Fedora ARM QA testing

Adam Williamson awilliam at redhat.com
Sat Mar 24 05:21:23 UTC 2012


On Fri, 2012-03-23 at 20:14 -0700, John Reiser wrote:
> On 03/23/2012 02:36 PM, Paul Whalen wrote [has been snipped]:
> 
> > We have recently opened up discussions with developers and the community,
> > engaging in a variety of topics to understand the requirements of moving
> > the ARM secondary arch to primary ( details -https://fedoraproject.org/wiki/Features/FedoraARM)
> > - with no disruption to the existing Fedora ecosphere.
> 
> The Fedora ecosphere already has spent over one thousand hours of developer time
> discussing such issues, as witnessed by at least two recent threads in
> devel at lists.fedoraproject.org:
>    ARM as primary architecture
>    RFC: Primary architecture promotion requirements
> Such time likely has not been totally productive [regardless of outcome],
> and certainly cannot be recouped.  This has been a major disruption already.

Hey, writing provocative emails to devel@ is something I do for
recreation while my brain is thinking about important things. =)

> > Introducing existing QE tests will prove problematic as currently
> > there is no anaconda support for ARM, ...
> 
> The topic of hardware+software architecture (ABI) should not be forgotten.
> Some groups have decided that "armhf" is the only worthwhile variant.
> I feel that ignoring "arm eabi[45]" (and specifically, v5te hardware)
> is a short-sighted choice.

For me, that feels more like the kind of thing that the ARM group should
discuss, to be honest. As I've always said, personally I think it's up
to those who do the development to decide what the development targets
are, and for QA to design our QA processes to be appropriate to whatever
the developers want to be developing for. We can of course all subscribe
to devel@ and the arm list and throw our two cents into those
discussions if we feel passionate about them, but I think those are the
appropriate contexts to talk about that. To me it feels inappropriate to
try and dictate from a QA context what development goals ought to be.

Paul - in one of those interminable devel@ threads I put a little sketch
of what I would expect early-stage Fedora ARM QA to look like, just
thinking off the top of my head. I don't actually think it needs to be
particularly complicated or involved, to be honest. It could probably be
just about as simple as a Does It Boot / Does It Install? matrix with a
row for each of the specific ARM systems we intend to support and
provide images for; we make sure that, collectively, we have at least
one of every system, and when a release point rolls around, releng spits
out the images for each system, and we fire them up and see if they
work. That at least feels like the natural starting point, to me.

By all means reanimate the ticket - honestly, I kind of got buried in
other stuff and hadn't really had a minute to get back to it lately.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net



More information about the test mailing list