WS-Eventing joins the WS-Thingy working group proposal

The original proposal for a “WS Resource Access Working Group” mentioned that WS-Eventing might later join the party. It’s now done, and the proposed name for this expanded W3C working group is “WS Resource Interaction Working Group”.

It takes me no effort to imagine the discussions that turned “access” into “interaction”. Which means I am not cured yet, after a year of post-standards therapy.

IBM hurried to “clarify” how, in their view, this proposal relates to the existing WS-Notification standard. The logic seems to be: WS-notification is a great general-purpose pub/sub spec, WS-Eventing is a pub/sub spec used in the device management spec, to prevent confusion we will make them overlap completely by making WS-Eventing another general purpose pub/sub spec.

Someone who’s been paying attention asks how this relates to the WSDM/WS-Management convergence. IBM’s answer is a model of understatement: “other activities in the WS community should not delay their work in anticipation of new documents being produced”.

As the sign at New York’s pier 59 might have read in 1912: “visitors expecting to great RMS Titanic passengers should not delay their activities in anticipation of the boat arriving in the harbor”.


Filed under Everything, IBM, IT Systems Mgmt, SOAP, Specs, Standards, W3C

2 Responses to WS-Eventing joins the WS-Thingy working group proposal

  1. Most of the W3C have not yet figured out that there SHOULD be a difference between protocols for managing the service interface and a management protocol for the service.

    Trying to roll everything into 1 protocol will doom the result into the same oblivion as the entire ISO protocol suite and for the exact same reasons.

    Stop the “kitchen sink” insanity and drive for clarity via separating these two very different concerns! The IBM proposal, perhaps for the wrong reasons, is actually a small step in the right direction.

    Meanwhile if WS-Man is to survive as a management protocol, it urgently needs a strict subset profile defined based on minimalist simplicity that rips out the fluff of optimizations and over specification.

    There is a lesson to be learned from the history and success of SNMP here.

  2. Pingback: William Vambenepe’s blog » Blog Archive » State modeling: party over, go home now.