[Webkit-unassigned] [Bug 72841] NRWT: option --skip-pixel-test-if-no-baseline support on DRT

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Mar 8 12:40:21 PST 2012


https://bugs.webkit.org/show_bug.cgi?id=72841





--- Comment #9 from Dirk Pranke <dpranke at chromium.org>  2012-03-08 12:40:21 PST ---
(In reply to comment #8)
> > given that we know whether we should run pixel tests or not in worker.py (as your change indicates), we should combine pass test_input.should_run_pixel_test to driver_input and pass run_pixel_test to driver_input and just use a different DRT instance to run the test in this case. 
> 
> I can't found the run_pixel_test you have mentioned. Is it a method, or a member? I don't understand how should the combined passing mechanism work.
>

I'm sorry, I could have been a little clearer. Change the is_reftest member of DriverInput to run_pixel_test and assign it the value we compute in worker and singletestrunner.

> > If we do that, then you don't need to modify DRT at all and the logic of needing pixel tests or not stays where it should be, in worker.py and single_test_runner.py.
> 
> Should we remove all of the modifications from WebkitTestRunner that were made by https://bugs.webkit.org/show_bug.cgi?id=70484, and we don't need those ones in DRT that can be found in the patch above too? I mean we should handle it only with Python code?

I probably would remove the previous changes in WebkitTestRunner, yes. I am saying we can handle it completely in Python code (for now; eventually we want to pass in whether we need pixel results or not as part of the per-test input, but that's a whole different discussion you should ignore for now :).

> If I understood it correctly, we should always skip pixeltest for a test if there isn't any expected png which belongs to it. So we won't use the command line argument --skip-pixel-test-if-no-baseline for evoking this behavior, will we?

Different webkit ports want different behavior. The Qt port wants to skip the pixel tests, but the Chromium port doesn't, for example. So we still need the flag on new-run-webkit-tests.

I hope this helps; let me know if you have further questions?

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.



More information about the webkit-unassigned mailing list