image-level tests.
And when something broken slips past the tests, you fix it, improve the
tests, and repeat.

In this model, breakage related to one flavor cannot break the image for
another flavor.  It's possible that it can cause the latest good image for a
flavor to be outdated; but the alpha milestone model *ensures* that this is
so.  The alpha image is always at least a day out of date vs. the archive by
the time it's released...

> As has already been said, lots of the automatic QA is so still *u*buntu
> specific, so at least kubuntu is still mostly relying on manual QA, and
> it's not like every develop pays/can pay attention to not break any
> flavor, so the "continuous installability guarantee" either
> *just* covers the packages, not the images, or it just covers ubuntu
> again.

It's only a guarantee if flavors choose to exercise the option of running
automated tests on their images before promotion to "current".  As it's been
mentioned above that automated image testing for flavors has stalled
somewhere, that seems like a good place to focus our attention to save
everyone time in the long run.

> So please leave us the possibility to be able to publish tested and
> somewhat guaranteed-to-work images during development, even if not all of
> us make use of it all the time.

I would much rather help you have the guarantee that every image published
as "current" is tested and works :)

-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
Ubuntu Developer                                    http://www.debian.org/
slangasek at ubuntu.com                                     vorlon at debian.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <https://lists.ubuntu.com/archives/ubuntu-release/attachments/20160127/722bbe68/attachment.pgp>


More information about the Ubuntu-release mailing list