OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

oslc-core message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: OSLC CORE TC Minutes October 1, 2015


Please note that one of today's action items is for all TC members to review the issue list https://issues.oasis-open.org/browse/oslccore and determine whether there any issues preventing us from going to a public review of the specification.

 

Chair

  • Martin Sarabura

Scribe

  • David Honey

Attendees

  • David Honey (IBM)
  • Harish Krishnaswamy (Sofware AG)
  • Jean-Luc Johnson (Airbus)
  • Jim Amsden (IBM)
  • Martin Pain (IBM)
  • Martin Sarabura (PTC)
  • Nick Crossley (IBM)

Regrets

Resolutions

  • Use new property oslc:iconSrcSet in addition to oslc:icon when a list of icon sizes is available on the server

·         V3 deprecates the use of resource type LocalResource. In all Core shapes, LocalResource shall be replaced with AnyResource with a representation of Inline. The TC believes this change is backward-compatible, and recommends that all domains make the same change to shapes.

·         In resolution to issue OSLCCORE-23 the TC adopts the proposal that ServiceProviderCatalog, ServiceProvider and Service resources from OSLC2 are specific kinds of LDPC resources, and that a ServiceProviderCatalog members include ServiceProviderCatalogs and ServiceProvider LDPCs and ServiceProvider members include Service resources.

Actions

  • Martin S: To determine what other things need to be done for review.
  • All TC members: Look through remaining issues prior to public review especially in light of our resolution of issue 23 above.

Chat transcript from room: oslc

[07:05] David Honey (IBM): David volunteered to be scribe

[07:05] Martin Sarabura (PTC): https://wiki.oasis-open.org/oslc-core/Meetings/Telecon2015.09.17

[07:05] David Honey (IBM): Meeting minutes accepted

[07:06] David Honey (IBM): [15:06] List of attendees: David Honey (IBM), Harish (SoftwareAG), Jean-Luc Johnson (Airbus), Jim Amsden (IBM), Martin Pain (IBM), Martin Sarabura (PTC), Nick Crossley (IBM)

[07:06] Nick Crossley (IBM): Nick also attending


[07:07] Martin Sarabura (PTC): https://issues.oasis-open.org/browse/OSLCCORE-15

[07:10] David Honey (IBM): Proposed vote by Jim, seconded Nick

[07:10] Nick Crossley (IBM): +1

[07:11] David Honey (IBM): Proposal is for new property oslc:iconSrcSet

[07:11] David Honey (IBM): +1

[07:11] Martin Sarabura (PTC): +1

[07:11] Harish (SoftwareAG): +1

[07:11] Martin Pain (IBM): +1

[07:11] Jean-Luc Johnson (Airbus): +1

[07:11] Jim Amsden (IBM): +1

[07:12] David Honey (IBM): proposal approved.


[07:12] Martin Sarabura (PTC): https://issues.oasis-open.org/browse/OSLCCORE-23

[07:15] David Honey (IBM): Jim: Looked at all operations a client might want to do for discovery, assuming OSLC 2 ServiceProviderCatalog, ServiceProvider etc were LDPCs. Raised 4 issues from that review.

[07:16] Nick Crossley (IBM): Is it possible to get a link to the relevant email in the archive?

[07:17] David Honey (IBM): 1) Separation of OSLC 2.0 from 3.0 if server wants to implement only one of them.

[07:20] David Honey (IBM): 2) Increased complexity? OSLC 3 client has access to usual service provider properties, and/or can use link headers.

[07:23] David Honey (IBM): 3) Overloads GET for OSLC 2 clients. Will GET service provider properties, plus members of LPDC.

[07:26] David Honey (IBM): A service provider needs to provide its member services inline. So no issue.

[07:29] David Honey (IBM): 4) if a service was an LPDC, it could be a referenceable as a resource. Is there a conflict between requirements for inline versus having a URI for a referencable resource? No, can include multiple resources in the same RDF graph.

[07:32] David Honey (IBM): If service provider catalog, service provider, and service are LPDCs, the service LPDC members could be top-level LPDCs.

[07:33] David Honey (IBM): Martin: Adding layers might be more burdensome for clients to discover.

[07:36] Martin Pain (IBM): Decision can be left to individual domains, as services are tied to a domain

[07:36] David Honey (IBM): Implementations can choose whether to use nested LPDCs, part of a domain spec.

[07:39] David Honey (IBM): David: Cost of multiple gets each time discovery is needed.

[07:41] David Honey (IBM): Martin: Could use caching/ETags, and HEAD to determine whether there are new service providers.

[07:43] David Honey (IBM): Nick: Selective properties being carried from OSLC 2 to 3?

[07:43] David Honey (IBM): Jim: Has to be for compatibilty.

[07:44] David Honey (IBM): Jim: Future direction might be to leverage LDP query.

[07:46] Nick Crossley (IBM): BTW, I have to drop from this call at 8am Pacific time.

[07:46] David Honey (IBM): Martin P: GET on LDPC must contain ldp:contains triples.

