Re: Better emoji support

[Kévin Le Gouguec]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Kévin Le Gouguec <kevin.legouguec@gmail.com>
>> Date: Sun, 19 Sep 2021 21:43:17 +0200
>> Cc: emacs-devel@gnu.org
>> 
>> > (set-fontset-font t 'emoji "Noto Color Emoji" nil 'prepend)
>> 
>> After evaluating that, it's down to:
>> 
>> > #*0©®🛝🟰🥹🧌🩻🪩🪷🫃🫗🫠🫰
>> 
>> So there's still something that overrides Noto Color Emoji for those
>> four codepoints, IIUC.
>
> If you mean these four:
>
>    ↔♠♣♥
>
> then why do you expect them to be displayed using Noto Color Emoji?

*In the long run*, when Emacs adheres fully to UTS #51, I don't expect
those to be displayed with a color font by default, since their
Emoji_Presentation property is "No".

(IIUC, ideally we'd want Emacs to display them with a color font only if
(1) users enforce it with set-fontset-font or (2) the characters are
followed by U+FE0F VARIATION SELECTOR-16)


*Right now* (as of commit 995a623594), I'd expect them to be displayed
 with Noto Color Emoji, since

(1) (char-table-range char-script-table ?↔) yields "emoji",
(2) fontset.el indiscriminately slaps "Noto Color Emoji" on the whole
    "emoji" script.

IOW my expectation comes from trying to understand what the code
currently does, not necessarily from thinking it's The Right Thing.
Sorry for being unclear.


IIUC Robert's tentative patch[1] gets us closer to The Right Thing, by
making sure (char-table-range char-script-table C) is "emoji" iff C's
Emoji_Presentation property is "Yes".


[1] <87a6k8l6pp.fsf@gmail.com>
    https://lists.gnu.org/archive/html/emacs-devel/2021-09/msg01449.html