bug#71716: [PATCH] Add new completion-preview-insert-{word, sexp} commands

[Eshel Yaron]

Hi Jules,

Jules Tamagnan <jtamagnan@gmail.com> writes:

> Eshel Yaron <me@eshelyaron.com> writes:
>
>> One important point is that I'm a bit hesitant about adding the sexp
>> variant, rather then defining only completion-preview-insert-word, and
>> mentioning in the documentation that other variants are trivial to
>> define (and how).  The reason is that I don't have a good idea of when
>> a completion candidate would span multiple sexps (if you have such an
>> example, please share it), so I'm not sure how much utility this
>> command would bring in practice.
>
> The use case that I have for the sexp variant is when completing eshell
> history. Both because: parts of shell commands such as file names can be
> considered sexp's, but also because eshell itself can interpret "full"
> elisp forms.

Thanks, I tried it out with cape-history from cape.el and I can see how
it may be useful for such use cases.

[...]
> Another idea would be to turn `c-p-partial-insert` into a macro that
> uses the `interactive-form` function to generate a sensible
> insert-partial function. I'm more than happy to take this tweak on as
> well.

That may be a nice addition.  In particular we could have a macro that
defines a partial insertion command, and takes care of setting the
completion-predicate of the defined command such that it's only
available when the completion preview is active.

[...]
> From 74d8efceaf8f64f7cf61e36f8a5e8a4fc86e558d Mon Sep 17 00:00:00 2001
> From: Jules Tamagnan <jtamagnan@gmail.com>
> Date: Mon, 24 Jun 2024 08:53:23 -0700
> Subject: [PATCH] Add new completion-preview-insert-{word,sexp} commands

Thank you, pushed to master as commit b3017e7c252, after some tweaks to
the commit message.  I've also pushed a follow up commit (9cb2a204088)
with some minor refinements, see the commit message for details.  One
notable change is that completion-preview-partial-insert does not force
point to the position of the preview overlay ("end") before calling the
motion function.  This makes completion-preview-insert-word behave more
like forward-word when point is in the middle of a multi-word symbol,
with the completion preview at the end of that symbol.  I've added
another test case that demonstrates this behavior.

Could you please give it a try to make sure that everything still works
as you expect?


Thanks,

Eshel