Lars Ingebrigtsen <larsi@gnus.org> writes:


Nice!

Perhaps we already get what I am about to say below from what you have,
but given that there are 256 variation selectors, it might be useful to
have a bigram/trigram model for chars that can meaningfully compose as
an emoji?

Once we had such a bigram/trigram model, the choices could be
progressively filtered without creating a giant list of choices  ---
this would permit the user to focus in on the   emojis that make sense
as an emoji.
> Lars Ingebrigtsen <larsi@gnus.org> writes:
>
>> https://unicode.org/Public/emoji/14.0/emoji-test.txt
>>
>> So I'll be rewriting that bit tomorrow.  That file also seems to allow
>> me to easy make the derivation groups without parsing all the other
>> files, so it should be faster, too.  So...  better all around, I think.
>
> Yup.  This is now implemented, and it's way less hacky, and I seem to be
> able to get all the derivations I'd expect (and I don't see any false
> positives).

-- 

Thanks,

--Raman(I Search, I Find, I Misplace, I Research)
♈ Id: kg:/m/0285kf1  🦮