[lsb-discuss] LSB fonts

Jeff Licquia jeff at licquia.org
Tue Nov 21 11:51:35 PST 2006


On Fri, 2006-11-17 at 07:59 -0800, Camp, TracyX E wrote:
> Part of the Xft tests I'm working on including rendering a variety of
> UTF-8 strings and doing a bitmap comparision against a reference set of
> renderings (even if we mandated fonts in LSB the tests will still need
> to use a controlled set of fonts for these tests in order to ensure this
> continues to work even if the fonts are revised).  This doesn't
> necissarily mean the fonts look good or are even correct - I only read
> english.

We should be cautious when relying on distro-provided stuff for our
testing, based on our experiences with distro-provided Xvfb.

I would feel more comfortable shipping fonts as part of the tests, since
we then have much tighter control of the input to those tests.  It's
quite possible and desireable to improve a font's rendering capabilities
(for example, improving hinting, or replacing composed code points with
unique glyphs), but this improvement would wreak havoc with our tests.

Which isn't to say that mandating a minimum font set is a bad idea; just
that I'd rather keep those tasks separate.

> Personally I'd propose simply mandating that X11 ucs fonts since these
> are already defacto in the xorg distribution and do provide full utf-8
> coverage.  The truetype/opentype fonts certain look better, but this
> might be a case of settling for what is easy to do and at least be able
> to guarantee that UTF-8 rendering will work.

In the spirit of mandating fonts, yes.  In the spirit of testing Xft,
don't we want to do some OpenType/TrueType font rendering tests?

(I'm less sure about PostScript fonts, PEX, etc., but to the extent
those make sense to support, we should probably test them too.)

Of course, if we want to test OpenType but not mandate any fonts, this
adds to the case for shipping fonts with the tests.




More information about the lsb-discuss mailing list