[MPEG-OTSPEC] Shared GSUB/GPOS notes, was Re: dmap proposal

[Jonathan Kew]

On 28/12/2023 16:02, John Hudson wrote:
>
> I’ve begun leaning the other way: perhaps we should ditch the 
> multivalency of langsys, restrict it to tags that map to languages, 
> and only use it in that way?
>
> JH
>

Would we then expect font developers to use certain "typical" language 
tags to represent typographic conventions, and expect users to tag 
content as being that language in order to get the desired behavior? Or 
do we expect applications to offer an "override" mechanism for tagging 
content as actually being in language XXX, but requiring the OpenType 
rendering behavior of language YYY?

I'm not really seeing how this makes things better than explicitly 
recognizing that typographic conventions are inherently distinct from 
language, and need to be managed in their own right (although in many 
cases there's a reasonable default mapping that can be used).

JK