Hi Aki, Sorry but I'm in vacations this week. Can you ping me next week (IRC : ch007m) to discuss that.
Regards, Charles On Tue, Oct 29, 2013 at 1:11 PM, Aki Yoshida <elak...@gmail.com> wrote: > Hi all, > I would like to get back to this websocket topic and I was wondering > if you could give feedback/direction toward how we can proceed. > > Concretely, I think these components do not fit naturally into the > existing related components camel-servlet and camel-ahc because these > components have only either the consumer-side mode or the > producer-side mode, respectively. For the websocket scenario, it's > good to have both in the server side as well as the client side so > that you can pass data to the socket in the duplex mode. > > and so on with further questions... > > feedback appreciated. > > thanks. > regards, aki > > > > 2013/9/30 Aki Yoshida <elak...@gmail.com>: > > Hi, > > A few weeks ago, I posted to this list about my thought on the > camel-websocket > > > http://camel.465427.n5.nabble.com/thoughts-on-the-camel-websocket-component-td5739005.html#a5739013 > > > > I did some prototyping and I wanted to share this with you to get some > > feedback. But I will be on vacation and not brining my notebook with > > me, so I'll be making a break for a while and get back working on it > > after I come back. > > > > And here it goes. I created two endpoints. One is for the server side > > websocket handling based on Atmosphere. It supports both the consumer > > and producer modes to have an asynchronous duplex data transfer on the > > socket, as done in the current camel-websocket component. > > > > The implementation is based on the camel-servlet component and uses > > its own servlet that delegates calls to the Atmosphere's websocket > > handler that is connected to the camel component. > > > > This code [1] is available at my github. Please note that I just used > > an apache camel package name and version during this prototyping and I > > hope this was okay for this kind of work. > > > > I prepared a test scenario [2] that can be executed on karat. This > > requires a service bundle [3] that publishes camel-websocket2's > > CamelWebSocketServlet to the OSGi registry. There is a short > > instruction [4] for setting up karaf for running this scenario. > > > > This scenario is written in Blueprint but I made also the > > corresponding Spring version (see those without "_bp"). In either > > case, the scenario has two routes: one performing a single websocket > > echoing and the other performing a broadcast websocket echoing. > > > > The second endpoint component is for the client side websocket based > > on async-http-client. This also supports both consumer and producer to > > have a duplex data transfer on the socket as in the server side > > processing. I put this endpoint in another component [5], but we could > > consider consolidating these two into one or maybe with others. > > > > As I mentioned at the beginning, I am going on vacation, so I just put > > together all this before making a break. There is some stuff that I > > will clean up or refactor or consolidate after I come back from my > > vacation. > > > > I would appreciate if you can look at it and give some feedback. > > > > Thanks. > > > > [1] > https://github.com/elakito/sandbox-camel/tree/master/components/camel-websocket2 > > [2] > https://github.com/elakito/testzone/tree/master/samples/osgi_camel_websocket_sample_route_bp > > [3] > https://github.com/elakito/testzone/tree/master/samples/osgi_camel_websocket_service_bp > > [4] > https://github.com/elakito/testzone/blob/master/samples/osgi_camel_websocket_sample_route_instruction.txt > > [5] > https://github.com/elakito/sandbox-camel/tree/master/components/camel-ws > > > > regards, aki > -- Charles Moulliard Apache Committer / Architect @RedHat Twitter : @cmoulliard | Blog : http://cmoulliard.github.io
