On 11/13/18 8:50 AM, Daniel P. Berrangé wrote:
> On Tue, Nov 13, 2018 at 10:39:57AM +0100, Markus Armbruster wrote:
>> Drive-by comment...
>>
>> Cleber Rosa <cr...@redhat.com> writes:
>>
>> [...]
>>> My impression is that the "infrastructure for block tests" is not that
>>> different from the infrastructure needed by other tests, specially other
>>> QEMU tests.
>> [...]
>>
>> Yes. The actual reason for having a completely separate testing
>> infrastructure for block tests is that it predates testing
>> infrastructure for anything else.
>>
>> Moving the tests to common infrastructure would be a sizable one-time
>> effort we can ill afford. Maintaining multiple testing infrastructures
>> is an ongoing effort we can also ill afford.
>
> If we do want to move to a common infrastructure, then IMHO this patch
> series should do the work to convert qemu-iotests to use it.
>
> Currently this is simply introducing a 2nd way to write block tests.
>
> We know from bitter experiance that when we introduce a new system in
> QEMU without converting the old system, the old system will live forever
> increasing our maint burden & creatnig confusion for developers about
> which is the preferred approach.
>
>
> Regards,
> Daniel
>
Right, I don't think there's enough momentum or interest in doing that
right now. As myself and Kevin suggested, the best solution at this
point may be wrapping the qemu-iotests without touching them.
Regards,
- Cleber.
