checkbox-dev team mailing list archive

[Ara Pulido <ara.pulido@xxxxxxxxxxxxx>]

On 25/03/15 12:42, Zygmunt Krynicki wrote:
> I agree there are some setup considerations.
> 
> We should think about what we currently do (e.g. extra storage drive
> or thumb drive or blank/rewritable optical disk) and see how we want
> to handle things like this.
> My gut feeling is that we should have something other than a job and
> that for certification testing we should just do manifests and ignore
> all the mumbo-jumbo flaky detection. It's not worth it and it doesn't
> solve the problem. Let's focus on how to incorporate manifests into
> our workflow. Then knowing the canonical identifier of the machine
> being tested we can just load the manifest and test without flaky
> guesses.

I agree manifests seems to be the way to go. But this is a long term
plan, that I agree we should pursue, but not relevant to TB.

For TB, as we need the tests now, let's stick to the asking the user
thing for now.

Let's discuss manifest in parallel.

Thanks,
Ara.

> 
> Thanks
> ZK
> 
> 
> 
> On Wed, Mar 25, 2015 at 12:38 PM, Ara Pulido <ara.pulido@xxxxxxxxxxxxx> wrote:
>>
>>
>> On 25/03/15 11:00, Zygmunt Krynicki wrote:
>>> Hey.
>>>
>>> So my opinion on thunderbolt is that it is actually pretty good that
>>> we cannot detect it. The last thing we should do is ask the user. At
>>> the end of the day, we'll clone all the tests and replace the word
>>> "display port" with "thunderbolt".
>>
>> So, we have three tests to create:
>>
>> 1) Displays - I agree that in this case it is equivalent to DP
>> 2) TB Storage
>> 3) Ability to put a TB storage device connected to the TB monitor, which
>> is connected to the system.
>>
>>>
>>> The only exception is the daisy-chain test which should IMHO follow
>>> our dependency step process where we ask the user to setup a
>>> particular configuration before doing a off-the-mill storage test.
>>
>> Exactly, that's the exception, but that's a test that we need to create,
>> and only offer to those systems that have a TB port.
>>
>> That's why we need to identify those TB (by asking the user), just
>> because of the daisy chain one, but that's enough to have the test.
>>
>> Specially because we have just 1 TB monitor, so imagine that you are
>> testing a system (from a pile that you need to test) and you don't
>> realize it is a TB one:
>>
>> 1) You start testing
>> 2) Checkbox thinks is a DP, so asks you to plug a DP monitor
>> 3) You do it
>> 4) Then it asks you to do daisy chain, if you have a TB port, and then
>> you realize it is actually a TB one, so you need to remove that monitor
>> and go and get the TB one. This wastes time.
>>
>> I think we should ask the user whether this is a TB system, and do it at
>> the very beginning of the run.
>>
>> Thoughts?
>> Ara.
>>
>>>
>>> Thunderbolt is display port. Let's not clone display tests because the
>>> color and shape of the plug is different. Storage tests are different
>>> because we're not testing thunderbolt. We're testing the add-on dongle
>>> that adapts SATA or other storage device to work over PCI-E. This is
>>> just a new shape for a storage controller card. We have tests for
>>> storage. Nothing to see here, move along.
>>>
>>> Thanks
>>> ZK
>>>
>>
>> --
>> Mailing list: https://launchpad.net/~checkbox-dev
>> Post to     : checkbox-dev@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~checkbox-dev
>> More help   : https://help.launchpad.net/ListHelp