308981 - X error on partial expose of window when logo not in dirty region

Status: RESOLVED WORKSFORME

(Core :: SVG, defect)

Description

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

[ Filing in SVG for lack of a cairo component. ]

I'm seeing an X error due to 0 width and height parameters to XCreatePixmap when
doing a partial expose of the window in a W3COrg.svg (a diagram of the W3C org
chart).  This is a regression from a few months ago, when I used that chart in a
demo I have at the W3C AC Meeting.  I'll attach a simplified testcase.

Steps to reproduce:
 * load attached testcase
 * drag another window over the browser such that it does a partial repaint of
an area that does not include the little piece of the W3C logo that's left in
the simplified testcase (the shadow of the C).

Actual results: X error (which calls |exit|).  I'll attach the stack.

Expected results: No program termination.

Testing in a trunk build from yesterday afternoon.

Comment 1

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

Attachment: testcase causing X error on partial expose not including the path

Comment 2

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

Attachment: stack of X error

I think the expose may need to be to the right of the logo.  I've been testing
by dragging a window horizontally across, so that the expose is the full height
of the window but covers a horizontal segment to the right of the logo.

Comment 3

[tor]

Branch or trunk?

Comment 4

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

(In reply to comment #3)
> Branch or trunk?

comment #0 says:
> Testing in a trunk build from yesterday afternoon.

Comment 5

[tor]

Can you reproduce this on the branch or MOZILLA_1_8_SVG_BRANCH?

Comment 6

[Boris Zbarsky [:bzbarsky]]

This seems to work for me on current trunk... David, are you still seeing this?

Comment 7

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

Cairo is a lot better about protecting against creating 0-width/height xlib drawables now than it was when this was filed; this is probably WORKSFORME at this point, assuming David can't reproduce it.