It's been a long time since I looked at those tests. I think the regression does look like something we should worry about and address, though I would bet it's not a user-visible issue. It's likely more of a situation where the "guilty" commit here has influenced the signal that we're waiting for, in such a way that it reduces our ability to measure regressions into the future. I think (?) the test is measuring time-to-first-paint after pageload, where "paint" is defined using the MozAfterPaint event, per: https://searchfox.org/mozilla-central/rev/c0bed29d643393af6ebe77aa31455f283f169202/testing/talos/talos/test.py#956,965,975,983 So: maybe the guilty change here is inadvertently causing a delay to when we dispatch MozAfterPaint, for some reason?
Bug 1783674 Comment 6 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
It's been a long time since I looked at those tests. I think the regression does look like something we should worry about and address ~though I would bet it's not a user-visible issue. It's likely more of a situation where the "guilty" commit here has influenced the signal that we're waiting for, in such a way that it reduces our ability to measure regressions into the future.~ EDIT: it may in fact be user-visible, as indicated by profiles in my next comment. I think (?) the test is measuring time-to-first-paint after pageload, where "paint" is defined using the MozAfterPaint event, per: https://searchfox.org/mozilla-central/rev/c0bed29d643393af6ebe77aa31455f283f169202/testing/talos/talos/test.py#956,965,975,983 So: maybe the guilty change here is inadvertently causing a delay to when we dispatch MozAfterPaint, for some reason?