Raymond Toy writes:
I confess that I don't know the sentinel stuff at all,
If I say it's just event-driven (callback-based) code, does that help?
The point is that the sentinel receives all events, and it receives
them concurrently -- you can't know what state the server is in when
the sentinel is activated. I don't know if the process ever emits
events other than "stop" (or whatever it's called), but if it does it
presumably could do so "simultaneously" with the QUIT message.
I'm happy to supply a new patch not to kill the process and
the process-status is not exit.
Yes, please. After some more experience maybe we will decide that the
killing the process and buffer is always the right thing to do
(actually, I think in the long run that's quite likely), but precisely
because this is concurrent code, it's likely that if there is a bug,
it will be intermittent. It would be nice if we could get at the
*first* process and buffer where it happens. :-)
XEmacs-Beta mailing list