[cairo] stroking tests and fixes

[Jeff Muizelaar]

On Fri, Aug 04, 2006 at 05:10:10PM -0700, Carl Worth wrote:
> On Thu, 3 Aug 2006 16:39:52 -0400, Jeff Muizelaar wrote:
> > Any thoughts Carl?
> 
> Sorry for the delay Jeff, and thanks for the reminder.
> 
> I've fetched the changes, and they all look very good! You did an
> excellent job of adding all the test cases first and then fixing the
> bug, (a two-liner no less), which is really appreciated.
> 
> And I just about pushed this out already. But I did notice that the
> SVG backend is failing the new degenerate-path case. Can someone take
> a close look at the images below and decide if this might be a cairo
> bug in generating the PDF, a librsvg bug in parsing/rendering the PDF,
> or perhaps a cairo bug in rendering what librsvg hands it.

It looks like it is a librsvg bug. The svg that is generated looks sane
to me. Batik has a slightly better interpretation of it, except it
doesn't draw the "M 31 31 Z" paths. I also checked a firefox nightly
build it and seems like it would also be correct if the changes were
made.

-Jeff