[test] Update font-awesome CSS file in regression tests fixture#43745
[test] Update font-awesome CSS file in regression tests fixture#43745Janpot merged 1 commit intomui:masterfrom
Conversation
Netlify deploy previewhttps://deploy-preview-43745--material-ui.netlify.app/ Bundle size report |
|
cc @DiegoAndai who's working on #43684 |
aarongarciah
left a comment
There was a problem hiding this comment.
Looks good, but leaving the approval to @DiegoAndai in case he has some thoughts.
|
@Janpot, thanks for taking this! To clarify, AFAIU, the flaky tests are the Pigment CSS regression ones added here: #43280 On the pigment fixtures, we have some cases of the icons disappearing: https://app.argos-ci.com/mui/material-ui/builds/31895/107984073 And some with the icons appearing: https://app.argos-ci.com/mui/material-ui/builds/31900/108029355 So, my thought was that network latency was involved. Is that not the case? |
@DiegoAndai It seems to match with this PR. We were not waiting for the icon font to load before taking the screenshot. |
Yes, there's a race condition between the font loading and the moment the screenshot is captured. I noticed there was setup code in the tests that waits until all fonts are loaded before actually capturing the screenshot. Only, it seems like it's waiting for a different CSS file than the one referenced in the test. I didn't deeply test this, I just corrected the faulty url. It looks like the screenshot contains the icons, so it worked at least once. If this doesn't fix the flakeyness, I plan to debug this fully, it just didn't feel justified to spend more time on it at the moment since this is a clear bug that could explain the flakeyness, with low risk of breakage, we will notice soon enough if it didn't work. |
|
Seems to not have worked https://app.argos-ci.com/mui/material-ui/builds/32164/109227549 |
|
@Janpot, my take is that we should remove them. |
Attempt at stabilizing the icons test. It looks like the referenced css has gone out of sync with the regression tests font loader.
Let's see if this does the trick.
Potentially closes #43684