Re: master 47e313e9805: Unbind 'C-M-i' in Text mode

[Eshel Yaron]

Dmitry Gutov <dmitry@gutov.dev> writes:

> On 06/12/2023 09:53, Eshel Yaron wrote:
>> Hi Dmitry,
>> Dmitry Gutov <dmitry@gutov.dev> writes:
>>
>>>> +(defun ispell-completion-at-point ()
>>>> +  "Word completion function for use in `completion-at-point-functions'."
>>>> [...]
>>>> +         (list beg end (cdr all)
>>>> +               :annotation-function (lambda (_) " Dictionary word")
>>>
>>> I have to say that "Dictionary word" is not a great annotation: it's
>>> long, longer than most of the completions coming from that backend, so
>>> these words, repeated, constitute most of the text inside the
>>> Completions buffer. And it's the same string for all completions.
>> I see what you mean.  How do you feel about the following
>> alternatives?
>> - "Word"
>> - "Dict"
>> - "Spell"/"Ispell"
>> - "🕮"
>> - No annotation at all
>> - One of the above, and make it configurable
>
> My preference would be "no annotation at all", if we're talking about
> simple solutions.

Fine by me.  I'm attaching a simple patch that does that.

> The fourth option was a newspaper emoji?

Well, it's supposed to be a book, you get it with `C-x 8 RET book RET`.

> I think the more comprehensive approach here would be to implement
> support for `:company-kind` in the default UI, and then return `text'
> in that function.

I agree, that would be nice in general and help here too.  For this case
I was also thinking about using the common part-of-speech of each word
as its annotation (e.g. "n" for nouns, "v" for verbs...) but we don't
get that information from the spelling dictionary, unfortunately.


Here's that patch:

0001-Remove-long-annotation-for-word-completion-candidate.patch
Description: Text Data