967074 - CSS repeating-linear-gradient rendering bug

Status: NEW

(Core :: Graphics: WebRender, defect)

Description

[Binyamin]

Attachment: WebKit vs Firefox

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/32.0.1700.102 Safari/537.36




Expected results:

CSS3 repeating-linear-gradient rendering bug, the stops should repeat every 3 pixels.
Also the visual appearance diffs on any browser.

Test cases:
http://jsfiddle.net/laukstein/K6XVw/2/ fine on IE11
http://jsfiddle.net/laukstein/K6XVw/14/ fine on Chrome 32, Safari 5.1.7

Related issues
https://code.google.com/p/chromium/issues/detail?id=155812
https://code.google.com/p/chromium/issues/detail?id=314383

Comment 1

[Binyamin]

I disagree with https://bugzilla.mozilla.org/show_bug.cgi?id=644444#c23 that Firefox rendering is the must correct.

1. Firefox has "transparency" bug like earlier had with CSS "border-color" - compare the visual appearance of http://jsfiddle.net/laukstein/K6XVw/15/ and http://jsfiddle.net/laukstein/K6XVw/2/. http://jsfiddle.net/laukstein/K6XVw/2/ must be as excatly http://jsfiddle.net/laukstein/K6XVw/15/.
2. Check http://jsfiddle.net/laukstein/K6XVw/17/ and http://jsfiddle.net/laukstein/K6XVw/15/ - they must return visualy accurately same result, but doesn't.

Comment 2

[sjw]

Please move this to
Product: Core
Component: Layout: View Rendering

Comment 3

[Liz Henry (:lizzard) (relman/hg->git project)]

It looks different in every browser so I'll confirm this bug as NEW for now, though I'm not sure what the correct implementation would be. 

dbaron, what do you think?

Comment 4

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

Attachment: first testcase, from http://jsfiddle.net/laukstein/K6XVw/2/

Comment 5

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

Attachment: second testcase, from http://jsfiddle.net/laukstein/K6XVw/14/

Testcases should *always* be attached to bugs so we can have a conversation about them that will make sense in the future (in case the URLs referenced change or disappear).

Comment 6

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

I'm seeing rather different results from the screenshot on Linux.  Either way, this is a Graphics bug... though I'm not sure exactly how cross-platform these results should be *expected* to be.  I don't think any relevant specifications describe the pixel-rounding rules needed to get interoperable behavior here, and I'm not sure if platform APIs are powerful enough to give us that level of consistency with good performance at the same time.

Comment 7

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

Attachment: screenshot on Linux

trunk from a few days ago, about:support says:

 Graphics
Adapter Description	Tungsten Graphics, Inc -- Mesa DRI Intel(R) Sandybridge Mobile
Device ID	Mesa DRI Intel(R) Sandybridge Mobile
Driver Version	3.0 Mesa 8.0.4
GPU Accelerated Windows	2/2 OpenGL (OMTC)
Vendor ID	Tungsten Graphics, Inc
WebGL Renderer	Tungsten Graphics, Inc -- Mesa DRI Intel(R) Sandybridge Mobile
windowLayerManagerRemote	true
AzureCanvasBackend	cairo
AzureContentBackend	cairo
AzureFallbackCanvasBackend	none
AzureSkiaAccelerated	0

Comment 8

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

And the Linux behavior looks the same, to my eyes anyway, if I drop the layers.acceleration.force-enabled pref and have about:support showing:

Graphics
Adapter Description	Tungsten Graphics, Inc -- Mesa DRI Intel(R) Sandybridge Mobile
Device ID	Mesa DRI Intel(R) Sandybridge Mobile
Driver Version	3.0 Mesa 8.0.4
GPU Accelerated Windows	0/1 Basic
Vendor ID	Tungsten Graphics, Inc
WebGL Renderer	Tungsten Graphics, Inc -- Mesa DRI Intel(R) Sandybridge Mobile
windowLayerManagerRemote	false
AzureCanvasBackend	cairo
AzureContentBackend	cairo
AzureFallbackCanvasBackend	none
AzureSkiaAccelerated	0

Comment 9

[Gregory Pappas [:gregp]]

I think the second test case looks the same in Firefox, Safari, and Chromium now. The first test case still exposes some subtle differences.

Comment 10

[Gregory Pappas [:gregp]]

Attachment: Comparison.png

Comment 11

[Gregory Pappas [:gregp]]

Attachment: Comparison.webm