Re: elpa.gnu.org packages requiring external packages

[Dmitry Gutov <dgutov@yandex.ru>]

On 2/3/18 03:43, Eric Abrahamsen wrote:

> company primarily does popup-on-a-timer;

popup-on-a-timer is irrelevant. company supports 
completion-at-point-functions via company-capf.

The only major feature that completion-at-point-functions don't get 
(that company backends do) is asynchronous operation.

 > I think calling it as an
 > explicit completion command (while supported) isn't how it's expected
 > to be used.

M-x company-<backend> is a well-established way of using it (not the 
most popular one, of course).

>> And overall, I really wish someone could sit down, take the ivy,
>> company, helm, and completion-at-point-function APIs and design a new
>> API which can be used by all of those UIs so you don't have to implement
>> N different slight variations of the same thing.

It would be nice to see some input from Ivy's author: it does support 
asynchronous operation, and even operating on an incomplete set of 
results. That's something CAPF does not have yet. Add fuzzy matching, 
and that's basically everything we want such API to have.

> But they do their things in such different ways. Some work on a timer
> with a popup (company), others tie into the existing completion
> mechanisms (helm and ivy), and others provide their own versions of
> basic Emacs commands (helm and counsel). How much unification is
> possible?

company-capf is an example of one such unification. And given a 
powerful-enough API, all packages could have the ability, at least, to 
switch to it.