716347 - feDisplacementMap test shows unexpected colors

Status: VERIFIED FIXED

(Core :: SVG, defect)

Description

[Cameron McCormack (:heycam)]

+++ This bug was initially created as a clone of Bug #619967 +++

http://dev.w3.org/SVG/profiles/1.1F2/test/harness/htmlObjectMiniApproved/filters-displace-02-f.html currently fails due to some strange colors near the edge of the displaced image.

Comment 1

[Markus Stange [:mstange]]

I don't know why but my patches in bug 924102 and bug 924103 fix this test. I added some bilinear interpolation for fractional source locations, and maybe I've changed the rounding in some way.

Comment 2

[Markus Stange [:mstange]]

Resolving as fixed by bug 924103, even though the test now fails because of the new feDisplacement taintedness restrictions (bug 941887). But before bug 941887 landed, this was fixed.

Comment 3

[Manuela Muntean [Away]]

Attachment: screenshots.zip

I'm not sure I was able to reproduce this issue: I've tried with both the Nightly from 2012-01-09 and Firefox 4. Does anyone have any thoughts/suggestions?

Please see the attached screenshots for more details (there is also the result I'm seeing on beta)

Comment 4

[Robert Longson [:longsonr]]

28 beta 6 seems completely broken but not in the way that Cam saw as far as I know. Nightly and Firefox 4 look OK.

Comment 5

[Markus Stange [:mstange]]

(In reply to Robert Longson from comment #4)
> 28 beta 6 seems completely broken but not in the way that Cam saw as far as
> I know.

It's broken due to bug 941887, intentionally.

> Nightly and Firefox 4 look OK.

This is surprising to me. I recall a purple fringe at the end of the gradient, but it's not visible in these screenshots. Maybe it only showed on Mac?

Comment 6

[Robert Longson [:longsonr]]

Yes, the fringe was mac only.

Comment 7

[Manuela Muntean [Away]]

Attachment: Screen Shot beta.png

I was able to reproduce the issue shown here https://bug619967.bugzilla.mozilla.org/attachment.cgi?id=498401 with Firefox 4 on Mac OS X 10.8.5.

On the same machine, with Firefox 28 beta 6 I see the same behavior as on Windows (also shown in this attachement). Considering beta it's broken due to bug 941887 intentionally, is it safe to call this bug verified on 28 branch?

Comment 8

[Markus Stange [:mstange]]

Yes.