Echo overriding messages

Echo overriding messages

[Jay Belanger]

Emacs used to not always cover up messages with keystroke
echoing.  For example, in Emacs 24.5, setting

(local-set-key "ab"
	       (lambda ()
		 (interactive)
		 (message "Hello")
                 (let ((echo-keystrokes 0))
                   (push ?a unread-command-events))))

and then keying in "ab" would display
  Hello
in the minibuffer.  In the git master, it will display
  a-
Even setting `echo-keystrokes' to 0, the message is overwritten
and the echo area is blank (although the message does show up in the
*Messages* buffer). 

Parts of the Calc help mechanism rely on the old behavior.
Is the new behavior intended, or is there a simple way of
reproducing the old behavior?

Jay

Re: Echo overriding messages

[Stefan Monnier]

> Emacs used to not always cover up messages with keystroke
> echoing.  For example, in Emacs 24.5, setting

> (local-set-key "ab"
> 	       (lambda ()
> 		 (interactive)
> 		 (message "Hello")
>                  (let ((echo-keystrokes 0))
>                    (push ?a unread-command-events))))

The let-binding has no effect since the echo doesn't take place during
the `push', but after the command finished execution.

> and then keying in "ab" would display
>   Hello
> in the minibuffer.  In the git master, it will display
>   a-

Hmm... This is not an intended change.  Both behaviors look
acceptable/reasonable, but can you make a bug report for it?

> Even setting `echo-keystrokes' to 0, the message is overwritten
> and the echo area is blank (although the message does show up in the
> *Messages* buffer). 

Hmm.. indeed, it shouldn't be blank.


        Stefan

Re: Echo overriding messages

[Jay Belanger]

>> Emacs used to not always cover up messages with keystroke
>> echoing.  For example, in Emacs 24.5, setting
>
>> (local-set-key "ab"
>> 	       (lambda ()
>> 		 (interactive)
>> 		 (message "Hello")
>>                  (let ((echo-keystrokes 0))
>>                    (push ?a unread-command-events))))
>
> The let-binding has no effect since the echo doesn't take place during
> the `push', but after the command finished execution.

Oh; the let part was just me trying something out; I forgot to remove
it.  The behavior I described was from the same thing without that
binding. 

> Hmm... This is not an intended change.  Both behaviors look
> acceptable/reasonable, but can you make a bug report for it?

Done; thanks.

Jay