Re: emacs-27 60c84ad: ; * etc/TODO: Fix last change.

[Eli Zaretskii <eliz@gnu.org>]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Mon, 02 Mar 2020 16:06:17 +0100
> 
>     Eli> On second thought: why do you need regexp-opt in this case?  None of
>     Eli> the other composition rules we have (search lisp/language/*.el for
>     Eli> composition-function-table) use that, so why is Emoji different?
> 
> The ones in lisp/language/*.el were presumably written by hand, unlike
> the Emoji ones.
> 
> Hereʼs an example of the patterns we want to match for U+1F3C3 (there
> are longer ones):
> 
>  "\N{U+1F3C3}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+200D}\N{U+2642}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FB}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FB}\N{U+200D}\N{U+2642}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FC}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FC}\N{U+200D}\N{U+2642}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FD}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FD}\N{U+200D}\N{U+2642}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FE}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FE}\N{U+200D}\N{U+2642}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FF}\N{U+200D}\N{U+2640}\N{U+FE0F}"
>  "\N{U+1F3C3}\N{U+1F3FF}\N{U+200D}\N{U+2642}\N{U+FE0F}"
> 
> Now we could add 12 rules here, one for each pattern, or 1 rule with
> all the patterns as alternatives, or we could run regexp-opt and add
> one optimized pattern.

If this is easier done by hand, maybe we should just do that.  We
could instead have an automated way of _checking_ the patterns against
emoji-*.txt files and flagging the new ones to add.  After all,
Unicode files don't change too frequently, and we already use similar
practices with other Unicode data files we import, see
admin/notes/unicode.