8201474: (so) Socket adaptor connect(InetAddress, timeout) succeeds when connection fails

[Alan Bateman]

On 13/04/2018 16:17, Brian Burkhalter wrote:
>
> I think this looks OK. However in the test, testConnect() has a 
> ‘shouldFail’ parameter which is always passed ‘true’. Should this be 
> remove or cases set to ‘false’ added?
>
I've tried it keep the test consistent with the other tests. The first 
test connects to a DayTimeServer, both with and without a timeout, so 
that code is being exercised. The new test connect to a RefusingServer, 
both with and without a timeout, as that was the case missing from the 
original tests. We could drop the `shouldFail` parameter but that makes 
it less useful if expanded. We could add a "false" case but that would 
mean connecting to a DayTimeServer or EchoServer and it already does 
this. So I'm inclined to leave it as in.

-Alan