Websockets: wait or implement now?

[Jeff G.]

I'm building a cluster system using netty for part of the networking, as well
as a client library for the cluster system (it's a storage cluster), and I
want to build in a mechanism to push notifications to clients if the cluster
topology changes.  I initially tried using ActiveMQ to do a pub/sub
mechanism, but I've run up against a wall where clients in different
networks can't subscribe to the topic (ActiveMQ is redirecting the clients
to the private IP of the server it's on instead of allowing them to connect
via the public IP, so connections are failing), so I'm thinking I'm not
going to waste any more time on the message queue (I've already wasted about
3 weeks trying to make it work and have not had much success).  

Since websockets is still a draft, should I wait for that to finalize and
netty's support to be more complete, or it is safe to implement websockets
in the client and the server right now (this is a production system that is
also going to be licensed to third parties)?  Keep in mind that clients of
this system will not be running in browsers, so I have to be able to have
the cluster's client library be able to speak websocket if I do it this way. 
My only other viable option is long polling, but I'd rather not have the
client lib create any more threads (the servers it runs inside already
create pretty much every thread in the known universe when they start up, so
my code gets close to starvation on initialization).  Alternatively, if
there's an idea I haven't though of yet, I'm open to suggestions.

--
View this message in context: http://netty-forums-and-mailing-lists.685743.n2.nabble.com/Websockets-wait-or-implement-now-tp6538889p6538889.html
Sent from the Netty User Group mailing list archive at Nabble.com.