widget-complete and C-g to hide *Completions*

widget-complete and C-g to hide *Completions*

[Mauro Aranda]

When in some mode like emacs-lisp-mode, after the *Completions* buffer
pops up, typing C-g will hide it (at least in emacs -Q).

But when using widget-complete, which uses completion-in-region instead
of completion-at-point, that doesn't happen.

Test case would be like this:

(defcustom foo "black" "..."
  :group 'emacs
  :type 'color)

Eval and then M-x customize-option RET foo
Move to the color widget and type C-k to erase the value.
Then write something like "bl" and type C-M-i, to get the *Completions*
buffer, with values "black", "blue", at least, hopefully.
Type C-g: Quitting doesn't hide *Completions*, like in other modes
(e.g., emacs-lisp-mode).

I'd like to know why the use of completion-in-region in widget-complete
doesn't turn on completion-in-region-mode, and what would be the steps
to make that happen, but I got lost quickly in the completion code, so
perhaps someone else can take a look.  Is it required that
completion-in-region-mode-predicate is non-nil? That is not documented,
but the code seems to expect that.

Re: widget-complete and C-g to hide *Completions*

[Stefan Monnier]

> I'd like to know why the use of completion-in-region in widget-complete
> doesn't turn on completion-in-region-mode, and what would be the steps
> to make that happen, but I got lost quickly in the completion code, so
> perhaps someone else can take a look.  Is it required that
> completion-in-region-mode-predicate is non-nil?

Yes, `completion-in-region-mode` requires
`completion-in-region-mode-predicate` to do its job, so the caller needs
to set that somehow.


        Stefan

Re: widget-complete and C-g to hide *Completions*

[Mauro Aranda]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> I'd like to know why the use of completion-in-region in widget-complete
>> doesn't turn on completion-in-region-mode, and what would be the steps
>> to make that happen, but I got lost quickly in the completion code, so
>> perhaps someone else can take a look.  Is it required that
>> completion-in-region-mode-predicate is non-nil?
>
> Yes, `completion-in-region-mode` requires
> `completion-in-region-mode-predicate` to do its job, so the caller needs
> to set that somehow.
>
>
>         Stefan

OK, thanks!