[XeTeX] Proper way to set up OT Features

[mskala at ansuz.sooke.bc.ca]

On Tue, 15 Feb 2011, Jonathan Kew wrote:
> Yes, you can't do lookups that involve <space>, because of how xetex
> handles word spacing.

If it doesn't work in XeTeX, there will probably be other applications in
which it also doesn't work - so changing XeTeX wouldn't solve the problem.
In general, I think it's a bad idea to try to write features that depend
on a <space> glyph; spaces really are not glyphs, and it's reasonable that
an application won't pass them to the font.

The Adobe specification does discuss word boundary detection here:
   http://www.adobe.com/devnet/opentype/afdko/topic_feature_file_syntax.html#5.f.ii

They use an "ignore" rule to block substitution when the search string is
preceded by a letter glyph.  So instead of saying "substitute only when
preceded by a space" it means "substitute always, except when preceded by
a letter".  At the start of a word, there is no preceding glyph, so the
ignore doesn't match and the main rule does match.  Of course, something
similar could be done at the end of a word.  This approach seems more
reliable than asking the application to pass in a space glyph for the
rule to detect.

The document linked above also mentions a possiblity of limiting an
entire feature (all rules in the feature) to match only at the start or
end of word, using a "feature tag registry" entry.  I'm not sure how one
does that in practice, or whether it may be a hardcoded thing associated
with the four-letter name of the feature.
-- 
Matthew Skala
mskala at ansuz.sooke.bc.ca                 People before principles.
http://ansuz.sooke.bc.ca/