ID for language-invariant strings

[Peter Constable]

> From: ietf-languages-bounces at alvestrand.no [mailto:ietf-languages-
> bounces at alvestrand.no] On Behalf Of Michael Everson


> I can't even begin to understand what "language-invariant" is meant
> to mean. You say you are talking about some group of language names.
> Favour us please with the actual language and or names, please, or
> this idea should just fade away into the vagueness you have given us
> about it.

OK; here's the specific scenario: name strings in OpenType fonts.

An OpenType font can have strings for family name (or other various data categories) in multiple languages, but for good internationalization design, what should be passed in programming APIs or put in markup for purposes of selecting a font resource should be completely independent of the current user's UI language. Thus, we want a way to designate a family name string that would get used for those purposes. The string used for that purpose may happen to be an English name, a Japanese name, a Farsi name, or whatever (it's probably the same string as used for one of the languages covered in the font's resources). What language it is is irrelevant; the main thing that matters is that it is stable since it is what would get used to reference that font resource programmatically in the future.

The language identifier (for which I'm spec'ing the use of BCP 47) constitutes part of the key used to index all of the string resources in an OpenType font, so I need to use an ID from the same identifier space. I could make up a private-use ID, but given that this is probably not an isolated case for needing this, it seems to me to make sense to ask first if there shouldn't be a conventional recommendation in the LTR.



Peter