I am curious to what extent ECN has been tested on both IPv6 and IPv6 tunnels, given all the bufferbloat questions flying around.
http://gettys.wordpress.com/what-is-bufferbloat-anyway/
The recent MIT study didn't address it.
http://gettys.wordpress.com/2011/02/22/caida-workshop-aims-2011-bauer-and-beverly-ecn-results/
I think this has more to do with specific IPv6 implementations than IPv6 itself.
and how many well tested IPv6 ECN implementations are there?
Few or none would be my guess. IPv6 is pretty new territory from a practical implementation standpoint.