-----Original Message----- From: ogsa-wg-bounces@ogf.org [mailto:ogsa-wg-bounces@ogf.org] On Behalf Of Donal K. Fellows Sent: Friday, April 13, 2007 4:38 AM To: Strong, Paul Cc: ogsa-wg; rm-wg@ogf.org Subject: Re: [ogsa-wg] Glossary terms for Thursday's call
Strong, Paul wrote:
Within the RM-WG we have split attributes into configuration attributes and state attributes. Configuration attributes to some degree define behavior, defining or constraining the set of allowed states, as well as the possible values of state attributes and allowed transitions between states. Clearly configuration attributes are "interesting", and
according to the definition below they would be considered part of
Paul/Donal/All, Thanks for the comments, and sorry to be slow in responding - I'm not able to work on this quite as much these days :( Now that I've taken time to read Paul's comment and Donal's response a little more carefully, I think Donal makes an excellent point - I would expect that an entity would exist only after it has been configured. Its configuration attributes would *contribute* to its state, but would not be a part of it unless the details are associated with (*maybe* contained within) the entity, are visible to some external observer (possibly a debugger), and might change. I'm happy to add a little more explanation in the definition if anyone thinks it would be helpful. For example, at the end of Para 1 I could add: Configuration attributes typically affect an entity's initial state, but would not be considered to be part of the state unless they can be changed during the entity's lifetime, and thus affect its behavior. Other suggestions welcome! - Jem thus the
state. I'm not sure that this is the case.
An argument that I made during the call (and which it might have been nice to have delved into in more detail if we'd had a few days instead of 15 minutes!) is that an altered configuration creates a different resource/stateful entity[*], but an altered state is something that is expected during the normal working lifetime of a resource and does not change the fundamental nature of that resource. That is, a change of config is qualitatively different to a change of state, even if both are (represented as) attributes.
In general, there are loads of definitions of state (I've got a background which leads me to regard it as really a partial function whose domain is the cartesian product of time and names for "stateful" things) but they're just different ways of looking at the same thing. The good aspects of the current definition are that it is observable and changes in it are events, things which are not implicit in the standard CS definitions of the term and yet are very useful for grids.
Donal. [* As you can tell, I've lost track of what kind of thing contains such attributes. I know *I* think of them as resources, but that doesn't make me right. :-) ] -- ogsa-wg mailing list ogsa-wg@ogf.org http://www.ogf.org/mailman/listinfo/ogsa-wg