Matt Mayer

I agree the country codes aren't ideal. It partly depends if you expect to add data for eg phone module which has country codes , location module which has cities...

Wikipedia uses ku for Kurmanji (with a toggle between ku-latn and ku-arab) and ckb for Sorani

Mozilla uses https://pontoon.mozilla.org/sdh/ https://pontoon.mozilla.org/kmr/ https://pontoon.mozilla.org/ckb/

I think the cleanest would probably be: * `ku_kmr_latin` → Kurdish (Kurmanji, Latin) - variant `kmr_latin` in metadata, fakerKU_kmr_latin when precompiled * `ku_kmr_arab` → Kurdish (Kurmanji, Arabic) - variant `kmr_arab`...

For ease for review please don't add any more modules for now. We prefer to get a small PR reviewed first, once it is approved you can follow up with...

> > i though we should use only contry code like IQ after the ku_ , but if it is fine to use kmr and ckb like you said that...

I think we don't have a really good definition for what goes in "variant" at the moment, but i think its basically "the minimum needed to disambiguate versions of a...

In theory could a generic Kurdish locale just contain things like phone numbers and domain names which would be the same in both scripts?

> I checked the workflows and noticed that the test failure comes from the snapshot file in the "ko" section. I didn’t modify that part when creating the pull request...

Yes noted. That is unrelated to your changes.