Closed dgrebb closed 9 months ago
@garris let's discuss this - I see some evidence that new
headless is actually slower.
It appears that one can pass headless: old
as the mode, which apparently silences the deprecation warnings.
Maybe that's an option for now? The Chromium team says "we're going to make that the default later this year" back in January. So ¯_(ツ)_/¯
I'll look into ways perf testing can be done.
Few thoughts.
1) pretty sure this is exposed in engineOptions
so risk of defaulting this to the "wrong" setting is fairly low.
2) we should probably default this to headless: new
but provide code snippet in docs for users who might hit an incompatibility.
3) backstop stashes time started and time completed and spits out duration into the CLI. So testing for performance should be pretty easy. You could run sanity test sequentially 20 times and grep for the timing value to get a cumulative average.
4) I'm open to other approaches. Just sharing gut reactions. Cheers.
Closed by https://github.com/garris/BackstopJS/pull/1535
Super happy I haven't caused any nightmares yet.
Let's make the full change in source to use
new
headless mode as the default.Test regressions, test fallbacks, and ensure support for
old
mode is supported whennew
becomes vendor standard.Notes
new
by EOY 2023new
headless