On 27/09/06, John O'Hara <[EMAIL PROTECTED]> wrote:
For testing I understand.
But what are we testing at that point?

It is extremely useful to be able to test without the latency
introduced by the network. We have found a number of timing-related
bugs this way. It is also useful to know that the higher layers in the
broker and client are not intimately bound to TCP/IP.

You should be able to run one test harness against both the C++ and Java
servers... that would be real leverage; and that would use IP.

Of course and nobody is suggesting that we should not be able to do this.

I think popping out the notion of a VM transport sends a bad message to
prospective users.

I don't follow. What bad message?

RG

Reply via email to