951541 - animateMotion-flattening-1.svg fails on Windows 8

Status: NEW

(Core :: SVG, defect)

Description

[Brian Birtles (:birtles)]

The reftest I created for bug 946540 fails consistently on Windows 8. Reftest logs forthcoming.

Comment 1

[Brian Birtles (:birtles)]

Reftest failures:

https://tbpl.mozilla.org/?tree=Try&rev=ba71af2814ef
https://tbpl.mozilla.org/?tree=Try&rev=a74f28d6fc4d

Note that animateMotion-flattening-1.svg has yet to be pushed (waiting to make sure the fails-if clause pleases try server)

Comment 2

[Daniel Holbert [:dholbert]]

The failure is in the extreme bottom strip of pixels

Testcase has:
 rgb(1,255,1)
whereas reference has:
 rgb(0,255,0)

So it's *very* close. Might be worth using fuzzy-if instead of fails-if, since it's so close. (That way, we'll be able to catch actual failures in this test on Win8.)

Still, odd that that bottom strip isn't matching. Does this reftest do anything near the bottom of the window?

Comment 3

[Brian Birtles (:birtles)]

(In reply to Daniel Holbert [:dholbert] from comment #2)
> The failure is in the extreme bottom strip of pixels
> 
> Testcase has:
>  rgb(1,255,1)
> whereas reference has:
>  rgb(0,255,0)
> 
> So it's *very* close. Might be worth using fuzzy-if instead of fails-if,
> since it's so close. (That way, we'll be able to catch actual failures in
> this test on Win8.)

Done: https://hg.mozilla.org/integration/mozilla-inbound/rev/fb1b9398aa86

> Still, odd that that bottom strip isn't matching. Does this reftest do
> anything near the bottom of the window?

It moves a red rectangle from roughly the top-left of the viewport to the bottom/bottom-right so it could well be an artifact from moving the rectangle which may have been layerized.

Comment 4

[Brian Birtles (:birtles)]

(In reply to Brian Birtles (:birtles) from comment #3)
> (In reply to Daniel Holbert [:dholbert] from comment #2)
...
> > Still, odd that that bottom strip isn't matching. Does this reftest do
> > anything near the bottom of the window?
> 
> It moves a red rectangle from roughly the top-left of the viewport to the
> bottom/bottom-right so it could well be an artifact from moving the
> rectangle which may have been layerized.

Oh, except that the whole test is paused and seeked. So maybe that has nothing to do with it.