> To tell the truth I'm not sure if the resident stack for VM can handle > that area, but the one for Linux certainly can. A Service Machine > running Linux and simply forwarding things between the two would work.
It's not that easy, especially with SSL and encryption requirements -- you have to deal with hard philosophical questions like what the actual endpoint of a socket really is and which certificate is valid in that case. There are also lots and lots of nasty problems wrt to maintaining state with v4/v6 connections -- for example, if the connection initiates from the v4 side, how do you specify the network prefix to get the other half of the v6 address to the remote side? You end up with a bunch of really icky NAT6 translation tables. It's messy as hell. The VM TCP stack does have basic IPv6 support, but none of the shipped apps except TELNET and PING have IPv6 support, and there are no plans (that I'm aware of) to fix that. And yes, requirements have been submitted. I have replacements for most of them ready, but ....
