641506 - Seams that do not go away after panning in some pages

Status: VERIFIED FIXED

(Firefox for Android Graveyard :: Panning/Zooming, defect)

Description

[Matt Brubeck (:mbrubeck)]

Since bug 634397 and its dependencies landed, I see seaming while panning (expected) but sometimes the seam does not disappear when panning stops (unexpected).  Unfortunately I do not have a consistent repro case.

Comment 1

[Matt Brubeck (:mbrubeck)]

Attachment: screenshot

This screenshot from tarend (see the horizontal line of repeated pixels near the middle of the image) is typical of what I'm seeing.

Comment 2

[Matt Brubeck (:mbrubeck)]

Attachment: Galaxy Tab screenshot

Here's another example.  See the line in the middle of the last paragraph that begins "imagine now...".  It's hard to see, but if you blow up this screenshot and move that line up, you'll find that it comes from the first line in the second paragraph, "The next wave..."

I've also seen examples where lines of pixels where omitted, rather than added.

I got this by panning down this page after double-tapping in the content to reflow zoom:
http://www.fastcompany.com/1736602/the-future-of-the-tablet-and-it-isn-t-the-ipad-2

Comment 3

[Matt Brubeck (:mbrubeck)]

Added logging to see whether we are calling redraw() when this happens.  It appears that we *are* calling it as expected, but the seam is still present.

Comment 5

[Brad Lassey [:blassey] (use needinfo?)]

Attachment: put the redraw in a timeout

putting the redraw in a timeout seems to make this better in my testing, which leads me to believe that we have a race somewhere. Matt, can you confirm?

Comment 6

[Mark Finkle (:mfinkle) (use needinfo?)]

Comment on attachment 519597 [details] [diff] [review]
put the redraw in a timeout

>+          win.setTimeout(
>+            function() {
>+              winCwu20.redraw();
>+            }, 10);

Use 0, which basically is enough to "fix" races and is actually 10 in Mozilla

Let's get more verification that this fixes the problem.

Comment 7

[Matt Brubeck (:mbrubeck)]

Comment on attachment 519597 [details] [diff] [review]
put the redraw in a timeout

I was not able to reproduce the problem with this patch applied.

Comment 8

[Mark Finkle (:mfinkle) (use needinfo?)]

Comment on attachment 519597 [details] [diff] [review]
put the redraw in a timeout

r+ with the 10 -> 0 change

Comment 9

[Brad Lassey [:blassey] (use needinfo?)]

pushed https://hg.mozilla.org/mobile-browser/rev/90d505ee8f9b

Comment 10

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Verified:

Mozilla/5.0 (Android; Linux armv71; rv2.0b13pre) Gecko/20110316 Firefox/4.0b13pre Fennec/4.0b6pre
Device: Droid 2 
OS: Android 2.2