GNU bug report logs - #76870
31.0.50; server-start with non-nil arg LEAVE-DEAD can write confusing warnings

Previous Next

Full log

Message #8 received at 76870 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Jens Schmidt <jschmidt4gnu <at> vodafonemail.de>
Cc: 76870 <at> debbugs.gnu.org
Subject: Re: bug#76870: 31.0.50;
 server-start with non-nil arg LEAVE-DEAD can write confusing warnings
Date: Sat, 08 Mar 2025 18:04:04 +0200

> Date: Sat, 08 Mar 2025 16:45:27 +0100
> From:  Jens Schmidt via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org>
> 
> * Simplified Reproducer
> 
> Start an Emacs session A with emacs -Q, start an Emacs server in that
> with M-x server-start RET.
> 
> Now start another Emacs session B with emacs -Q and evaluate the
> following with C-j in its *scratch* buffer:
> 
>   (require 'server)
>   server
> 
>   (server-start t)
>   nil
> 
> You get the same warnings as shown in the original reproducer.

What happens in this recipe is expected, and is not a bug.

> However, if there is an external Emacs server running in an Emacs
> session A, then `(server-start t)' from another session B obviously
> cannot just stop that server.  However, the warning issued in session B
> as quoted above mentions that the server cannot be *started*.

Which is the truth: server-start attempted to stop the (non-existent)
local server process, which silently succeeded, and then attempted to
start a new server process, which fails with the above message.

> 1. So I think in case of non-nil LEAVE-DEAD and a running external
>    server, a more appropriate warning would be:
> 
>      ⛔ Warning (server): Unable to stop an external Emacs server.
>      There is an existing Emacs server, named "server"
>      Stop it either in its Emacs session or call M-x server-force-delete to forcibly disconnect it.

This doesn't seem to be very different from the current message.  I
don't think the "in its Emacs session" part is significantly more
clear than "stop the existing server".

> 2. However, one could also argue that, if there is no local server
>    running, a call to `(server-start t)' should be completely silent.
>    That is at least what `(server-start t)' does if there is no server
>    (neither local nor external) running at all.
> 
> 3. Or at least it should be silent if called from a non-interactive
>    context as shown in the original reproducer.
> 
> 4. Or at the very least it should be silent if called from
>    `server-unload-function'.

The only bug here is 4.  Unloading the server feature should not cause
this warning.  All the rest is the intended behavior.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.