[erlang-questions] Sockets and linking processes

[Liam Clarke]

Hi all,

I have two processes (A and B to be totally original) - process B
handles a socket, based on messages from A. A parses data received
from B and responds accordingly.

I've linked A and B, but what I'm wondering is if A fails, which will
kill B also, if B's socket is still open, will it close it gracefully
or will it leave the remote end hanging?

If so, is there a way I can handle the exit signal so as to close the
socket without making B a system process? I still want B to exit with
A, just that I want B to tidy up before it does.

Regards,

Liam Clarke

PS - I'm up to Chapter 3 of Mr Armstrong's thesis - whoever said it
was the best introduction to Erlang is right. It's good to see the
_why_ of certain implementation decisions, as it helps in
comprehending the _how_.

And, as far as dissertations that I've read go, it's the most readable
by far, so kudos to Mr Armstrong for avoiding the sirens of academic
short hand.