Mathieu's write-up of Atmosphere vs Cometd seems to cover it.

When I first looked into providing non-chrome support for Wave I considered
Cometd. At the time I decided that it would be easier to write a Java
backend for Socket.IO than it would to modify the Wave server to work with
Cometd. Now, I think that was a bad decision.

I think the best approach now would be to use Cometd. And, if needed, Cometd
can be deployed into non-jetty containers. I've deployed Cometd in JBoss 5.1
and know that it'll work in any Servlet 2.5 container. However, it must be
deployed in Jetty or a Servlet 3.0 container inorder to take advantage of
jetty Continuations. Atmosphere has a Cometd module for non-jetty containers
but it appears to be based on the Cometd 1.x version.

Cometd has some message acknowledgment extensions that should allow similar
message semantics to that achieved with websockets.

-Tad

On Mon, Oct 17, 2011 at 2:17 AM, Vicente J. Ruiz Jurado <v...@ourproject.org
> wrote:

> El 17/10/11 07:45, Joseph Gentle escribió:
> > On Sat, Oct 15, 2011 at 11:14 PM, Vicente J. Ruiz Jurado
> > <v...@ourproject.org> wrote:
> >> El 15/10/11 00:37, Yuri Z escribió:
> >>> I think someone mentioned that probably just updating to  Jetty 7 will
> >>> suffice.
> >>
> >> Sorry, suffice for what? to solve the problem of compatibility between
> >> chrome and socket.io 0.6? I don't think so.
> >>
> >> El 14/10/11 21:08, Nelson Silva escribió:
> >>> Perhaps we should just update to socket.io 0.8 for now and try to
> >>> gather some pros and cons for atmosphere.
> >>
> >> I was looking on that, and seems that is not so easy to update to
> >> socket.io 0.8, or not for me.
> >>
> >> Bests,
> >> --
> >> Vicente J. Ruiz Jurado
> >
> > I don't think there is a socket.io 0.8 server implementation in java.
> > The wire protocol is completely different. As for flashsocket, as well
> > as using an old version of the websocket protocol, it can't traverse
> > HTTP proxies.
> >
> > I'm struggling with the same problems with ShareJS. I've implemented a
> > BrowserChannel server for node.js and I'm looking at extending it with
> > websocket support (where available). It'd be much less work for wave
> > in a box to use atmosphere or something though. .. Or a bayeux
> > implementation of some sort.
> >
> > -J
> >
>
> Yes, seems that socket.io 0.8 is a total rewrite. There is a migration
> document
> https://github.com/LearnBoost/Socket.IO/wiki/Migrating-0.6-to-0.7+
> but...
>
> I was this weekend playing with atmosphere and suffering the Cons
> described in:
> http://www.ibm.com/developerworks/web/library/wa-reverseajax4/index.html
> that, by the way, is a good (and recent) series about this topic. It's
> written by the author of (I think an unmaintained) fork of socketio-java:
> http://code.google.com/p/socketio-java/issues/detail?id=6
>
> Bests,
> --
> Vicente J. Ruiz Jurado
>
> http://comunes.org
> http://ourproject.org
> http://homes.ourproject.org/~vjrj/blog
>
>  "Whenever there is a conflict between human rights and property rights,
>  human rights must prevail." [Abraham Lincoln]
>
>
>
>
>
>
>

Reply via email to