Closed Bug 1495962 Opened 3 years ago Closed 3 years ago

Floating table header moves and snaps back when scrolling down (should'nt move at all)


(Core :: Graphics: WebRender, defect, P2)




Tracking Status
geckoview62 --- disabled
firefox-esr60 --- disabled
firefox62 --- disabled
firefox63 --- disabled
firefox64 --- wontfix
firefox65 --- wontfix
firefox66 --- verified
firefox67 --- verified


(Reporter: mayankleoboy1, Assigned: kats)


(Blocks 2 open bugs, )


(Keywords: nightly-community, regression)


(3 files)

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:64.0) Gecko/20100101 Firefox/64.0

Steps to reproduce:

1. Enable WR
2. Go to
3.  There is a comparison table in the middle of the page
4. Scroll down from the table

Actual results:

the "floating" table header moves down, and then snaps back continously

From the looks of it, this is a APZC+WR bug

Expected results:

not so.
ni? ,as this looks like a APZ issue
Flags: needinfo?(kats)
I am scrolling using the mousepad on a laptop
A regression date would have to be before 2018-01-15.
It's broken since it was implemented:
> Emilio Cobos Álvarez — Bug 975644: Enable position sticky in table parts. r=bz
Flags: needinfo?(jan)
well, this reproduces only with WR for me.
Thanks. I can look at this when I'm done my current set of bugs. Leaving unassigned for now in case somebody else wants to do it first.
Flags: needinfo?(kats)
Marking fix-optional for 64. We could still take a patch for 65, and if it's verified and doesn't seem risky, could still take fixes for 64 as well.
Priority: P3 → P2
Assignee: nobody → kats
Attached file Minimal test case
Minimal testcase to repro. Basically, both the <tr> and the <th> are sticky-positioned, so presumably we end up applying the sticky offset twice, or something like that. With async scrolling the header ends up moving against the scroll instead of staying in place.
Pushed by
Add reftests for nested sticky-positioning r=kvark
Looks like it needs a little fuzz on win7
Flags: needinfo?(kats)
Pushed by
Add reftests for nested sticky-positioning r=kvark
Pushed by
Follow-up to fix overeager fuzzing. r=me
Pushed by
Loosen fuzz range for win 7 reftest jobs. r=me
WR only, not severe/common enough to uplift, I would say.


Confirming this issue as verified fixed on Windows 10x64 with 66.0b14 and 67.0a1 (2019-03-08)on an nVidia GTX 1060 GPU.

You need to log in before you can comment on or make changes to this bug.