Improving TerminalOp tests

[Paul Sandoz]

On Sep 27, 2012, at 6:18 AM, Brian Goetz <brian.goetz at oracle.com> wrote:
>>> I also included some more TestData instances specifically for sequential streams and map streams whose size is unknown, thus exercising the case of sink.begin(-1).
>> 
>> Did the new tests find anything yet?
> 

No, previously sink.begin(-1) was being tested albeit in a very specific way, so i suspect the new way is testing the same code paths but doing so in a more generic way.


> More likely to start finding errors once we start enforcing the invariants for XxxStreamBuilder (hint)  :)
> 

:-)

Paul.