RE: [External] : Re: [PATCH] `completing-read`: Add `group-function` support to completion metadata (REVISED PATCH)

[Drew Adams]

> > If not, just what do you think is needed, and why?
> 
> Check out elisp--company-kind, for example. And other 
> :company-* stuff in elisp-completion-at-point.

No thanks.  Either you'll say what you think is needed
or you won't.  That's up to you.

I've said what I wanted to about this.  You apparently
agree that, in general, the completion apparatus should
not be concerned with presentation.  It shouldn't, in
general, need to encode info to support presentation.  
To implement presentation features, we should not need,
in general, to bake some settings into the completion
apparatus ahead of time (e.g. when defining the "table").

You've agreed, I think, that there can be exceptions to
this general rule.

It's the general rule I wanted to get across.  Let's
not design the completion table etc. so it becomes our
means to encode presentation or info for presentation.

We've already gone down that road for completion-styles
and sorting.  IMO that was a mistake.  It's better to
be able to control such things more flexibly, from
anywhere, on the fly, instead of pretty much setting
them once and for all for a given call to, say,
`completing-read'.