Macro language should not prevent entries of languages

[hatton]

In language chooser dialog type kmr.
Expected: you should see "Northern Kurdish".
Actual: you see nothing.

In language chooser dialog type Kurdish.
Expected: you should see "Kurdish, Central Kurdish, Southern Kurdish, Northern Kurdish".
Actual: you see only "Kurdish, Central Kurdish, Southern Kurdish".

Cause: there appears to be a problem in the logic of DeriveTagsFromJsonEntries. When considering the entry for Norther Kurdish, it sees tag:"ku-Cyrl-AZ" and decides that since it already has an entry for "ku" (the macro language), it should skip this entry.

[mhosken]

I can only speak for the data set, langtags.json. The name "Kurdish, Northern" was added to the "ku" tagset on 6/May/2021 and it had been missing before that. So it may be that the software release isn't up to date with released langtags.json. Secondly, I would not expect to see kmr as a resulting tag in the language picker since kmr is a sub language of the ku macro language (well, it's the primary language for ku).

In effect, there is no such thing as the ku language. There is no such thing as any macro language. Nobody speaks a macro language. Instead they speak one of the languages in the macro set. But since most people want to use the macro language to represent the language most people speak, the primary sub language of the macro set is made equivalent (the same tag set) as the macro language and the default identifier of that tagset is the macro language and not the sublanguage. Thus langtags.json says "kmr" == "ku", use "ku" (if you type kmr it should bring up ku).

Remember also that langtags.json is an orthography (or writing system) tagset database and not a language list, per se. You can derive a language list from it. But as quickly becomes evident, the deeper you dive into this stuff, the more complicated and non-obvious it becomes.