[Fwd: Re: Two way channel roles?]

Sorry. I hadn't realized I'd not sent this to the whole list.

Andy Dent wrote:

>Should this be two channels or one?
>  
>
I would do it as one channel, assuming the messages can by handled 
synchronously and have essentially the same format. That is, if you have 
a problem with the server and client both updating the same value at the 
same time, you need to think about this more. Otherwise, you should be 
OK. Making it one channel probably makes it a little easier to implement.

>I *think* these are two profiles, each running over one channel.
>
That's how I would do it. You want the parameter settings to be 
asymchronous relative to the events.

>This is where the lack of really detailed examples has confused me when we
>get into race territory.
>
It's always better (IMHO) to have detailed formal specifications than 
extensive examples. Of course, if you can have both (and keep them 
consistant) then all the better.

>Take the example of the Triggers, say a client sends back a CancelTrigger
>after a Trigger has been running for a while - are both the following
>valid, which is what I've assumed?
>
>
>I:	MSG 1 TriggerRequest
>L:	RPY 1 ok
>L:	MSG 1 TriggerEvent
>I:	RPY 1 ok
>...
>L:	MSG 1 TriggerEvent
>I:	MSG 4 CancelTrigger     // cancel sent before processing the
>TriggerEvent
>L:	RPY 4 ok
>I:	RPY 1 ok
>
>OR
>L:	MSG 1 TriggerEvent
>I:	MSG 4 CancelTrigger     // cancel sent before processing the
>TriggerEvent
>I:	RPY 1 ok
>L:	RPY 4 ok  // Listener has received the MSG and RPY before having
>time to send the RPY
>  
>
Yes. The way you know is to look at the section of the RFC 3080 
(basically, near page 11) that says basically "a message is valid if..."






-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com