Closed Bug 1375269 Opened 7 years ago Closed 7 years ago

Graphical artifacts left behind by WhatsApp Web "fly-in" load animation

Categories

(Core :: Web Painting, defect)

defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 1373479

People

(Reporter: dholbert, Unassigned)

Details

(Keywords: regression)

Attachments

(1 file)

Do all of the STR in a fullscreen browser.

STR:
If you're already logged in to WhatsApp Web:
 1. Visit (or reload) https://web.whatsapp.com/
...otherwise, log in:
 1. Visit https://web.whatsapp.com/
 2. Log in using a whatsapp client on your phone)
 3. After the login process completes, watch the fly-in animation

ACTUAL RESULTS:
The "fly-in" loading animation leaves behind a bunch of graphical artifacts. These stick around until I do something that forces a repaint (e.g. change focus).

EXPECTED RESULTS:
No graphical artifacts.

Regression range:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=958d2a5d10091401fd5e900e8e063d21940c137e&tochange=7f894f791cdf170d788507d0eff30024ce699523

(This is the smallest range that mozregression would give me, I think because the intermediate builds had unrelated bustage which was eventually fixed in the tipmost cset in this range)

In that range, Matt Woodrow's changes for bug 1359709 & bug 1361970 look most suspiciously related.
Here's a screencast showing the bug.  (starting from a blank tab, in which I load WhatsApp Web. I popped up a text editor to redact the contents of my conversations, but you can still see the artifacts left behind at the edges of the screen at the end of the video.
(The artifacts are all the weird white semitranparent blocks on each side of the screen -- everything outside of the main floating "dialog" that has my cartoon avatar on its upper left corner.)
(In reply to Daniel Holbert [:dholbert] from comment #0)
> Do all of the STR in a fullscreen browser.

(Sorry for not clarifying -- this ^^ is simply to be sure that your browser is large enough to show the edges of the page where artifacts appear. If your browser isn't large enough, then all you see is the dialog and there are no artifacts left in view anywhere.)

> In that range, Matt Woodrow's changes for bug 1359709 & bug 1361970 look most suspiciously related.

I just confirmed that a local backout of Bug 1361970 (https://hg.mozilla.org/mozilla-central/rev/68932b1a5b71 ) is sufficient to fix this bug here.  --> Marking as regression from that bug.
Blocks: 1361970
Flags: needinfo?(matt.woodrow)
(I looked at other regressions from that bug -- it looks like this might be a dupe of one of them.  I tested bug 1373479's fix to see if it helps here -- but it does not help here.  BUT, it also doesn't seem to help with that bug's own testcases, so maybe it's just incomplete? I'll leave some comments over there...)
Whiteboard: [possibly dupe of bug 1373479?]
OK, mattwoodrow's updated patch on bug 1373479 [1] *does* indeed fix this, based on my testing. Hooray! Marking as dupe.

[1] https://bugzilla.mozilla.org/attachment.cgi?id=8880176
Status: NEW → RESOLVED
Closed: 7 years ago
Flags: needinfo?(matt.woodrow)
Resolution: --- → DUPLICATE
Whiteboard: [possibly dupe of bug 1373479?]
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: