Fedora ARM QA testing

[Peter Robinson]

On Sat, Mar 24, 2012 at 7:21 PM, Adam Williamson <awilliam at redhat.com> wrote:
> On Sat, 2012-03-24 at 16:37 +0000, "Jóhann B. Guðmundsson" wrote:
>> On 03/24/2012 04:22 PM, Peter Robinson wrote:
>> > That's not entirely true, in the short term we likely won't use
>> > anaconda, in the medium to long term we plan to use anacona
>> > exclusively, whether that be on the server style hardware using
>> > traditional install methods (gui, text or kickstart) or to build the
>> > images in similar ways to how we currently use it to build LiveCDs or
>> > amazon images. Having spoken with the Anaconda guys in Milan and
>> > Blacksburg they have some even more interesting ideas that I could
>> > have imagined but they're all RFE stages at the moment. In the more
>> > distant future (A15 and arm64) alot of non embedded ARM platforms will
>> > even use uEFI..... whether that's a good thing or not I'll leave up to
>> > the reader to decide:)
>>
>> If and when the arm sig decides to start using Anaconda ( which is
>> probably based on this when Anaconda supports it )reporters would have
>> to follow the install matrix until then they simply report if the image
>> that got created for their particular hw works or not.
>>
>> You guys might want to give us a list of applications to test so we can
>> see if we already cover them.
>> ( We obviously need to created test cases for those that we don't cover
>> already.)
>>
>> Other than the above I dont think there is nothing more we from the arm
>> sig.
>>
>> If the arm sig needs some kind of blessing from QA considered it done (
>> unless someone object then he can do so now ).
>
> I pretty much agree, every time we've talked about it we've figured
> there's not actually a whole lot that needs to be done. The biggest
> thing really is to make sure there are people with every supported ARM
> target device who are willing to file a 'does it work' report (and,
> obviously, bugs where it doesn't) for each compose, I can't see a lot
> beyond that at present. Peter, can you? Are there any big problems with
> actually running ARM at the moment of the kind that would be appropriate
> to turn into test cases for the future, or is it really just a case
> right now of development grind in getting things to build?

No, there will likely always be weird hardware bugs with particular
devices but I don't think that's any different to x86 other than ARM
isn't quite as mature. On the higher levels of the stack most of it
should work just like work like x86 platforms. If it boots and X works
most of the bugs will be the same as x86, it's little endian so there
shouldn't be any of the endian style bugs you might see on something
like PPC.

I'm sure as we evolve we might find there's other particular things we
might want to test but I see that as being QA for every release
anyway.

As for supported platforms for testing the ARM guys have a number of
them (I have 5 currently for example) and we plan on having means of
getting devices into people's hands for testing as well as things move
forward, and devices like the Raspberry Pi at $35 aren't exactly
expensive.

Peter