- From: Kohei Honda <kohei@dcs.qmul.ac.uk>
- Date: Mon, 30 Oct 2006 09:22:14 +0000
- To: Gary Brown <gary@pi4tech.com>
- CC: "Monica J. Martin" <Monica.Martin@Sun.COM>, Martin Chapman <martin.chapman@oracle.com>, 'WS-Choreography List' <public-ws-chor@w3.org>
A note on co-relation identity and session: I wrote about corelations: I would like to note its historical context. As fart as I know, this idea was first presented in scientific texts in definite forms in two ways: (1) Nonces (challenge strings) in Needham-Shroeder protocol. (2) Private name passing in the pi-calculus. Further going back in the history, similar ideas can be found in actor models, TCP's ack, and the use of document identities in financial processing. Making interactions collected into one by this device into types for conversation (in the way we know of as session types now) were done by many people including myself since 1994. As noted, these are called session types. I do not know from where this term "corelation" comes from. I think it is also a possible name. Personally I myself think "session identities" (and associated "session instances") may denote its meaning more clearly. Session is a fundamental idea in all high-level (and low-level) communications among computing agents (just consider your session with any web service for human, filling fields etc.). It emphasises more in what context conversations are going on rather than individually relating this message and others. So the latter (co-relation) does have a place different from session. My preference is to use the corelation for more application-oriented situations, such as invoice numbers. After understanding the concepts clearly, a good terminology will naturally arise. kohei
Received on Monday, 30 October 2006 09:22:48 UTC