bug#29518: 27.0.50; Compilation errors grab frame focus

[Eric Abrahamsen]

On 01/06/18 15:31 PM, Noam Postavsky wrote:
> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Thanks for the test! That triggers the same focus-grabbing behavior for
>> me -- from other Emacs frames, and other applications' frames, using
>> "emacs -Q".
>>
>> You tried this also with switching to a different workspace? And it
>> didn't turn the workspace's tab red?
>
> I tried now switching to a different workspace after starting the
> compile, the tab did not turn red.
>
>> Perhaps something's configured differently on my system (though I don't
>> know why it would be, I didn't customize anything). Is there anything I
>> might be looking for in config.log?
>
> I can't think of anything.  If you M-x debug-on-error RET x-focus-frame
> RET do you get any hits?

Sorry about the slow response...

That was an odd recipe, but turning on debug-on-error, then running
several varieties of (x-focus-frame (window-frame)), including focusing
the current frame or other frames, never did anything other than what I
would expect it to -- returning nil when the frame was already focused,
and focusing the frame when it wasn't.

>>> There have been some changes to the way timeouts are handles, which
>>> could affect frame/focusing stuff (see #24091, #25521, and #29095), but
>>> I don't think it should cause what you are seeing.
>>
>> I fooled with `x-wait-for-event-timeout' but that didn't do anything. I
>> suppose 6a788d2fc18c23dcfc5d0352649b2f690e9cbff7 could be related
>> (FRAME_VISIBLE_P certainly sounds relevant) but I don't actually
>> understand what's happening in there. Should I try just reverting the
>> commit and re-building?
>
> I don't expect it to affect this, but since we have no other leads, I
> suppose you may as well try it.  You would probably need to revert [1:
> e1f6e3127a] first.
>
> [1: e1f6e3127a]: 2017-09-29 18:40:06 -0400
>   Bring back the busy wait after x_make_frame_visible (Bug#25521)
>   
> https://git.savannah.gnu.org/cgit/emacs.git/commit/?id=e1f6e3127a292e6ba66d27c49ddda4fe949569f5

I wasn't able to revert those commits cleanly, and I guess I don't
really think that's the issue. I suspect this is a change in X behavior
(Arch has xorg-server 1.19), and maybe it's not worth chasing after at
this point. Maybe it's enough just to leave this report hanging here,
and see if anyone else runs into it?

Eric