Re: emacs pgtk doesn't set WAYLAND-DISPLAY

[Eli Zaretskii <eliz@gnu.org>]

> From: Po Lu <luangruo@yahoo.com>
> Cc: j.cubizolles@free.fr,  emacs-devel@gnu.org
> Date: Thu, 15 Feb 2024 18:46:43 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> From: Po Lu <luangruo@yahoo.com>
> >> Cc: Julien Cubizolles <j.cubizolles@free.fr>,  emacs-devel@gnu.org
> >> Date: Thu, 15 Feb 2024 16:11:05 +0800
> >> 
> >> Eli Zaretskii <eliz@gnu.org> writes:
> >> 
> >> >> From: Julien Cubizolles <j.cubizolles@free.fr>
> >> >> Date: Thu, 08 Feb 2024 17:18:36 +0100
> >> >> 
> >> >> In Wayland, the following environment variables are set in a shell
> >> >> 
> >> >> WAYLAND_DISPLAY=wayland-0
> >> >> DISPLAY=:0
> >> >> 
> >> >> In Emacs built with pgtk, only DISPLAY is set, with
> >> >> DISPLAY=wayland-0.
> >> >> 
> >> >> I don't know if there is a good reason for that but it causes the
> >> >> matplotlib python program to fail as it seems to expect a "valid" value
> >> >> for DISPLAY or WAYLAND_DISPLAY to be set (see
> >> >> https://github.com/matplotlib/matplotlib/issues/18377).
> >> >> 
> >> >> Maybe Emacs could just copy both the WAYLAND_DISPLAY and DISPLAY
> >> >> environment variables instead of merging both ?
> >> >
> >> > Po Lu, any suggestions or comments?
> >> 
> >> I think this is was Emacs's treatment of WAYLAND_DISPLAY until we
> >> received several bug reports that couldn't be resolved except by
> >> refraining from changing that variable.
> >
> > Can you point to some of those bug reports?
> 
> bug#53969 comes to mind, though several more problems were reported, to
> both this list and the bug tracker.

OK, so I guess the solution to this mess could only be in the
application's own code, whereby, if WAYLAND_DISPLAY is necessary, the
Lisp program that invokes an external utility should inject
WAYLAND_DISPLAY into the environment, is that so?