W3C home > Mailing lists > Public > xml-dist-app@w3.org > October 2000

RE: XMail and XML Protocols (WAS: Removal (Time for XMail?))

From: John Evdemon <john.evdemon@xmls.com>
Date: Sun, 1 Oct 2000 21:03:17 -0400
Message-ID: <2E26643EEFEB974F81E01F521FF2BCECDFD012@NEW.xmlsmail.xmls.com>
To: "'xml-dist-app@w3.org'" <xml-dist-app@w3.org>
I would also like to hear your rationale as to why HTTP is less than
adequate for highly asynchronous messaging.  I have worked with some fairly
high volume systems that utilize HTTP for asynch messaging.  They simply
pass a unique token (such as a UUID or URI value) back and forth to track
the transaction.  Its a simple ACK/NAK protocol:

  Issue request w/token
  Receive ACK w/token (same session)
  Later receive the response w/token
  ACK response w/token (same session)

The token is used to track transaction state and match up responses with
requests.  ACKing in the same session (with the appropriate token value)
ensures that the request or response has been delivered.

I think this approach, properly implemented, enables the use of HTTP in
highly asynchronous, transaction-oriented environments.

Comments?

JohnE


-----Original Message-----
From:  S. Mike Dierken <mike@knownow.com>
Subject: RE: XMail and XML Protocols (WAS: Removal (Time for XMail?))

> HTTP effectively solves the case of relatively
> synchronous request/response architectures, but is less than adequate for
> highly asynchronous messaging systems,

Tell me again why HTTP is less than adequate for highly asynchronous
messaging systems?

MikeD
Received on Sunday, 1 October 2000 21:03:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:57 GMT