[Webkit-unassigned] [Bug 9268] Quirksmode: Backgrounds on table rows

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=9268





--- Comment #4 from Daniel Bates <dbates at webkit.org>  2009-12-02 19:35:14 PST ---
Created an attachment (id=44202)
 --> (https://bugs.webkit.org/attachment.cgi?id=44202)
Patch

Work in progress.

This patch correct the issue. However, there are some weird rendering artifacts
that may appear when you try to select text within a column in a row with a
background-image and/or the background of the selection will be some
re-positioned portion of the row's background image.

In the standalone example <https://bugs.webkit.org/attachment.cgi?id=26361>, if
you click anywhere to the right of the table you will see that the background
of the first row starts to tile on the right-most side of the "TH 3" cell.
Also, if you try to select the text "TH 3" then click somewhere to deselect the
selection, the selection rectangle will visibly disrupt the diagonal in the
background. That is, it looks like we didn't paint the same portion of
background that was under the selection. Instead, we painted some other portion
of the row background.

I suspect that the phase in x is wrong during the selection and I think the
correct phase in x should be the default (the one we calculated using
calculateBackgroundImageGeometry plus the bit phase += destRect.location() -
destOrigin), but I haven't looked at the values of destRect.location() and
destOrigin yet nor have a solid understanding how the selection code influences
what we should re-paint. It looks like we issue a repaint of only the rect of
the selection.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.