[cairo] PS backend

[Carl Worth]

On Thu, 06 May 2004 03:05:03 +0100, "Rob Taylor" wrote:
> Does PDF 1.4 suffer from the same precision problems and lack of features
> (are we talking about transparency here)?

The precision problems should still be there. It's fairly easy to get
PostScript to get the wrong answer when stroking --- just make the
stroke very wide and the radius of curvature very tight. Visible
ploygonization appears on the contour that exceeds the specified
flatness. The interaction of the specification of flattenpath and stroke
seem to require this behavior.

And I believe PDF follows the lead of PostScript with respect to these
operators.

As for "lack of features", PostScript is definitely missing
transparency. I would hope PDF 1.4 isn't missing anything that cairo
needs, as that specification has kitchen sink syndrome to some extent.

>                                          If not, wouldn't it be better
> to produce PDF 1.4 from a high-level hook-in? This would hopefully stop
> cairo from producing overy dumb and huge PDF files..

The low-level backend in cairo is geometric in nature, (lists of
trapezoids). Using it should not result in dumb and huge PDF files.

This is in contrast to the current PostScript backend implementation
which only hooks in after the image backend has already rasterized the
geometry. That's where the dumb and huge files come from.

-Carl