Oliver Jowett wrote:
> I haven't seen this before and there's nothing obviously wrong with the
> driver's protocol-level code that I can see.. so unless you can
> reproduce it, it's a bit of a dead end :(
Actually, I just dug through CVS and noticed a post-7.4.1 bugfix.
Without the fix, asynchronous notification messages (via LISTEN/NOTIFY)
would have caused the message stream to get out of sync.
Does your application use LISTEN/NOTIFY at all?
-O