[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: Re: Comment on section 3.3.4, Messaging Service Spec v0-1
In case I didn't make this clear (and I didn't!), I think these two properties are orthogonal, and shouldn't overload the same data element. I'm also unclear as to why we would specify the transport-level QoS in the TPA, but specify ebXML reliability/ack-mode/whatever in the message header on a message-by-message basis. -gvh- Gordon van Huizen wrote: > > I think that there are two issues here and that they need to be > decoupled: > > 1) Whether or not the messaging service is to provide ebXML positive > acknowledgments or not. I'd prefer to see this called "MSAck", > "MessagingServiceAcknowledgement" or some such. "AtMostOnce" has a > number of additional connotations that don't apply here, and it really > blurs the line between what the transport and the message service. > > 2) Whether or not the underlying protocol is to employ its own QoS for > reliable messaging. This could certainly be dealt with in the TPA, as > previously discussed. Note that we're pushing QoS normalization issues > off to the TPA discussion if we do this (not that doing so would be > inappropriate). > > WRT the reliability of the RM spec, I don't mean to discredit the work > that's been done. My point is that I believe that we as a group are not > clearly targeting the same scope for this work. If our goal is to > provide ExactlyOnce semantics over HTTP, we need to do a lot more > work--and be clear that this is our intention (the current naming of the > flag points out that we are not all thinking the same thing). > > -gvh- > > Jim Hughes wrote: > > > > Rather than have an extended discussion now on the contents of the "RM > > Spec" (even if it isn't 'reliable', according to individual > > interpretations!), I think the proper conclusion to Change Request 137 is > > to have three values for the ReliableMessagingInfo element: > > > > 1) "None" - meaning, ebXML semantics are silent > > > > 2) "AtMostOnce" - meaning, refer to the "reliable messaging spec" for what > > the Sending MSH and Receiving MSH will do. Yes, there are overtones to the > > term "AtMostOnce", but as opposed to "SeeRMSpec", "AtMostOnce" is a better > > value, IMHO. I think we can require that all compliant MSHs must support > > this value for interoperability. > > > > 3) "TPAdefined" - (or similar) meaning, through the TPA or some other > > means, the two MSHs will agree to use some underlying transport's > > reliability semantics, whether or not some kind of ACK is expected. Support > > of this value is obviously MSH-implementation dependent. > > > > Comments? > > > > Jim
begin:vcard n:Van Huizen;Gordon tel;work:510-848-1988 x-mozilla-html:TRUE url:http://www.sonicmq.com org:Progress Software;XML and Internet Technology adr:;;14 Oak Park;Bedford;MA;01730; version:2.1 email;internet:gvh@progress.com title:Director, Product Management fn:Gordon Van Huizen end:vcard
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC