Alain Schneble <a.s@realize.ch> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> If the ioctl call issued by the implementation always fails on
>> sockets, then yes.
>
> On GNU/Linux (debian) the call to ioctl (fd, TIOCSWINSZ, &size) in
> set_window_size returns -1 if called with a socket fd and thus fails
> with errno=25 "Inappropriate ioctl for device".  I guess this might be
> the case on other systems as well...

If we will all agree, here is a patch for the feature/async-dns branch
to basically turn set-process-window-size into a "no-op" when called on
network processes.  And it simply makes the call to the "blocker"
function obsolete as a welcome side effect.