Re: [ELPA] New package: transient

[Eli Zaretskii]

> From: João Távora <address@hidden>
> Date: Mon, 4 May 2020 02:01:13 +0100
> Cc: Eli Zaretskii <address@hidden>, address@hidden,
>  Stefan Monnier <address@hidden>, Richard Stallman <address@hidden>,
>  emacs-devel <address@hidden>
> 
> [lgstring-char] should be renamed composite-lgstring-char

lgstring-char is an internal function, why do we care about its name?
It doesn't even have a doc string.  A better way with those is to
somehow mark them (perhaps by their symbol's property?), and make help
commands and completion avoid them by default.

OTOH, consider the use case of someone who would like to write
ligature support for Emacs (something which I hope will happen soon).
That person will need to find the various *lgstring* functions for the
job, even though they are internal, and naming them
composite-SOMETHING won't help them, since they will think of
ligatures, not of compositions.

I don't know how to solve this contradiction, but if we want our
completion to be a better tool, we should do something about these
conflicting goals.

> I can even see a completion system where you type "alist" and "assq"
> appears in the list by considering some source of truth (the manual).

It should be enough to look at the arguments as they appear in the doc
string.