[cairo] Zero length segment and SVG

[Jeff Muizelaar]

On Tue, Jun 27, 2006 at 03:36:03PM +0200, Carl Worth wrote:
> On Sun, 25 Jun 2006 16:31:42 -0400, Jeff Muizelaar wrote:
> > I think it would be good to add or modify an exisiting testcase to
> > exercise the closed-dashed-path-missing-leading-cap bug that Keith Wells
> > reported in the "cairo stroke problem with potential patch" thread.
> 
> OK, so there are still some things to look into, but I don't think we
> need to make this patch block on that first.
> 
> > I'd also like to look into the behaviour of degenerate dashed
> > sub-paths to make sure we have consistent behaviour.
> 
> I ran through things a bit as I wrote the documentation and I feel
> pretty good about the ways things are. We've got the test case you
> provided for deliberate degenerate paths, and I did a bit more
> interactive testing of zero-length dash segments (both round and
> square) via cairo-demo/X11/cairo-spline. Everything I've seen so far
> seems to be matching the documented semantics just fine, (except for
> the failures in the new test case of course).

The patches as merged don't draw degenerate dashed paths correctly.

i.e.

set_cap (ROUND)
set_dash ([2, 2], 0)
move_to (5, 5)
line_to (5, 5)
stroke ()

should draw a dot but doesn't.

I have fixes for this and some more extensive test cases for the above
and for the other problem reported by Keith. I originally had a patchset
that had the fixes and test integrated but I'll redo them on top of
current HEAD when I get the chance.

-Jeff