Re: Plug treesit.el into other emacs constructs

[Yuan Fu <casouri@gmail.com>]

> On Dec 13, 2022, at 3:19 PM, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> 
>>>> I mean, what construct is each one expected to jump over?
>>> In my book "sexp" movement should jump over subtrees of the AST.
>> It’s pretty hard to judge which subtree to move over at a given point in an
>> AST. For example, when point is at | in the following text:
>> 
>> (|X.y(z), alpha)
>> 
>> Should point move over X, or X.y, or X.y(z)? All three subtrees has their
>> beg=(point).
> 
> Exactly.  It's even a bit worse: I'd also argue that an additional valid
> choice is to move over the whole "X.y(z), alpha".
> 
> The semantics I opted for in SMIE is to choose the smallest/deepest
> subtree.  That's what best matches the previous behavior of
> `forward-sexp`.  If the users want to move over larger units they have
> to place their point elsewhere (e.g. if it's just before ".", then
> moving over "y" wouldn't make sense because "y" is attached to "y(z)"
> and not to ".", only "y(z)" is attached to ".").

I would argue that the purpose of forward-sexp is to move over items in a list. Always going for the smallest subtree doesn’t seem to align with it. Take that example above, going across the smallest subtree means moving over X, then moving over “.”, that doesn’t feel like what forward-sexp should do to me. I think I'm misunderstanding what you mean.

> 
>> Just a thought, but maybe we can let major modes define what’s an “abstract
>> list”, and sexp-forward would move across the immediate children of abstract
>> lists. Eg, abstract lists in C would contain block, argument list,
>> statement, etc. And in the example above forward-sexp would move across
>> X.y(z) because it’s an immediate children of the enclosing abstract list,
>> the argument list.
> 
> Using the semantics I advocate, the user needs to place his point just
> to the left of `;` in order for `forward-sexp` to jump over the next
> instruction (or to the right of the `;` in order to jump over the
> previous instruction with `backward-sexp`).

You mean in the following code

int a = 0[1];
int b = 1;[2]

Forward-sexp would move [1] to [2]? But if we move over the smallest subtree, I’d imagine it only move across the semicolon after [1]. Even if it moves from [1] to [2], needing to adjust point feels very inconvenient to me, at least I wouldn’t want to use something like that. I want to type a single binding and move to where I want, and type that binding multiple times to move multiple steps. Both doesn’t seem to be possible with what you described.

Yuan