bug#36745: 27.0.50; completing-read with require-match nil does not accept spaces

[Jean Louis]

* Drew Adams <drew.adams@oracle.com> [2019-07-21 02:07]:
> > So far I understand, icicle is some external
> > library?
> 
> Yes, but it's not important to my point about the subject
> of this bug.  I mentioned it to say that (1) the behavior
> you said you're looking for (a) is easily made available
> and (b) has existed for a long time in external libraries,
> and (2) has already been suggested for vanilla Emacs,
> several times in the past, to no avail.

Sure, thank you. However, I am looking into built-in way of
completing it, so that it has less dependencies, and that
later those completing libraries can take over. That is how
I understood how various completing libraries are working,
like helm. So the completion I would leave to future choice.

> But you're apparently also OK with having the doc just
> remind users that they can input a space char using `C-q
> SPC'.

Now I found more references in Emacs Lisp manual, so there
is the minibuffer-local-completion-map and I think function
shall be described better with references to it.

And in my previous email on this bug, you can see my
proposal, which sounds logical, and which is when user is
entering first non-matching word and press SPC which should
complete the word by minibuffer-local-completion-map, then
there is message [No match] for short time, so in case of
require-match being nil, then space could be added without
hurting any other behavior of the function.

Because SPC can be added after first matching word. And
additional non-matching word can be added to first matching
word without warnings that it is no match.

So I think the function could be perfected that space can be
added also after first non-matching word in which case it is
even more logical to allow it when require-match is nil.