[OpenType] Allowing DFLT to take languages

Eric Muller eric.muller at efele.net
Tue Apr 12 17:51:35 CEST 2016


I remember debating with myself whether to allow language-specific 
behavior under the DFLT script, back in 2001 when we (Adobe) proposed it.

I have been searching my memory for the rationale of not allowing it, 
and this is what I reconstructed:

- it seemed weird that you (font designer) would care to have language 
specific behaviors when you have no idea of the script. That may be more 
apparent if you think of OT script tag + OT language tag as (an 
approximation of) a locale; I believe that in the locale world, for all 
practical purposes, "fr" is really to be treated as if it were 
"fr-Latn-FR", ie a locale is always completed to have the three 
components (surely, if I ask for "what's the word for Monday in 'fr'" I 
don't expect the answer to come back in the Deseret script)

- if you really need to do something specific for language l, then you 
also know that it is commonly written in scripts x, y, z, and you can add
(x, l, feature) -> lookup, (y, l, feature) -> lookup, (z, l, feature) -> 
lookup. Not necessarily very satisfactory, but there is a practical 
escape valve. And remember that if you have x, y or z in you font (for 
any language or feature), you need those anyway, (DFLT, dflt, feature) 
is not going to be used in that case.

- it is a lot easier to open the door to (DFLT, specific language) than 
to close it, from a system point of view; without a good case for it, it 
seemed prudent to leave the door closed at first.

While right now I am not opposed to opening the door, the third point 
still makes a lot of sense, and I would prefer to hear of a real use 
case before opening the door.

Eric.



More information about the mpeg-otspec mailing list