[LTP] [Automated-testing] [PATCH v2 00/30] Introduce runtime and conver tests

[Richard Palethorpe]

Hello,

Petr Vorel <pvorel@suse.cz> writes:

> Hi all,
>
>> > TODO:
>
>> > This patchset is reasonably complete in a sense that it removes the
>> > timeout API at the end. Still there are a few things to consider:
>
>> > - is reusing the -I parameter a good idea? Wouldn't adding new parameter
>> >   (-r) be better?
>
> But -r is using only when .max_iteration_runtime set, right?
> How should it behave on tests which don't set it?
> Should -I and -r be mutually exclusive?
> Maybe reusing -I would be simpler for users.
>
>> Perhaps we could deprecate -I, but convert it to use the new mechanism
>> while trying to keep the behaviour similar.
>
>> I'm not sure we need a global -r option, but unlike -i it's not much
>> effort to support. It would not surprise me if there are existing tests
>> which don't work with -i.
>                         ^^ Do you mean -I ?

Nope, I mean -i, my point is that global parameters add another failure
mode and -i is the best example of this. However -r has less
implications than -i.

>
> Kind regards,
> Petr
>
>> > - there are quite likely tests that run for more than a second or a two
>> >   and should be made runtime aware
>
>> > - anything else?


-- 
Thank you,
Richard.