[cairo] ghostcript-8.54 causes most cairo-1.2.6 ps surface tests to fail

[Gary V. Vaughan]

On Tue, Jan 23, 2007 at 10:25:00PM -0500, Behdad Esfahbod wrote:
> On Tue, 2007-01-23 at 18:01 -0600, Gary V. Vaughan wrote:
> > 
> > Hope that helps tracking down a way to allow use of gs-8.54 with the
> > cairo tests.
> 
> From a quick look it seems like the failing tests are all the XFAIL ones
> and all the tests containing text.  This is a known problem with the PS
> backend.  gs uses very different font rendering and so is very hard to
> test.  Not sure how we want to resolve this.  The tolerance value does
> not help I believe.

Thanks for taking the time to look at these.

I wonder then whether the PS backend should have its own reference
images to compare against, along with shipping images from the most
commonly used 2 or 3 releases of ghostscript.  At configure time, cairo
could check what version of ghostscript is going to be used for the
ps surface tests, and if it is not one of the flavours supported by
the shipped reference images, configure should warn that the PS tests
will not be performed -- and perhaps add a list of the supported
ghostscript versions.

If you decide to pursue this idea, I'd be happy to generate the
reference images for any of the versions of ghostscript I have access
to (7.07, 8.16 and 8.54 at least).

Are the PDF surface test failures I posted about a similar known problem
with the current testsuite?

Cheers,
	Gary
-- 
Gary V. Vaughan (gary at thewrittenword.com)