On Thursday 2008.08.14, at 09:19 , Marc Larue wrote:
[...]
I'm sure that in theory everything is really complex, but in
practice when
the client disconnects the TCP socket a "Socket closed" IOException
will
be trown on the server.
Sigh. With that kind of attitude, I'll say my final bit here and then
stop as that's just too depressing.
It's not "theory", it's reality. As L. Peter Deutsch and so many
others have said in so many different ways, too many developers
believe such completely insane things when it comes to networking and
network-based systems that it's really sad and frightening. He didn't
use the term "fallacies" by accident.
It's really trivial to cause the majority of (professionally
developed) networking applications to e.g., hang indefinitely. I
recently did it in a number of ways to the JavaMail library (as well
as a number of other libraries written in 5 or 6 different
languages). To be clear, that means I was able to get the system into
a state where it would *never* recognize the death of the network.
If you want to choose to disregard reality, that's certainly your
business but please stop infecting other people with that nonsense.
Thanks,
John
