keymanapp/lexical-models

Add `bal` to `bcc` lexical-models

Closed this issue · 6 comments

darcywong00 commented

Yes, we should update those models to also include bal (don't remove bcc at this point to maintain compatibility with 13.0 keyboards)

Originally posted by @mcdurdin in keymanapp/keyboards#1452 (comment)

DavidLRowe commented

IIUC bal is a macrolanguage tag for Balochi and bcc is the tag for Southern Balochi. I don't know enough about this language situation to know whether that is going to cause problems, for example, someone from another Balochi language selecting bal and getting Southern Balochi predictive text. If @rmlockwood is available, it would be good to get his opinion.

darcywong00 commented

keyboards PR keymanapp/keyboards#1452 was abandoned, so this issue may not need to be resolved.

rmlockwood commented
DavidLRowe commented

The issue is that 'bal' macrolanguage uses 'bcc' as its default, so 'bal' is preferred over 'bcc' as the canonical tag.

It seems that keyboards and lexical models that currently have 'bcc' should at least include 'bal', and perhaps should include both 'bcc' and 'bal'. @mcdurdin Can you confirm?

mcdurdin commented

At present, lexical models should list both bcc and bal as the matching for BCP 47 here does not do this level of canonicalization. Keyboards should only include one.

DavidLRowe commented

The lexical model for bcc-arab includes entries for bal, so that should be fine.

The sil.bcc-latn.upp_ptwl1 lists "bcc-Latn", but not "bal-Latn", in *.kps and *.model_info files.
Ideally, this should be changed to include "bal-Latn" (but keep "bcc-Latn" for backward compatibility).

Two keyboards for Latin-script Balochi need to switch from bcc-Latn to bal-Latn (see keymanapp/keyboards#1593), but after the change to the lexical model.

Note also: #119