445410 - Regression in filter rendering inside transform

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

See attached testcase. When you scroll up and down, the filtered rectangle is not repainted correctly. This seems to be a recent regression, so I'm guessing bug 418201. The testcase was reduced from bug 431984.

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: testcase

Comment 2

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I'm pretty sure this is https://bugs.freedesktop.org/show_bug.cgi?id=16469. I'll see if I can make any progress on that.

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: better testcase

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: testcase reference

better testcase reference

Comment 5

[Robert O'Callahan (:roc) (email my personal email if necessary)]

To look closely at this problem via Gecko, break where nsSVGFilterFrame::FilterPaint calls nsSVGUtils::CompositeSurfaceMatrix. In attachment #329764 [details] we have a 200x400 image surface that's all blue (actually there's an extra row and column that are transparent but ignore that), it has a device offset of -20,-40 because filter space is 10% bigger along each edge (SVG default) but we've computed a tight bounding box and created the temporary surface with the right size and positioned it appropriately in filter space. We then go ahead and draw that with the 'fini' transform which maps filter space to device space.

Comment 6

[Jeff Muizelaar [:jrmuizel]]

The patch attached to https://bugs.freedesktop.org/show_bug.cgi?id=16469 fixes this.

Comment 7

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Thanks Jeff! Can we land that patch in our cairo now, or do we need to wait? I'd like to know if it fixes the issue in bug 431984.

Comment 8

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

We could, I'd rather upstream it first though -- I'm trying to get bug 446323 landed and would prefer to avoid having another patch (albeit a tiny one).

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Great, this is mostly fixed now. However, the reftest doesn't pass because there's a blurry edge. This is probably the same rounding issue affecting bug 445079.

Comment 10

[Tallak tveide]

hi. I am not sure if this is the exact same issue, but it might be as it regards the use of filters inside transformed groups in SVG documents.

Problem summary: When a filter is used within a group that is transformed, it is not painted at the correct position.

I am adding a test case to show the problem.

Comment 11

[Tallak tveide]

Attachment: SVG show problem with transformed filtering. Move the mouse over the rectangles to see the problem.

SVG show problem with transformed filtering. Some more comments inside the file.

Comment 12

[Robert Longson [:longsonr]]

Comment on attachment 346408 [details]
SVG show problem with transformed filtering. Move the mouse over the rectangles to see the problem.

This is bug 443114

Comment 13

[Robert O'Callahan (:roc) (email my personal email if necessary)]

This bug was fixed by Jeff's cairo patch.