bug#81598: [PATCH] Add new input method for toki pona

> From: Joshua Murphy <[email protected]>
> Date: Tue, 11 Aug 2026 19:20:32 +0000
> 
> This adds an input method for Toki Pona that allows typing sitelen pona
> characters using sitelen lasina (Latin script). You need a font that
> supports the UCSUR range that sitelen pona covers to view the glyphs.
> GNU Unifont covers it as well as some other more specialized free
> software fonts [1].
> 
> I haven't added an etc/NEWS entry yet because I didn't know the best
> header to put it under.

Thanks.  Unfortunately, I don't think we can accept this patch.  The
main reason is that the characters of this script are not yet part of
Unicode.  Emacs uses a lot of character properties defined by Unicode,
lifted from the Unicode data file UnicodeData.txt, and that file does
not currently have any data for codepoints in Private Use Areas (PUA)
which this script uses.  Without this data, many Emacs features will
not handle these characters correctly.  For example, the default
syntax for characters is word-constituent, so all of the codepoints in
PUA have this syntax, which AFAIU is wrong for the characters of this
script.  Also, AFAIU there are certain character-composition rules for
these characters (like when a noun is followed by an adjective), but
the shaping engines will not generally work with PUA codepoints (and
your patch doesn't define any composition rules to begin with).
Filling features will likely not work correctly as well.  And that is
just the tip of a very large iceberg.

So I'm afraid we need to wait for this script to be added to Unicode.

If and when the script is officially part of the Unicode Standard,
there's more to supporting it in Emacs than just adding an input
method.  We need to have a suitable language environment, defined by
its info (see language-info-alist); the input method then becomes the
default for that language environment.  We also should add a line to
etc/HELLO showing the script.  If there are composition rules for this
script, they should be added to the appropriate file under
lisp/language, as we have for other supported languages.  Last, but
not least, lisp/international/fontset.el should be updated with
suitable settings for the script (representative characters etc.), so
that Emacs would be able to search and find fonts suitable for
displaying the characters of this script.

If you want to have this support in Emacs before the script is
officially adopted by Unicode, may I suggest to do this as a GNU ELPA
package instead?  You will still need to provide at least some of the
information mentioned above, like your own data to replace the missing
information in UnicodeData.txt, but at least it will only affect the
enthusiasts of this script and no one else.