[cairo] text extents bugs in the test suite

[Keith Packard]

I am having regressions against the test suite with the new glyph
caching code and I think they're actually bugs in the old code (and,
perhaps the test suite too).

It looks like the bounding box computation for glyphs is obviously
wrong. Check out the image for the text-antialias-gray test and you'll
see that the 'b' glyph which occurs in both 'black' and 'blue' has very
faint blue 
cast in the pixel at the top of the stem. Then note that the same pixel
is not displayed for 'black'. The test places the glyph right at
-extents.y_bearing, so it should display exactly the whole glyph.

A good side effect is that I've now spent an entire day checking and
rechecking all of the metric and placement code in the new glyph
functions and I believe they're correct.

-keith

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://lists.freedesktop.org/archives/cairo/attachments/20050824/13aef2c2/attachment-0001.pgp