Turns out the problem wasn't the duplex communication after all.
My callback was trying to write to the UI, but that thread was being
used call the server, so it was blocking there.
I solved the problem by decorating the callback class with
[CallbackBehaviour(UseSynchronizationContext=false)] and writing to
the UI differently.
Thanks anyway,
Tim
On Oct 30, 11:37 am, "mr.tim.long" <[EMAIL PROTECTED]> wrote:
> Hi,
>
> This issue has been driving me a little crazy.
>
> I've setup a wcf service with duplex communication and it works well
> for one way functions. i.e. my contract looks a bit like this
>
> [ServiceContract(SessionMode=SessionMode.Required,
> CallbackContract=typeof(ILogMessaging)]
> public interface IMyContract
> {
> [OperationContract(IsOneWay=true)]
> void DoSomething();
>
> }
>
> The implementation of DoSomething() invokes the callback of
> effectively sends back a log message to the client, and it all works
> nicely, the message appears at the client.
>
> The problem is when I want to put in a function that returns a value
> and uses the callback. So, for example:
>
> [OperationContract]
> DateTime WhatIsTheTime();
>
> with implementation:
> DateTime WhatIsTheTime()
> {
> Callback.LogMessage("you asked for the time");
> return DateTime.Now;
>
> }
>
> The client ends up timing out, and throws a timeoutexception. I've
> read that because the service is single-threaded you get a deadlock
> situation where client and server are both waiting for each other to
> finish, so I guess this is what is causing the timeout. The solution
> they say is to set the concurrency behaviour to Reentrant or Multiple.
>
> But I don't understand this. I've set my callback function to
> IsOneWay=True, and besides, when debugging I can see that execution
> moves past the Callback statement. So I don't know what the client is
> waiting for. I've tried changing the concurrency behaviour to
> Reentrant or Multiple, but that didn't make any difference (was i
> doing it wrong?)
>
> Is there a way around this, or is it not possible to have this type of
> behaviour? Am I trying to have my cake and eat it?
>
> Thanks for your help,
> Tim
