On 12/8/21 13:56, Ulrich Mueller wrote:
> So, can we please think about a better solution, and not knee-jerk
> something half-baked into Emacs 28, like checking for yet another
> environment variable?

Agreed, now is not the time for significant innovation. It could be that 
any significant improvement will have to wait for Emacs 29.


> Even reverting to the Emacs 27 behaviour would be better than what has
> been suggested above: In Emacs 27, you can set EMACS_SOCKET_NAME to make
> things work.

Setting EMACS_SOCKET_NAME to an absolute file name should work just as 
well in emacs-28 as it did in Emacs 27, because in that case the code 
doesn't consult either XDG_RUNTIME_DIR or TMPDIR. So if that's an 
adequate solution, we should be able to continue with that.


> Given the goal to release Emacs 28.1 soon, I'm not sure there's time to come up with (and be confident in) a better solution for 28. In that case, I guess the available options are:
> 
> * If the security issue is considered minor enough, keep the current Emacs 28 behavior and silence the warning.

I'd rather not silence the warning as there is a real issue here 
(admittedly obscure...).


> * Otherwise, revert to the Emacs 27 behavior and come up with a better solution for Emacs 29. 

That sounds better, if the EMACS_SOCKET_NAME approach is good enough. 
Proposed patch attached. This patch attempts to be reasonably minimal vs 
what's in emacs-28 now (as opposed to cleaning up this somewhat-messy area).