Re: Plug treesit.el into other emacs constructs

[Theodor Thornhill]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Yuan Fu <casouri@gmail.com>
>> Date: Wed, 14 Dec 2022 16:05:10 -0800
>> Cc: Theodor Thornhill <theo@thornhill.no>,
>>  emacs-devel <emacs-devel@gnu.org>,
>>  eliz@gnu.org
>> 
>> > I see, so maybe sexp means the general, flexible AST entity. And thinking 
>> > of it, my idea is just forward-list :-) And we definitely should include 
>> > forward-list into the list of navigation commands we want to support, 
>> > among the ones that are already brought up.
>> 
>> Ok, upon closer inspection, forward-list isn’t exactly what I’m thinking 
>> about, it moves over lists, but I’m thinking about moving over elements of a 
>> list.
>> 
>> If you think about it, there are two kinds of constructs in an AST/grammar: 
>> those are repeatable, and those are not. Statements are repeatable, you can 
>> stack multiple ones together and it still makes sense. Identifiers are not 
>> repeatable, put two together doesn’t make sense (in most languages).
>> 
>> And these repeatable constructs appears at every level of the AST, from 
>> top-level stuff like function definition, to statements, to small stuff like 
>> arguments in an argument list. That makes them very good unit of navigation.
>> 
>> Navigation commands, IMO, needs to be extremely easy to predict and requires 
>> no thinking. That’s the reason why avy never grow on me: yes you can move to 
>> anywhere you want, but it takes so much cognitive load to use. Much better 
>> to type a few C-n, C-M-f, etc to get their, because I don’t need to think 
>> about it.
>
> FWIW, for me, "sexp" in its C/C++ interpretation always means
> "expression".  So in
>
>    foobar (a, b + c, 2 * d - f / 10 + pow (g, x), y);
>
> C-M-f should move to the next comma on the top level (i.e. the comma
> inside the 'pow' call doesn't count), and C-M-f inside the 'pow' call
> should move by commas on that level.
>
> More generally, C-M-f should move to the next expression
> _on_the_same_level_, without entering inner levels.  Exactly like we
> do in Lisp.

Agreed