bug#48356: [External] : bug#48356: 28.0.50; choose-completion discards the suffix after the completion boundary

[Drew Adams]

FWIW, my opinion (no doubt a minority of one)
is that all such approaches, including what's
in vanilla Emacs now (since `minibuffer.el',
which I guess means Emacs 22/23), are inferior
to the original vanilla behavior.

Icicles uses that original behavior, in which
it _makes no difference where point is_ in
the minibuffer content.

That is, the entire minibuffer content is the
pattern to be matched (whether for completion
or reading by `read-from-minibuffer' etc.).

I find this more flexible & saner - doesn't
matter where point is.  Whether or not you've
made an edit in the middle of the content,
e.g. yanking or typing or deleting there, all
of the current text is used.

If someone really ever wants the text that
follows point not to be taken into account
then it's simple enough to hit a key to remove
it (and it can be restored with undo etc.).

You may say it's also simple enough otherwise
to move point to the end of input (e.g. `M-v').
Fair enough, but I think that's more bother.

It's more common, I think, to edit text in the
middle somewhere, and then either keep the
text that follows point or kill/delete it.

There's never any need to move point just to
get the pattern you want.  You never need to
pay any attention to point in the minibuffer.
___

[I suppose that in some sense this is kind of
a bottle half-full/half-empty choice.  Maybe
akin to views on `delete-selection-mode':
convenience of not having to first use `C-w'
(to replace) versus convenience of not having
to first use `C-g' (to not replace).]