> > Addison wrote: > > > No, actually we can't "program implementations" without > > breaking existing "well-formed" implementations (which I > > think represents the majority of implementations). > > I am not sure I understand this. What you said was: > You can easily programme into implementations the scenarios for when a > user > has used an extlang as a primary language subtag and you don't need a > type 2 > on the macrolanguage to say that it can be used as a standalone primary > language subtag - that's covered by the text. "Programming into implementations" information about which extlang subtags can be used as primary language subtags requires a table of items from the registry. Implementations that rely on remove-from-right solely (i.e. "well-formed" implementations, as opposed to "validating" implementations) will not have this information and will get it wrong. > > > Reliance on metadata in the registry is, of course, something > > you are familiar and comfortable with---ISO 639-6 relies on > > it. But BCP 47--and the matching schemes in particular--have > > historically eschewed such reliance. If we introduce reliance > > on the registry, it breaks the implementations that rely > > solely on tag structure. Arguments between extlangistas and > > non-extlangistas have greatly revolved around this particular problem. > > Or this... I take it you mean that implementations are just looking at > the > syntax to see if it is wellformed and not hardcoding the registry > within. Yes: well-formedness relies solely on the syntax rules. We don't require a copy of the registry for that. Validation requires a copy of the registry. > Can you give an example? > > > If this were just a one-time thing, such as the grandfathered > > list, it might not be so bad, since we could give the list to > > hardcode in the RFC. But there is no indication that the > > macrolanguage list is closed or particularly stable (even if > > entries are stable, new ones will probably be created). > > But that is same across the Registry - new entities are created all the > time. New entries are created all the time. Validating implementations rely on the registry as of some particular registry date. But well-formedness is not affected by this. You don't have to know what the tags mean to do many operations on them. What I'm suggesting is that: if we are going to *require* that tags of the form "yue-*" be treated as equivalent to tags of the form "zh-yue-*", that might be okay (I'm not convinced that it is, please note) if the list were short, closed, and stable. If, however, it changes over time, it is hard on implementations that otherwise have no use for the registry. Addison Addison Phillips Globalization Architect -- Lab126 Internationalization is not a feature. It is an architecture. _______________________________________________ Ltru mailing list Ltru at ietf.org https://www.ietf.org/mailman/listinfo/ltru
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.