hanging select processes

Hi,
I am using the odbc driver to provide MS Acess with a view on our
postgres db, securing the connection using stunnel.

Everthing works fine, and it seems to be a fantastic solution to give
our clients a user friendly look at the data.
the only problem we have discovered when testing this out is that if the
client runs a query in access and then kills the access process on their
computer. we end up with a postgres SELECT thread running at 99% cpu
time on our backend. I have found a thread possible hinting at this in
the archives - regarding 'zombie' processes, but these are not zombie
processes - we can kill them though a restart of postgres is required to
get it all working again.

Now we cant stop our clients being idiots - i guess we'd all be a lot
happier if we could! Is there a way to mitigate or even stop this errant
behaviour? surely postgres should stop the select if the client
dissapears in a puff of smoke?
cheers
John