bug#48669: Inconsistent overlay placement between minibuffer-message and set-minibuffer-message

[Eli Zaretskii]

> From: Juri Linkov <juri@linkov.net>
> Cc: joca.bt@gmail.com,  48669@debbugs.gnu.org
> Date: Sun, 06 Jun 2021 23:54:58 +0300
> 
> >> -    (let ((ol (make-overlay (point-max) (point-max) nil t t))
> >> +    (let* ((ovpos (minibuffer--message-overlay-pos))
> >> +           (ol (make-overlay ovpos ovpos nil t t))
> >
> > Doesn't this rely too much on the internal details of
> > minibuffer--message-overlay-pos?  At least, without any comments, this
> > call looks like a riddle that isn't easy to unlock.
> 
> If minibuffer--message-overlay-pos serves its purpose for
> set-minibuffer-message, it seems suitable for
> minibuffer-message as well.

So you object to adding a comment there which would explain what that
call does?

> >> -              (put-text-property 0 1 'cursor t message))
> >> +              (put-text-property 0 1 'cursor 1 message))
> >
> > Why this change?
> 
> Only to make minibuffer-message the identical copy of
> set-minibuffer-message.

I'd say make it the other way around.  Handling the 'cursor' property
with a numeric value is more expensive and tricky than if the value is
t, and in this case the effect is exactly the same.