New headless #11099
New headless #11099
Conversation
Codecov ReportBase: 52.31% // Head: 52.31% // Decreases project coverage by
Additional details and impacted files@@ Coverage Diff @@ ## trunk #11099 +/- ## ========================================== - Coverage 52.31% 52.31% -0.01% ========================================== Files 82 82 Lines 5503 5505 +2 Branches 198 199 +1 ========================================== + Hits 2879 2880 +1 Misses 2426 2426 - Partials 198 199 +1
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
| Hi @titusfortner , Looks like there is an issue with https://www.w3.org/TR/webdriver/#dfn-print-page when set The test runs in headless mode but still fails with error Is this an issue with chromedriver ? Thanks, |
| @harsha509 yeah, looks like that functionality only works with "old" headless mode, so that's a confounding factor for our decision. I'm also not sure why JS is the only language testing it. :) Was going to look into it more. |
It's because headless isn't being passed in the new way in the |
| It was decided that instead of changing this method that we would mark it as deprecated and advertise using the new mode as the recommended way to run in headless. |
titusfortner force-pushed the new_headless branch from d38d8b2 to 06e5bff Compare 
titusfortner force-pushed the new_headless branch from 06e5bff to dabf5e8 Compare 
| SonarCloud Quality Gate failed.
|
| Decided not to do this one, tracking with the new Issue |
| Why? I thought we were going to eventually merge this? |
| This PR is wrong. Created new issue to track actual requirement |
Most Selenium languages have implemented convenience methods for Headless mode in Chrome & Firefox.
Currently implemented:
The Chrome Headless mode we've set is not the production browser, and has many limitations (including not being able to load extensions).
Thanks to @mdmintz I learned that Chrome implemented a new Headless mode that is the full production browser:
https://chromium.googlesource.com/chromium/src/+/833543a511a970f44fb074b9f30c6b6268e0be71
It's been available since Chrome 94, but this change should be backwards compatible since it picks up anything it doesn't recognize (e.g.,
--headless=xx) as just--headless.There may be slight performance benefit to using the old mode of headless (because it strips out functionality), but I believe the new mode more accurately represents what Selenium should encourage. If we are going to maintain a convenience feature (tbh, I would also be ok with deprecating/removing all of these), then I believe this new mode is the correct one.
I'm not adding Unsetters to Ruby or JS, or Getters to non-Python because I just don't see them as that important. I am implementing Headless in .NET (both Chrome & Firefox) to match Java implementation.
I've verified that the new headless mode works on a Linux box without a GUI / xvfb. I have not evaluated performance.
Which do you prefer: