bug#31855: 27.0.50; Async shell command finished message not visible any more

[Daniel Colascione <dancol@dancol.org>]

On 06/16/2018 01:48 AM, Eli Zaretskii wrote:
>> From: Michael Heerdegen <michael_heerdegen@web.de>
>> Date: Sat, 16 Jun 2018 07:12:50 +0200
>>
>> In emacs25 -Q, I type, for example, M-! t r u e SPC & <return>, and the
>> echo area displays "true: finished".
>>
>> In master emacs -Q, doing the same gives you no visible feedback any
>> more.  A message is logged in *Messages*, but it seems to be dismissed
>> immediately.
> 
> Daniel, this seems to be due to your recent changes in keyboard.c.  If
> I make the change below, which reverts a small part of your changes,
> the problem is gone:
> 
> diff --git a/src/keyboard.c b/src/keyboard.c
> index 5409918..76097e8 100644
> --- a/src/keyboard.c
> +++ b/src/keyboard.c
> @@ -8999,7 +8999,7 @@ read_key_sequence (Lisp_Object *keybuf, Lisp_Object prompt,
>     /* These are no-ops the first time through, but if we restart, they
>        revert the echo area and this_command_keys to their original state.  */
>     this_command_key_count = keys_start;
> -  if (INTERACTIVE && is_replay)
> +  if (INTERACTIVE && t < mock_input)
>       echo_truncate (echo_start);
>     is_replay = true;

Yep. You're right. I'll bring us back to the old logic here.

I just worry that the "t < mock_input" condition is wrong in the case 
where we use a function key map to rewrite an input to nothing, but 
maybe the improved last_nonmenu_event tracking will fix that class of 
problem by itself.