[wdspec] Only resize and re-position window in session setup and teardown if needed #43853
Conversation
wpt-pr-bot
requested review from
AutomatedTester,
jgraham,
jrandolf,
juliandescottes,
sadym-chromium and
shs96c
whimboo
force-pushed
the
chromedriver_window_bounds
branch
3 times, most recently
from
972af8e
to
a1802c1
Compare
|
The Safari wpt.fyi results are somewhat busted because the
This timeout only happens with my patch applied and I'm not able to reproduce it locally at all. Also the log output on Azure stops at that point, so maybe it's an issue with the provider? |
|
@whimboo great, could it be the reason for the slowness we investigated previously? |
whimboo
force-pushed
the
chromedriver_window_bounds
branch
from
a1802c1
to
dd03258
Compare
whimboo
force-pushed
the
chromedriver_window_bounds
branch
from
dd03258
to
70175ba
Compare
Not sure which slowness you mean specifically but that is / was probably a longer term issue. But since recently when the prompt handling and network interception tests landed there are quite a few of extra Timeout results for BiDi and Chrome. Increasing the timeout here doesn't make sense given that the tests would still hang. So one of you might want to take a look at those. |
|
The chrome stability job finished but looks like it missed synching and as such is still listed as running. I've checked the logs of the other jobs and everything looks fine. @jgraham would you mind admin merge this PR? Thanks! |
|
@sadym-chromium and @nechaev-chromium I noticed this change in chromedriver from mid of March. Does that actually mean that we can remove the workaround as put in place with this PR? |
|
Yes. The fix is supposed to eliminate the unconditional sleep on window bound change. There is no longer any performance benefit in checking the bounds before setting them in the tests. |
|
@nechaev-chromium that is great to hear! Would you mind to provide a PR to get these changes reverted then? Thanks! |
I've created #46589 |
This is a workaround until the underlying bug in Chrome has been fixed. Potentially we can remove a lot of longer timeout requests as well. Lets see what tests results show us.