Closed Bug 1213156 Opened 9 years ago Closed 7 years ago

Poor performance on Travis pages

Categories

(Core :: Layout, defect)

Product:
Core
Component:
Layout
Platform:
x86_64
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED
Tracking Flags:
Tracking Status
firefox44 --- affected

People

(Reporter: marco, Unassigned)

References

Details

(Keywords: perf, Whiteboard: [perf:p3][perf:profile-needed])

I often need to kill the script running in a Travis page to get the browser responsive again.
Could you please provide more information on this bug? Ie. Steps, Screenshots, expected and actual resuts, etc.
Flags: needinfo?(mar.castelluccio)
To reproduce, simply open a Travis build results page (e.g. https://travis-ci.org/mozilla/pluotsorbet/builds/68736884) and do something (e.g. scroll). It should be fast, but it's pretty slow.
Flags: needinfo?(mar.castelluccio)
User Agent Mozilla/5.0 (X11; Linux x86_64; rv:47.0) Gecko/20100101 Firefox/47.0 Build ID 20160127030236 Reproduced this on Firefox 44 and Nightly 47.
Component: Untriaged → Layout
Product: Firefox → Core
I experience the same issue. I'm not sure it's a Linux specific problem; a friend using a Mac has the same issue.
Adding the quantum flow label, the build overview is still slow in Firefox, and there are other application affected by something similar as well (see 'see also').
See Also: → https://webcompat.com/issues/10103
Whiteboard: [qf]
Keywords: perf
Whiteboard: [qf] → [perf:p3]
Whiteboard: [perf:p3] → [perf:p3][perf:profile-needed]
Could you please add a performance profile for investigating the issue?
Flags: needinfo?(dschubert)
After I did not have any luck reproducing any performance issues on my MacBook, I did intensive testing on the Quantum Reference Device. Using 56 and 55, I have a fair bit of jank during scrolling. However, when using 57 Beta, everything is really smooth, even on long travis build result pages. I clicked around a bit to see all UI part, but could not make out any specific thing that's slow enough to cause any notable hang. I could do a bisect and see when we ixed it, but I consider that as a win out of general Quantum performance work, which is awesome. Because of that, I'll close this as fixed. If anyone feels like we should have more information about what caused jank in the past, I can do some profiling and analysis on old versions, but I don't think that's very useful. Also, if someone still has performance issues with 57, we can reopen. :)
Status: NEW → RESOLVED
Closed: 7 years ago
Flags: needinfo?(dschubert)
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.