Re: Who uses Icomplete-mode? Re: [Emacs-diffs] master b0e318d 2/2: Score flex-style completions according to match tightness

[Stefan Monnier]

> Which brings me to a question I've been meaning to ask: who
> here uses Icomplete-mode and just how attached are you to its
> interface?

I don't use it all the time, but I often like to use it.

> - A less cryptic left side hint as to what the "current" matched
>   thing is (in ido.el it's pretty obvious, in icomplete.el not so much)

Not sure what you mean by that

> - A way to have RET exit the minibuffer with the current thing,

    (define-key icomplete-minibuffer-map [?\C-m]
      'icomplete-force-complete-and-exit)

should do the trick.

>   except when finding files, where it should just enter the directory

This is the one "feature" of Ido which drove me completely mad.  I might
have gotten used to some of the other parts of the Ido UI but this one
just makes no sense to me: I much prefer typing / than RET to enter the
directory (and with partial-completion style this works nicely).

>   (though there should always be a binding for exiting the minibuffer
>   with whatever the current input is).

    (define-key icomplete-minibuffer-map [?\C-j] 'exit-minibuffer)

?

> - A way to delete buffers and kill files effectively without leaving
>   the minibuffer prompt.

Hmm... I thought you coded this up already, but I don't see it in
icomplete.el.  Did I dream it?

This is sliding towards Helm territory and is difficult to provide in
a clean generic way (icomplete-mode is supposed to work for all
completions), but it might make sense to offer "category-specific" key
bindings and commands, yes.

> - Eliminate the need for icomplete-compute-delay, using while-no-input
>   or maybe threads.

Looks like you have your work cut out !

> Some of these can be achieved through customization, and some
> would probably need more work.  Moreover, if there aren't a lot of
> people overly attached to the current interface, we could choose
> some new defaults.

I like the fact that the key-bindings are pretty much the same with and
without icomplete, so I'm not looking forward to changing the default
RET binding in icomplete-mode.  But that's my opinion as icomplete user,
so if I'm in the minority I can live with it.

> Otherwise, I'd propose a new icomplete-ido-emulation-mode with
> some less horrible name.

That might actually be a good approach (especially if it's built as just
"icomplete-mode with different defaults" so the new features can also be
used from icomplete-mode modulo simple customizations), also because it
will be new and so could attract new users more effectively than
"icomplete-mode has slightly different default behavior".


        Stefan