Re: Useful completion package in GNU ELPA

[Tassilo Horn]

Christopher Dimech <dimech@gmx.com> writes:

Hi Christopher,

> The way you say it sounds very good.  I wonder how it works with
> orderless and other completion styles such as partial-completion and
> initials.  Or does aggressive-completion have such capability?

It is agnostic of the used completion-styles but works best for those
where minibuffer-complete (or rather
aggressive-completion-auto-complete-fn) can complete the current
candidate further.  That's the case for styles whose candidates usually
have a common prefix which can then be completed.

With orderless, because of the orderlessiness, that's less common, e.g.,
with input fo most styles would have as candidates foo-bar, foo-quux,
foo-baz and aggressive-completion could complete to foo-.  With
orderless, baz-foo would also be a canditate so there is no common
prefix which could be completed.  (Maybe it could complete at least
"foo" and then it's the users choice if she wants to complete at the
front or at the end by moving point...).

Similar arguments hold for the flex completion style.

So all in all, I'd say aggressive-completion works very good for the
styles basic, partial-completion, and substring and less good for more
"open and flexible" styles.

Bye,
Tassilo