1659496 - Bugzilla dashboard hover effect is broken

Status: RESOLVED DUPLICATE of bug 1661316

(Core :: Layout, defect)

Description

[Miko Mynttinen]

Regressed by bug 1656418.

This seems resolution dependent, I was only able to reproduce this bug on my Mac with maximized window (2560x1440 resolution, scale 2.0).

Comment 1

[Miko Mynttinen]

Attachment: hover.mov

Comment 2

[Hiroyuki Ikezoe (:hiro)]

On my environment, no animation is involved on hover.

Miko, can you please provide a reduced test case? (Is there no animation in the first place?)

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1656418

Comment 4

[Miko Mynttinen]

Attachment: 1659496.html

Comment 5

[Hiroyuki Ikezoe (:hiro)]

Thank you, Miko. But it's still super unclear how this issue is related to the partial pre-render. There seems no animation involved and the hovered element size looks quite small....

Comment 6

[Hiroyuki Ikezoe (:hiro)]

Okay, main#bugzilla-body element has will-change:transform property (but there is no transform property specified), and the element is big enough so with partial pre-render it gets layerized, presumably this layerization is the cause of the issue. That means I think it's a pre-existing issue somewhere...

Comment 7

[Kartikaya Gupta (email:[e-mail])]

Maybe worth bisecting with the pref set to true?

Comment 8

[Hiroyuki Ikezoe (:hiro)]

Indeed!

Please someone try to narrow down regression range with layout.animation.prerender.partial=true. (It may be possible that it's been a long standing issue).

Comment 9

[Miko Mynttinen]

(In reply to Hiroyuki Ikezoe (:hiro) from comment #8)

Indeed!

Please someone try to narrow down regression range with layout.animation.prerender.partial=true. (It may be possible that it's been a long standing issue).

This resulted in https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=933c9f34edfab8d5cf2a5389304cf3708889eb1c&tochange=96cd358e3eedc981b0bf4e00294ed73ce099d1c0

Where bug 1649270 enabled WebRender on my Mac GPU, which suggests that this might be WebRender related.

I then ran mozregression with WR pref flipped on, and got this regression range:
4:55.16 INFO: Last good revision: 4d585c7edc7683e4b35eca6b18c9a646a1b8a78d (2019-11-05)
4:55.16 INFO: First bad revision: 96b58f95ed7333672e6dba134d091015328d299b (2019-11-06)
4:55.16 INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=4d585c7edc7683e4b35eca6b18c9a646a1b8a78d&tochange=96b58f95ed7333672e6dba134d091015328d299b
0:06.32 INFO: There are no build artifacts for these changesets (they are probably too old).

Comment 10

[Hiroyuki Ikezoe (:hiro)]

Thank you Miko! It's super helpful to identify the cause. But we are still on the way.

The range is including bug 1587713. Before bug 1587713 partial pre-render had been available on WebRender regardless of whether the pref value is false since bug 1508522.

So if the root cause landed during the window between bug 1508522 and bug 1587713, it's quite hard to track down..

Comment 11

[Hiroyuki Ikezoe (:hiro)]

(In reply to Hiroyuki Ikezoe (:hiro) from comment #10)

Thank you Miko! It's super helpful to identify the cause. But we are still on the way.

The range is including bug 1587713. Before bug 1587713 partial pre-render had been available on WebRender regardless of whether the pref value is false since bug 1508522.

Well, to be precise, it's full pre-render. Anyways the result should be the same in this bug's case.

Comment 12

[Hiroyuki Ikezoe (:hiro)]

Dropping bug 1656418, it didn't regress actually.

Comment 13

[Hiroyuki Ikezoe (:hiro)]

Miko, bug 1661316 has been landed on m-c, I hope it also fixes this issue. Would you mind trying to see whether this issue is still reproducible or not?

Comment 14

[Miko Mynttinen]

(In reply to Hiroyuki Ikezoe (:hiro) from comment #13)

Miko, bug 1661316 has been landed on m-c, I hope it also fixes this issue. Would you mind trying to see whether this issue is still reproducible or not?

I cannot reproduce this with the latest Nightly, thank you Hiro!

Comment 15

[Ciprian Georgiu, Desktop QA]

Since this has been closed as duplicate of bug 1661316, and the regression has been already provided, I think we can remove the regressionwindow-wanted keyword.