> -----Original Message-----
> From: Nicholas Bodley [mailto:nbodley@...]

> It finally came to the front of my consciousness that all the work
> for Arabic shaping and joining should be useful, with appropriate
> for rendering latin script.

In a cursive Latin OpenType font, you don't use features such as 'fina'
that are used for Arabic. Rather, you add lookups to select contextual
forms and associate them with the 'calt' feature. These lookups will
need to include logic to determine what glyphs to use according to word
position or adjacent glyphs. (Word position isn't hard; you can decide
how complex you want to get in picking alternates according to preceding
or following characters.)

> Unfortunately, fonts for such use are likely to require extensions of
> existing standards, perhaps specifying permissible locations on each
> for joining, as well as initial, medial, final and standalone forms,
> perhaps other data.

I'm not at all sure what you're referring to. There are several cursive
Latin OpenType fonts, but none required extensions of any standards.

> Another scheme would have lots of kerning pairs with joins appropriate
> the specfic letter forms of each pair. Each pair might contain
> about best connections to preceding and following letters.

Um, no. Kerning adjusts positions of glyphs, not their shapes. Getting
the right connections for Latin has mostly to do with getting the right

Peter Constable