On 29/06/2017 10:07, Jani Nikula wrote:
On Thu, 29 Jun 2017, Daniel Vetter <daniel@xxxxxxxx> wrote:
Just adding tags because tags sounds like a cool idea on top of all the
other things we do already still doesn't sound appealing to me.
Agreed.
Well it's not on top, but as a replacement, but anyway.
So I always thought it was silly when we relied on test names to select
tests to run or exclude in the past. I'm equally suspicious of adding
any sorts of categories or tags within the test sources. I don't think a
test source file should be changed if the category changes. Thus I think
maintaining the categories outside of the tests is a good idea.
Currently we maintain the categories as test lists. I think you could
call them tags just as well. Them being lists in files is just an
implementation detail. If a test is included in a test list, then it's
so tagged. If we wanted to, we could add more logic to running the tests
based on which lists the tests are in. For example, "in extended and not
in stress".
We started with the premise that tests lists are bad. :) If we have
multiple lists we will need someone to keep an eye on test changes and
keep all the lists up to date. To me that has a higher chance of failing
that having tags together with the code. It is easier to remember if it
stares at you from the screen while you are editing the test.
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
