732119 - MAPLE: parts of background never drawn on graph server

Status: RESOLVED DUPLICATE of bug 746120

(Firefox for Android Graveyard :: General, defect, P2)

Description

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

Attachment: screenshot

1. go to http://graphs-new.mozilla.org/graph.html#tests=[[16,65,20],[16,53,20],[16,27,20],[16,52,20],[16,63,20]]&sel=none&displayrange=7&datatype=running
2. while the page is loading, continually zoom out

Expected: things load according to relative position in the zoom level
Actual: things are loaded according to relative position of original zoom level

Note:
1. nexus Captivate, Maple build 3/1/2012

Comment 1

[Joe Drew (not getting mail)]

This page seems to load the same broken way regardless of zooming.

Comment 2

[Tony Chung [:tchung]]

nom for triage.   retest this after more changes land in maple

Comment 3

[Benoit Girard (:BenWa)]

I feel like this would be solved by the recent changes to viewport.

Comment 4

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

Unfortunately it is not fixed, and I don't know offhand what's wrong. It doesn't seem like a viewport problem, maybe layout?

Comment 5

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

Panning is an issue on this page this time.  The viewport only shows a portion of the page, and you cannot pan in either directions.  As kats stated in comment 4, this bug is not fixed.

Comment 6

[Chris Lord [:cwiiis]]

Copying my comment from bug 737021:

"I think this may be related to the android-specific code in CompositorParent - it finds the root-scrollable frame and applies the transform to the associated layer, but perhaps there are situations where this is the incorrect layer?"

Comment 7

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

(In reply to Chris Lord [:cwiiis] from comment #6)
> Copying my comment from bug 737021:
> 
> "I think this may be related to the android-specific code in
> CompositorParent - it finds the root-scrollable frame and applies the
> transform to the associated layer, but perhaps there are situations where
> this is the incorrect layer?"

The "CompositorParent" issue seems to have popped up in recent chatter on IRC as well. Is there a specific bug for the "CompositorParent" issue?

Comment 8

[Erin Lancaster [:elan]]

If this bug is relegated to the graph server and we are not seeing this reported on other pages, let's not block on this.

Comment 9

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

(In reply to Chris Lord [:cwiiis] from comment #6)
> Copying my comment from bug 737021:
> 
> "I think this may be related to the android-specific code in
> CompositorParent - it finds the root-scrollable frame and applies the
> transform to the associated layer, but perhaps there are situations where
> this is the incorrect layer?"

Chris, so is this related to bug 732971?

Comment 10

[Chris Lord [:cwiiis]]

(In reply to Brad Lassey [:blassey] from comment #9)
> (In reply to Chris Lord [:cwiiis] from comment #6)
> > Copying my comment from bug 737021:
> > 
> > "I think this may be related to the android-specific code in
> > CompositorParent - it finds the root-scrollable frame and applies the
> > transform to the associated layer, but perhaps there are situations where
> > this is the incorrect layer?"
> 
> Chris, so is this related to bug 732971?

I wouldn't have thought so, but possibly?

I think it'd fix this if we could apply the transform to the root frame instead of the root scroll frame, but doing so causes odd redrawing issues (likely due to calculated visibility in layers?)

I need to experiment more with this.

Comment 11

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

Martijn - can you try to create a minimal testcase?

Comment 12

[Aaron Train [:aaronmt]]

Is it possible that this is due to the overflow: hidden styling - I put up a test-case in bug 746120 where I see the same thing.

I also see this on duckduckgo.com when pinch-zooming inwards. If it is than my bug 746120 is a duplicate of this.

Comment 13

[Martijn Wargers (dead)]

Yeah, I agree, the page has overflow:hidden on the html element, that makes it basically the same as bug 746120.