[07:49] David Honey (IBM): Nick: an LPDC can have a default behaviour with a preference to exclude member properties.

[07:51] David Honey (IBM): Example from Config Mgmt spec: A component LPDC gets links to top level components.

[07:52] David Honey (IBM): Some domains might provide an empty LPDC if they don't have a capability to return all resources [of some type].

[07:54] David Honey (IBM): Worst case: OSLC 2 provider, maps its service to contain all the change requests, return all services, and each service includes a reference to all change requests for that service. We need recommended best practice so they don't do this.

[07:55] David Honey (IBM): Martin: Prefer headers - do they apply to nested LPDCs?

[07:55] David Honey (IBM): Nick: Undefined by current spec.

[07:56] David Honey (IBM): Two possible behaviours: Prefer applies to whole HTTP request (all resources returned), or only applies to the resource matching the request URI.

[07:57] David Honey (IBM): Should we ask W3C for clarification: action on Nick to do so.

[08:00] David Honey (IBM): Martin: Suggestion in LPD primer/spec, use single type in a container. Might be two types SPCs and SPs.

[08:03] David Honey (IBM): Martin P: Need to agree that a reference from a service provider to service can be relaxed - current OSLC 2 spec requires it to be a blank node. We should relax it so it has to be inline (in the same RDF graph), but not necessarily a blank node.

[08:03] Nick Crossley (IBM): I'm back - other meeting was cancelled

[08:08] Nick Crossley (IBM): Proposal: OSLC deprecate use of resource type LocalResource, and treat it as equivalent to AnyResource with a mandatory representation of Inline.

[08:08] David Honey (IBM): Martin P: Will raise a separate issue for relaxing Service provider to service from being a blank node/.

[08:10] Nick Crossley (IBM): Proposal: OSLC 3.0 to define resource type LocalResource as equivalent to AnyResource with a mandatory representation of Inline.

[08:10] Martin Pain (IBM): Proposal: The TC agrees that changing a resource shape property from being LocalResource to being AnyResource with a representation of Inline is not backwards-compatible.

[08:11] Martin Pain (IBM): Proposal: The TC agrees that changing a resource shape property from being LocalResource to being AnyResource with a representation of Inline IS backwards-compatible.

[08:14] Nick Crossley (IBM): Proposal: OSLC deprecate use of resource type LocalResource. In all Core shapes, LocalResource shall be replaced with AnyResource with a representation of Inline. The TC believes this change is backward-compatible, and recommends that all domains make the same change to shapes.

[08:16] Martin Pain (IBM): +1

[08:17] David Honey (IBM): Needs representation to be [re]included in OSLC 3.

[08:17] Nick Crossley (IBM): +1

[08:17] Jim Amsden (IBM): +1

[08:17] David Honey (IBM): +1

[08:17] Harish (SoftwareAG): +1

[08:17] Martin Sarabura (PTC): +1

[08:17] Martin Sarabura (PTC): Proposal accepted


[08:18] David Honey (IBM): https://issues.oasis-open.org/browse/OSLCCORE-23

[08:19] Martin Pain (IBM): Proposal: That v3 SPCs MUST be LDPCs (of SPCs and SPs). SPs MUST be LDPCs of Services. Services MUST be LDPCs, whose members are of type(s) defined by the Service's domain. The "best practice" suggestion to domain committees is to have one LDPC pre resource type within a Service.

[08:19] Jim Amsden (IBM): Proposal: In resolution to issue OSLCCORE-23 the TC adopts the proposal that ServiceProviderCatalog, ServiceProvider and Service resources from OSLC2 are specific kinds of LDPC resources, and that a ServiceProviderCatalog members are ServiceProviderCatalogs and ServiceProvider LDPCs and ServiceProvider members include Service resources.

[08:20] Jim Amsden (IBM): Proposal: In resolution to issue OSLCCORE-23 the TC adopts the proposal that ServiceProviderCatalog, ServiceProvider and Service resources from OSLC2 are specific kinds of LDPC resources, and that a ServiceProviderCatalog members include ServiceProviderCatalogs and ServiceProvider LDPCs and ServiceProvider members include Service resources.

[08:22] Jim Amsden (IBM): Note: Service is the point at which Domain specifications specify their specific service capabilities.

[08:22] Jim Amsden (IBM): +1

[08:22] Nick Crossley (IBM): +1

[08:22] Martin Pain (IBM): +1

[08:22] Harish (SoftwareAG): +1

[08:22] David Honey (IBM): +1

[08:22] Martin Sarabura (PTC): +1

[08:22] Jean-Luc Johnson (Airbus): +1

[08:23] Martin Sarabura (PTC): Proposal accepted


[08:24] David Honey (IBM): Jim: Ask TC members look through remaining issues prior to public review.

[08:25] David Honey (IBM): Nick: Are there rules on what needs to be complete before public review?

[08:30] David Honey (IBM): Nick: Action to work out why Config Mgmt use of Respec is broken.

[08:31] David Honey (IBM): Martin: To determine what other things need to be done for review.

[08:31] David Honey (IBM): Meeting adjourned.



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]