On 1/10/12 6:01 AM, Philipp Tarasiewicz wrote:
> Ok, I understand.
>
> Regarding the mentioned unidirectional heartbeats, I think that
> besides ping() a pong() method is equally meaningful.
>
>
> In the meantime, would you (or Ryan) mind posting some code showing
> how to send a ping using the current (2.1.8) API?
> Would be very helpful since then I could run some tests with the
> ping/pong feature.
We've logged;
http://java.net/jira/browse/GRIZZLY-1174
>
>
> Thanks! And keep on doing such a good work! :-)
>
> 2012/1/10 Justin Lee <jlee_at_antwerkz.com <mailto:jlee_at_antwerkz.com>>
>
> And if you ever find yourself doing something like "new
> DataFrame(...)," you're almost certainly doing it wrong. The API was
> designed such that you'd never have to do deal with DataFrame
> directly. But, yeah, it looks like the we probably could use a ping()
> on the websocket class.
>
> On Tue, Jan 10, 2012 at 8:00 AM, Ryan Lubke <ryan.lubke_at_oracle.com
> <mailto:ryan.lubke_at_oracle.com>> wrote:
> > On 1/10/12 3:26 AM, Philipp Tarasiewicz wrote:
> >
> > Ah okay, I see.
> >
> > So, in the case I want to perform ping/pong keepalives, the
> question is: How
> > can I send a valid PingFrame from the server to the client?
> > (Seems like I was too cheeky... :-))
> >
> >
> > However, the Pong chapter in the WebSocket RFC (ch. 5.5.3,
> > http://www.rfc-editor.org/rfc/rfc6455.txt) says something like:
> > "A Pong frame MAY be sent unsolicited. This serves as a
> unidirectional
> > heartbeat. A response to an unsolicited Pong frame is not expected."
> >
> > That's exactly what I want to achieve with the client code from
> my last
> > snippet. The client shall send heartbeats to the server at a
> specified
> > interval.
> > Consequently, I need to find a way to send valid PongFrames from
> the client
> > to the server.
> > Any hints on this one? :-)
> >
> > Looks like we'll need to work on that from our end. I'll follow
> up once
> > something is ready.
> >
> >
> > Thanks in advance!
> > Philipp
> >
> > 2012/1/10 Ryan Lubke <ryan.lubke_at_oracle.com
> <mailto:ryan.lubke_at_oracle.com>>
> >>
> >> On 1/9/12 2:42 PM, Philipp Tarasiewicz wrote:
> >>
> >> It turned out that I still have complications to send a valid
> PongFrame
> >> while implementing a heart-beat mechanism on my websocket client.
> >>
> >> Here is the relevant snippet:
> >>
> >> // My WebSocketApplication subclass
> >>
> >> @Override
> >> public void onPong(WebSocket socket, byte[] bytes) {
> >> System.out.println("*** *** onPong() *** ***");
> >> super.onPong(socket, bytes);
> >> }
> >>
> >> // My client's PongFrame send invocation
> >> webSocket.send(new DataFrame(new PongFrameType(), new byte[0],
> >> true).getBytes());
> >>
> >> That's not going to work since the binary representation of the
> frame you
> >> create will be the payload and not the frame itself.
> >>
> >> If the Server is sending the ping, the pong should be sent
> automatically
> >> so you shouldn't have to do this.
> >>
> >>
> >> Syntax highlighted version:
> >> http://pastebin.com/ADSGKBxS
> >>
> >>
> >> All I want to accomplish is to let the client send a valid
> PongFrame, such
> >> that my overriden onPong() callback gets invoked on the server
> side.
> >>
> >> Would you be so kind and provide a snippet showing a sending of
> a valid
> >> PongFrame? :-)
> >>
> >> 2012/1/9 Philipp Tarasiewicz <justphilmusic_at_googlemail.com
> <mailto:justphilmusic_at_googlemail.com>>
> >>>
> >>> Hi Ryan,
> >>>
> >>> I haven't any code that is worth mentioning, since I've obviously
> >>> misinterpreted the related API docs.
> >>>
> >>> My initial thoughts after having studied the docs were like:
> >>> Grizzly automatically performs ping/pong cycles every x seconds to
> >>> maintain the connection to every connected remote client and
> correspondingly
> >>> invokes the related callback methods in the
> WebSocketApplication class.
> >>> - onPing(WebSocket socket, byte[] bytes) : Server is about to
> ping
> >>> WebSocket 'socket' by sending 'bytes' bytes.
> >>> - onPong(WebSocket socket, byte[] bytes) : Server has
> received pong
> >>> information 'bytes' from remote client connected to WebSocket
> 'socket'.
> >>>
> >>> So, I've overriden these methods in my WebSocketApplication
> subclass
> >>> (just for logging purpose), started the server, connected a
> client to it and
> >>> have been expecting that server/client will automatically
> perform ping/pong
> >>> cycles every x seconds.
> >>>
> >>>
> >>> But finally, after having read your and Justin's responses +
> having a
> >>> closer look at your corresponding code (just cloned the repo),
> it is
> >>> completely enlightening that it could not work this way.
> >>>
> >>> In particular, I've realized how you send the ping / pong frames.
> >>> Consequently, it shouldn't be a big deal for me to implement
> the behavior
> >>> that I had initially expected.
> >>>
> >>> Nevertheless, I will also log a feature request. :-)
> >>>
> >>>
> >>>
> >>> There is one more question I'm interested in. But, I should
> probably ask
> >>> it in a new thread since it does not related to this topic.
> >>>
> >>> Thank you guys for your help!
> >>>
> >>> Phil
> >>>
> >>>
> >>> 2012/1/9 Ryan Lubke <ryan.lubke_at_oracle.com
> <mailto:ryan.lubke_at_oracle.com>>
> >>>>
> >>>> On 1/9/12 7:19 AM, justphilmusic_at_googlemail.com
> <mailto:justphilmusic_at_googlemail.com> wrote:
> >>>>>
> >>>>> Hi Ryan!
> >>>>>
> >>>>> Until now I've just played around with the WebSocketClient
> included in
> >>>>> Grizzly's 2.1.8 WebSocket API. So, it's the websocket
> version that the
> >>>>> WebSocketClient class is using.
> >>>>> I haven't used a brower yet.
> >>>>
> >>>> To go back to your original question, I've locally tested
> that pings are
> >>>> responded to automatically, however, I'm making some changes to
> >>>> make sending pings a little more straight forward.
> >>>>
> >>>> Can you share your code? Would like to see what you're doing
> >>>> different.
> >>>>
> >>>>>
> >>>>> The question is:
> >>>>> Does Grizzly's WebSocket implementation perform any connection
> >>>>> maintainance / heart-beating under the hood or do I have to
> implement
> >>>>> the logic by myself?
> >>>>
> >>>> At this point, there is no heart-beat feature. However, it
> probably
> >>>> wouldn't
> >>>> be a bad feature to offer. I would recommend logging a
> feature request.
> >>>>
> >>>>>
> >>>>>
> >>>>> Best regards,
> >>>>> Phil
> >>>>
> >>>>
> >>>
> >>
> >>
> >
> >
>
>
>
> --
> You can find me on the net at:
> http://antwerkz.com
> http://antwerkz.com/+
> http://antwerkz.com/linkedin
> http://antwerkz.com/twitter
>
>