Investigate GH crashes when page timeout is set multiple times

[uuf6429]

While adding tests in #38, we've hit a very strange bug where setting page load timeout in combination with window tests causes crashes, but only on GitHub actions.

We've had a long investigation already, but it was inconclusive, so we ended up disabling the tests in the specific cases.

This issue is to track this problem, investigate deeper, collect information and hopefully find a solution.

[uuf6429]

@aik099 I wonder if the crashes might have been related to php-webdriver/php-webdriver#1120.

[aik099]

Have you updated our tests to avoid using the NULL timeout value as a means for resetting a timeout in a JsonWireProtocol?

Maybe we need to stop/start the Selenium session to reset timeout (I thought we were already doing that).

[uuf6429]

Well at one point instead of null or not setting the timeouts, we started setting the default W3C values (e.g. 30s etc). Other than that I don't know. For now I propose waiting until php-webdriver-side is fixed then I can resume this task.

[aik099]

Well at one point instead of null or not setting the timeouts, we started setting the default W3C values (e.g. 30s etc). Other than that I don't know. For now I propose waiting until php-webdriver-side is fixed then I can resume this task.

OK. I hope, that applying this fix in the above-mentioned PR (we can use it only with a new WebDriver release) would fix the problem.