I have made some further headway with this and the latest patch is on people.a.o [1].
Work is still far from complete but the current position is that I have a simple sample application that runs in Chrome that sends a message to Tomcat and that message is correctly received. The most obvious remaining TODOs are: - Handle multiple messages - Enable writing of data back to the client - Allow apps to close the connection I want to get at least these working before I even think about a comit to trunk. Other major TODOs include: - Handle multi-frame messages - Handle control frames The focus has been on functionality rather than speed. I am sure there will be places that will need some significant refactoring for response time and GC performance but my plan is to worry about those once I have something that works and we can test it to see where the problems really are. I don't want to start premature optimisation - especially when the design may need to be thrown out if something just can't work with the current approach. The good news is that the integration points have remained largely unchanged with this latest iteration so the outlook for the stability of the integration points is good - so far ;). Mark [1] http://people.apache.org/~markt/patches/draft/2012-01-24-websocket.patch --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
