- From: Steve Graham <sggraham@us.ibm.com>
- Date: Wed, 16 Jul 2003 12:41:45 -0400
- To: "Sanjiva Weerawarana" <sanjiva@watson.ibm.com>
- Cc: www-ws-desc@w3.org, www-ws-desc-request@w3.org
Sorry for the late reply on this one. OGSI has the notion of serviceGroup. This concept in OGSI is more like a "collection" or "registry" or "directory" of service instances. Personally, I prefer to consider the concept of OGSI serviceGroup as analogous to collection classes in Java or Smalltalk. I am not sure the serviceGroup concept in OGSI is similar to what WSD is referring to by the term serviceGroup. What Sanjiva mentions is a WSDL description of a service and addtional metadata to describe the groups in which it participates. The serviceGroup concept in OGSI is more like a collection class, in that the relationship between the collection and each member is modelled by the collection and the individual members do not necessarily know which collections it is a part of. Therefore I must agree with Sanjiva on his point that @groups is not terribly useful, as most of the serviceGroups we in the OGSI WG are imagining are much more dynamic than should be modelled in WSDL. I would prefer WSD not add a serviceGroup element or attribute to WSDL. sgg ++++++++ Steve Graham sggraham@us.ibm.com (919)254-0615 (T/L 444) STSM, On Demand Architecture ++++++++ "Sanjiva Weerawarana" To: <www-ws-desc@w3.org> <sanjiva@watson.i cc: bm.com> Subject: Re: Can someone recap the differences between @serviceGroup vs. Sent by: definitions-targetNamespace ? www-ws-desc-reque st@w3.org 07/14/2003 04:12 AM "Jonathan Marsh" <jmarsh@microsoft.com> writes: > > If the remaining concept we wish to express is grouping, the > targetResource is not the most direct way to achieve this. Instead of > introducing the "resource" entity, we can introduce an entity > representing the "group". The relationship between endpoint(s) and the > group is that of "member of", instead of the fuzzier "manipulates". > > There are several ways this could be manifested in the syntax: > a) Add a "serviceGroup" attribute that takes a URI. The URI is an > arbitrary identifier for the "group" to which this service belongs. The > purpose of the group is defined by the creator of the URI. > b) Add a <serviceGroup> element to group <service>s together into an > anonymous group. > c) Both of a) and b). > d) Use external mechanisms such as RDF. > e) Conflate grouping with an existing construct such as <description> or > @targetNamespace. So one way I can read this is that by doing s/targetResource/serviceGroup/g (in SED syntax for those not fortunate enough to have had the pleasure of it) we're getting the concept passed. After all, the membership criteria for the group can be "all <service>s that manipulate the same resource", right? Several people have indicated that a <service> should be able to belong to zero or more groups: <service [groups="list-of-qnames"]> .. </service> I have to agree that a single group concept is rather artificial. OGSi (the gridders) have a similar concept *shockingly* called service group!! The idea is similar (or may even be precisely the same): its a group of services which meet the required membership criteria. So I suspect that Steve & Steve should like this approach as they can model service groups directly at the WSDL level. While I continue to prefer the more rigid @targetResource concept (yes I do believe its more rigid/focused than @groups), I can live with this solution too. I think in most scenarios the grouping is very dynamic; so I don't expect service/@groups to be widely used. That is, I don't believe service authors will statically declare what groups they belong to simply because they won't know that statically. Furthermore, the groups themselves are unlikely to be static. I'm a bit surprised that @groups isn't getting trashed like @targetResource was. To me @groups is even worse at supporting the multiple interface scenario .. Sanjiva.
Received on Wednesday, 16 July 2003 13:52:59 UTC