[Oslc-recon] requirement that a provider MUST provide an oslc:serviceProvider property for its resources
Tuan Dang
tdang at us.ibm.com
Wed Feb 20 15:25:00 EST 2013
The scenario (
http://open-services.net/wiki/reconciliation/Scenario-Coverage-Report/ )
calls for reporting on the delta between
monitored resources and discovered resources.
We'd be able to determine if a reconciled resource has a reconcilable
resource from a monitoring SP and a reconcilable resource from a
discoverySP
Thanks ! T
Tuan Dang
Tivoli OSLC governance, OSLC Reconciliation workgroup lead, Tivoli Common
Data Model
Internet: tdang at us.ibm.com
phone: (919) 224-1242 T/L 687-1242
From: Tuan Dang/Raleigh/IBM at IBMUS
To: oslc-recon at open-services.net,
Date: 02/20/2013 03:03 PM
Subject: [Oslc-recon] requirement that a provider MUST provide an
oslc:serviceProvider property for its resources
Sent by: "Oslc-Recon" <oslc-recon-bounces at open-services.net>
In the spec :
> Reconciliation service providers MUST provide a oslc:serviceProvider
property for their defined resources that will be the URI to a Service
Provider Resource.
Discussion so far:
>From Mike Fiedler whom I asked about the same req in the Automation spec
>The reason it is there is for the benefit of the consumer. For the V2
automation scenarios, the thought was that a Service Provider is the
starting point for generic >discovery of query capabilities, creation
factories and delegated UIs. Strictly speaking, I guess it does not have
to be a MUST, but consumers would require "special" >knowledge to know
which URLs to hit on the provider for these services.
..
>OK, I understand. Some folks are bigger fans of the Service Provider
concept than others. My feeling, from having been a consumer of
providers from many domains, is that they provide consumers with a
well-known starting point for interacting with a provider and figuring out
what it can or cannot do.
Reply from John Arwe:
>I think we're mentally asking the question "what scenario leads to its
inclusion" - channel Martin ;-) . It may be an unstated one (that either
should be rendered explicit >and adopted, or should be explicitly
disowned). If it's got anything to do with fanboi-ism, we're already off
The Path.
>It would be easier perhaps if there was one (in Core maybe) that drove
the need. Otherwise we need to do so w/in each domain. My strawman for
Recon follows.
>What I could hear between the lines here is looking at things from the
client end, a point of view not necessarily where the Recon weenies would
start. They start from: >you have 2 things and you want to know if
"they're really the same" or not, based on their properties. Dicey
business in URL space, but routine in meatspace.
>If OTOH you start from (ala "it was a dark and stormy night. a shot rang
out."): "someone hands you a URL, what now?" then you'll get a different
answer. If >(distancing ourselves from the Tivoli *implementation* for a
moment) you get a URL, you GET its representation, and it turns out that
it is of a type that has reconciliation >rules and it also "is
reconcilable" (satisfies a set of property-existence and content format
constraints), ... now what? The usual OSLC scenario approach is a brute
>force "how would a user do it" one, as a starting point ("the simplest
thing that could possibly work", as Bill Higgins was wont to say). We
have a least on scenario in >Reconciliation, the fabled coverage report.
Does that drive a need for oslc:SP in a non-Tivoli implementation? May-be.
>Simplest form of coverage report is for just this 1 input URL, who has
data about "it" (by which I mean, the set of URLs that is output from the
reconciliation process... to >an omniscient observer with perfect
reconciliation results in hand, which URLs would reconcile with the input
URL). For each of those, what you'd need for a coverage >report is
(minimally) product name and link to its details. Here, I think that
means scavenge around through every SP you know how to find (doesn't
matter how you find >them), query each for its resources of the same type,
GET each, decide if same/different from your starting point. There is NO
guarantee that the set of SPs you >already know about includes the one who
minted the input URL, so how do you find it (for product name) if not via
oslc:SP? That seems like a good scenario to render >explicit. It does
make a jump from SP to product, so just use "provider title" instead of
"product name", done.
Thanks ! T
Tuan Dang
Tivoli OSLC governance, OSLC Reconciliation workgroup lead, Tivoli Common
Data Model
Internet: tdang at us.ibm.com
phone: (919) 224-1242 T/L 687-1242
_______________________________________________
Oslc-Recon mailing list
Oslc-Recon at open-services.net
http://open-services.net/mailman/listinfo/oslc-recon_open-services.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://open-services.net/pipermail/oslc-recon_open-services.net/attachments/20130220/5a2a23f0/attachment-0003.html>
More information about the Oslc-Recon
mailing list