Re: Testing redisplay code in batch

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: psainty@orcon.net.nz,  emacs-devel@gnu.org
> Date: Thu, 24 Sep 2020 14:31:58 -0400
> 
> Based on your comments, I assume you consider that this feature is
> acceptable for `master` once it's clean.

Sure, why not?  It looks useful.

>         /* No actual display to update so the "update" is a nop and
>            obviously isn't interrupted by pending input.  */
>         paused_p = false;

OK.

> Without it I got an assertion failure in xdisp.c:18241
> 
>   if (MINI_WINDOW_P (w))
>     {
>       if (w == XWINDOW (echo_area_window)
> 	  && !NILP (echo_area_buffer[0]))
> 
> because XWINDOW complains that `echo_area_window` is not a window.
> 
> Maybe the better fix is to change `init_xdisp` so it also sets
> `echo_area_window` when `noninteractive`?

Yes, I think so.

> Do you think it's otherwise acceptable as-is?  I was thinking of only
> allowing redisplay in `FRAME_INITIAL_P` under the control of some
> special config var.

A command-line option, perhaps?  I agree that the "normal" -batch
should probably continue working as it does now.

And yes, I think it's okay to go in.  I bet it may not support too
many display features, but we can always extend it as we go.

> > How about a small NEWS item about this?
> 
> I'll see to it,

Thanks.