[cairo] Last couple of items on the 1.6 roadmap

[Carl Worth]

[I drafted this a few days ago, but apparently forgot to send it out.]

With the 1.5.12 snapshot out, I just updated cairo's roadmap:

	http://cairographics.org/roadmap/

And happily, 13 of the 15 items on the 1.6 roadmap are complete. It's
good we're close since "February 2008" is the date we have listed
there, and there are only a couple of hours left of that, (in my
timezone anyway).

And I think we are really close. One of the items is that we simply
need to make a pixman release, (need that for the CAIRO_EXTEND_PAD
fix, for example). I just want to add some macros and functions to
announce the pixman version, and then also adopt the same
version-numbering scheme that cairo has. And then I think that's
ready.[*]

Then, the other item left is supporting arbitrary X server visuals,
(including pseudocolor). I had the pleasure to sit down with Keith
Packard today and he walked me through much of the code needed to do
this, and that's nearly complete. About the only part missing is
color-cube allocation, (and that explains why Keith appeared earlier
on this list explaining how to do color cubes---I had to leave his
house this afternoon before he had the chance to explain that to me).

And then obviously, there are things that aren't on that roadmap, but
that form part of every release. That's mostly testing, (regression
testing with "make test" and "make perf" and also testing with "make
check-valgrind" to ensure we're not leaking anything).

We generally want to have no test-suite failures across the entire
test suite for all supported backends. Currently I'm only really
seeing failures in 7 different tests and in only the PDF and PS
backends as follows:

	pdf: clip-operator, operator-clear, operator-source,
	     push-group, bitmap-font

	ps: clip-operator, degenerate-path, miter-precision,
	    operator-source, push-group

Adrian's been looking into these and it looks like the 4 tests with
"operator" in the name might be a single bug. And the push-group
failure for PDF is definitely because of a poppler bug, (hard to hid
with a reference image since poppler fails differently in the case of
an integer offset case being applied). And the remainder might not be
that significant, (I know fairly well what's happening with
miter-precision, but I'd still like to understand the bitmap-font and
degenerate-path failures better).

Meanwhile, Adrian also just pointed out to me that extend-pad will be
failing for PS and SVG as soon as I remove it from the XFAIL list. So
that's something else that we'll need to fix.

Anyway, it's definitely exciting to be so close.

Happy hacking!

-Carl

[*] Fixing pixman to support accept coordinates larger than 16 bits
would be really nice too, but we can follow up with that later, (and
do a cairo 1.6.2 for that).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.cairographics.org/archives/cairo/attachments/20080304/ecbac84f/attachment.pgp