Re: Entering emojis

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> How am I supposed to go from GOLFER to that glyph?  From POLICE OFFICER
>> it's no problem getting to "woman police officer: light skin tone",
>> because those use the same name in the UCS file and in the zwj file, but
>> GOLFING isn't the same as GOLFER.
>
> How many such cases are there?  Can't you have a small database of
> such "translations"?

I'd prefer things to work automatically -- then there'll be no need to
maintain this stuff as Unicode adds new things every year.  But...
that's perhaps a forlorn hope.  We'll see; things seem to be working
pretty well now with:

>> So the first codepoint is what matters for determining the variants?
>
> Yes, AFAIK.

I'm now doing (some) mapping based on that instead, and that does indeed
fix the problem with golfing.  But it seems like it might give some
false positives (i.e., it thinks that some things that shouldn't be
derived are derived), so more tweaking might be needed in the algorithm.
I think it's "basically working", but I need to rewrite that algo
anyway, because it's a bit of a mess with all the tweaking back and
forth, and I may just have confused myself...

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no