[cairo] Bug in PDF output with independent path/pen transformations

[Baz]

On 31/10/2007, Carl Worth <cworth at cworth.org> wrote:
> The test can be viewed here:
>
>         http://cairographics.org/~cworth/images/degenerate-pen.c
>
> and the desired result as a png here:
>
>         http://cairographics.org/~cworth/images/degenerate-pen-ref.png
>
> This test accidentally exposed a new bug in the PDF backend. The
> result I'm getting is as follows:
>
>         http://cairographics.org/~cworth/images/degenerate-pen-pdf-argb32-out.pdf
>
> Both acroread and poppler agree that that third rectangle is seriously
> skewed, but it really should be axis-aligned.

I'm presuming I've not grabbed some later pdf here, but this looks
much worse than you described in Preview (on Leopard - which btw is
free of the font embedding bugs that we had problems with in Tiger).

I see what appears to be a wide top-to-bottom vertical bar covering
the first rectangle, a wide left-to-right horizontal bar covering the
second, and a diagonal bar with one edge running through the top left
of the third, with no apparent lower right edge. At least, those
positions would be consistent with the fill I see and bear some
relation to the pens - its a bit late for me to start digging into the
test tonight to confirm that.

The flourish sample works fine.

> Adrian, care to work some magic to diagnose this one?
>
> I was a bit surprised to find a bug here, since I had successfully
> tested independent transformations of the path and pen here:
>
>         http://cairographics.org/~cworth/images/flourish.c
>         http://cairographics.org/~cworth/images/flourish.png
>         http://cairographics.org/~cworth/images/flourish.pdf
>
> -Carl
>
> _______________________________________________
> cairo mailing list
> cairo at cairographics.org
> http://lists.cairographics.org/mailman/listinfo/cairo
>
>