bug#29789: 25.1; Emacs blocks user input when using visual-fill-column in wide terminals

[Luis Gerhorst]

I'm usually using iTerm 3.1.5 and just tried again with the native
Terminal.app from macOS 10.13.2 (same issue). It also occured on a
Debian system I was running Emacs on over SSH.

I was not able to test with a non-macOS terminal emulator. I can do that
tomorrow however.

With 'hang' I meant that Emacs does not respond to user input at
all. The whole screen is frozen.

Eli Zaretskii writes:

>> From: Luis Gerhorst <privat@luisgerhorst.de>
>> Date: Wed, 20 Dec 2017 20:28:27 +0100
>>
>>
>> 1. Run Emacs in a wide terminal (about 150 characters, 100 isn't
>> enough): emacs -nw -Q
>>
>> 2. Load a arbitrary theme: M-x load-theme adwaita
>>
>> 3. Load visual-fill-column. E.g. by copying
>> https://raw.githubusercontent.com/joostkremers/visual-fill-column/c0e5ec8f88a7598f7afdee565978410bee0de2a0/visual-fill-column.el
>> into the scratch buffer and typing M-x eval-buffer.
>>
>> There is nothing suspicious about the code, regular lisp code should not
>> be able to block Emacs (I believe). It does not seem to contain any loops 
>> that block the main thread. The maintainer knows about the issue but say's 
>> it may be an issue with low-level code. In GUI Emacs it works.
>>
>> 4. Open a buffer and enable visual-fill-column-mode: M-x
>> visual-fill-column-mode
>>
>> 5. Your Emacs is now blocked.
>
> I tried to reproduce this on GNU/Linux and on MS-Windows, but failed
> both times, both with Emacs 25.1 and the current emacs-26 release
> branch.
>
> What I see is that after enabling the mode, Emacs becomes somewhat
> sluggish in cursor motion (hardly surprising, given what the mode
> does), but it certainly doesn't hang.
>
> I wonder what's different on your system, or on that of those who
> confirmed the problem.