745934 - CSS transitions have horrible artifacts

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Vladan Djeric (:vladan)]

On my Windows laptop running Nightly 14, I am getting horrible artifacts during various CSS transitions. However, I do not get these artifacts on Beta 12 and Aurora 13.

Test sample: http://people.mozilla.com/~tglek/fosdem2012/

Comment 1

[Boris Zbarsky [:bzbarsky]]

Can you find a regression range?

Comment 2

[Vladan Djeric (:vladan)]

(In reply to Boris Zbarsky (:bz) from comment #1)
> Can you find a regression range?

Nightly from Friday morning (April 13th) doesn't have the bug:
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2012-04-13-04-02-25-mozilla-inbound/

Nightly from Saturday morning (April 14th) does have the bug:
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2012-04-14-04-02-27-mozilla-inbound/

Comment 3

[Vladan Djeric (:vladan)]

Attachment: Only ~25% of DIV shown during transition

Screenshot showing the bug: only about a quarter of two DIVs undergoing a CSS transition is shown during the animation. The other parts of the DIVs only "pop in" after the transition is complete.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

What's the regression range w.r.t mozilla-central?  That might provide additional data.

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=1587faab5e5c&tochange=4469d30b6943 is the pushlog range for your inbound regression range

Comment 6

[Lukas Blakk [:lsblakk] use ?needinfo]

[Triage Comment]
Seems a little early to be tracking this for 14, let's wait and see if a regression range and possible backout or forward fix is possible then re-nominate for tracking if something needs uplifting.

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

No, we need to track it; otherwise it could get lost.  If it's fixed we'll un-track it then.

Comment 8

[Boris Zbarsky [:bzbarsky]]

In the comment 5 range, bug 735898 and bug 744607 jump out at me.  Especially the latter.

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: testcase

This very simple static testcase shows the problem... it should be a transparent yellow rectangle with its top-left at 0,0 ... but it isn't.

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

Comment 11

[Robert O'Callahan (:roc) (email my personal email if necessary)]

This would have been a regression from changeset 007e4f53e361.

Comment 12

[Matt Woodrow (:mattwoodrow)]

Comment on attachment 615670 [details] [diff] [review]
fix

Review of attachment 615670 [details] [diff] [review]:
-----------------------------------------------------------------

::: layout/reftests/bugs/745934-1.html
@@ +5,5 @@
> +div {
> +    position: absolute;
> +	top: 200px; left: 200px;
> +    width: 600px;
> +    height: 500px;

Indenting

Comment 13

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 615670 [details] [diff] [review]
fix

Review of attachment 615670 [details] [diff] [review]:
-----------------------------------------------------------------

Bad regression. Fix only affects pages using -moz-transform-style:preserve-3d.

Comment 14

[Lukas Blakk [:lsblakk] use ?needinfo]

Comment on attachment 615670 [details] [diff] [review]
fix

[triage comment]
low/no risk to mobile, approving.

Comment 15

[(no longer active)]

Backed out for reftest failures: https://hg.mozilla.org/integration/mozilla-inbound/rev/c6316d2579e2

Comment 16

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Checked in, this wasn't the cause of the failures.

https://hg.mozilla.org/integration/mozilla-inbound/rev/48704ff71f85

Comment 17

[(no longer active)]

https://hg.mozilla.org/mozilla-central/rev/48704ff71f85

Comment 18

[gabriel]

Not sure if this should be re-opened or a new ticket opened, but when using transitioning with scale3d and the element has an outline, lots of artefacts appear

Test showcasing the bug at http://jsfiddle.net/gaby/SWJ4Z/2/

Comment 19

[gabriel]

Sorry guys.. missed the firefox14 milestone..

Comment 20

[Boris Zbarsky [:bzbarsky]]

New ticket please.  This bug was caused by another patch that landed in Firefox 14, and fixed 3-4 days after it first appeared.  So if you're not using a nightly build from one of those 3-4 days about a week ago, then you're seeing something different from this bug.

In particular, your testcase shows a problem in a nightly from 3 days ago, which certainly has the fix for this bug in it.  Do you want to file the new bug, or should I?

Comment 21

[gabriel]

I am seeing this bug as a result from upgrading from v11 to v12.0

You seem to know much more about the procedures here, so it might be better if you file it as a new bug, if it fits the criteria..

Comment 22

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Filed bug 749450. Thanks!