[XeTeX] Diacritics on ligature components

[Jonathan Kew]

On 18 Mar 2009, at 10:41, Peter Baker wrote:

> I'm working on a Latin-script font based on a medieval hand: it has  
> lots
> of ligatures and contextual substitutions, and it uses (or will use)
> lots of combining diacritics. I'm developing with FontForge and  
> testing
> with XeTeX.
>
> Because combining marks (especially marks of abbreviation) can come on
> so many characters in a medieval Latin text, I'm trying to implement
> ligature marks for anchoring diacritics. ("Ligature marks" is a very
> unpleasant term to Google, by the way.) As far as I can tell,  
> FontForge
> does this correctly: the output looks to be according to the
> specification, anyway.
>
> What I haven't been able to figure out is how to input text in XeTeX  
> (or
> any other program, for that matter) to test this. Documentation seems
> hard to come by (or perhaps is in languages where diacritics on
> ligatures are more common--ones I don't read). For example, I have a
> ligature "e_n" which a GSUB lookup substitutes for "en". Each  
> component
> (the e and the n) has an anchor, and there are matching anchors on
> <acute> and <macron>. If I type "en<acute><macron>" then the two marks
> end up over the "e", on top of each other. If I type  
> "e<acute>n<macron>"
> then the ligature never gets made.
>
> I think I must be missing something obvious, but what? Thanks for  
> any help.

It sounds like you didn't tell your GSUB ligature lookup to ignore  
marks. (Sorry, I don't know the exact FontForge way to do that, I've  
used VOLT for the OpenType work I've done in the past.) Or did you  
forget to classify the "e_n" as a ligature rather than a simple base  
character in the GDEF table?

Have you checked that FontForge is generating a "MarkToLigature  
Attachment" lookup (GPOS LookupType = 5) for the positioning?

(I think this is an OpenType font development issue rather than a  
XeTeX one; you might find more knowledgeable people on an OpenType  
mailing list or forum.)

JK