Bug 36604 – [Qt] Clip test case failure (X11 paintengine)

• Component: Layout and Rendering
• Status: RESOLVED DUPLICATE

[qi]

Test case: http://waplabdc.nokia-boston.com/browser/users/ligman/canvas%20tests/complex%20shapes/clip.html failed on Qt, but it works on Safari

[Laszlo Gombos]

According to Qi, a failure on http://www.netzgesta.de/reflex/ from bug 26256 could be related. Let's keep it in scope until it is proven otherwise.

[Dirk Schulze]

I can confirm this. It's an old issue. IIRC it was related to problems with the XServer.

[qi]

I found two issues on this case, but somehow we need 23817 fix first. So currently I can't check in my fix for this case.

1> At CanvasRenderingContext2D::clip(), we suppose to call m_path.closeSubpath(); This is based on spec "Open subpaths must be implicitly closed when computing the clipping region"

2> At GraphicsContextQt.cpp clearRect(), we suppose to calculate the intersection between rect and cliprect. This is based on spec "..must clear the pixels in the specified rectangle that also intersect the current clipping region to a fully transparent black..."

[Simon Hausmann]

Indeed, some platform information is needed here. It works for me with -graphicssystem raster, but it doesn't work for me with the x11 painting backend. Qi, what platform do you see this failure on?

I think it would be good to create a platform independent trace - using the trace graphics system. Then we can file this as a bug against Qt, which it most likely is.

[qi]

I am working on X11.

[Andreas Kling]

*** This bug has been marked as a duplicate of bug 23817 ***