Core Spec Review meetings - March 29 and 30, 2010
Here are the raw notes from the session. Issues from these notes will be documented on the
OslcCoreV2Issues page.
Notes
Monday March 29
10AM - Noon
- Overall comments
- OSLC defined resources
- Common properties
1:30PM - 3:30PM
- Service Resources
- Delegated UI
Tuesday March 30
10AM - Noon
- Representations
- Authentication
1:30PM - 3:30PM
- Resource Shapes
- Query
----------------------------------------------------------------------------
OSLC Core spec final review session #1/4
SteveS - X
TackT - X
DavidB - X
JamesC - X
PaulM - X
NickC - X
IanGreen
NickC - file/file descriptor guidance needed ASAP
TackT - questions need/use case for multiple shapes/query capability
DaveJ - Extended Error vs. Extended Message
DaveJ - Resouce Update, should use 409 conflict instead? 412?
JamesC - 409 for concurrency, 412 for other issues
JamesC - no If-Match then 412
PaulM - need two response codes here
AI - ensure that two response codes documented
JamesC - versioning solution not workable - separate URIs!?!
SteveS - may be able to use content-types
JamesC - use version header to tell v1 vs v2
JamesC - how about a "latest" header
AI - no header means send earliest/most compatible version
AI - kill the migration service stuff
NickC - common versioning? need to specify what the OSLC-Version means
AI - response & request should indicate core & domain verions
AI: ensure Service Resource says same thing as diagram, 0..N shapes / query
TackT: is one domain / service too restrictive
AI: discuss in service provider session
TackT: clients may not need to be able to discover stable vs. unstable
AI: follow up on paging: do we have redundant mechanisms
AI: do we need a URI type at all now that we have Node
Need to differentiate
a - Properties about a resource at a URI
b - Properties about a link/relationship
rdf:resource - simply refers to resource
rdf:about - adds propeties to
----------------------------------------------------------------------------
OSLC Core spec final review session #2/4
SteveS - X
ArthurR - X
TackT - X
DavidB - X
PaulM - X
JamesC - X
NickC - X
Sofia - X
ArthurR: nice explanation of stripeing
AI: replace word "Node" with 'Resource"
ArthurR: consider placing oslc:nextPage in oslc:Query, even inside plain old resource
Maybe a propery 'oslc:response'
ArthurR: use "strongly discouraged" rather than MUST on the new content-types section.
NickC: encourage use or standard types, discourage gratuitous new types
SteveS: elements vs. terms
AI: do we need anything other than the 'dc' namespace?
ArthurR: shape should declare what RDF type it is defining
AI: yes, add this
ArthurR: How to get shape or shapes given a URI?
JamesC: one instance of resource can have mutliple RDF types
AI: ensure that there's a way to get from resource back to its service provider
oslc:context points to service? oslc:serviceProvider points to provider?
TackT: in service resource can we have a title or type for the shapeURI properties
AI: explore Tack's user case, re: associating shapeURIs with Query Capabilities
SteveS: service provider example had a reporting service, provider supporting 2 services
ArthurR: service provider and elsewhere, use rdf:resource to carry URI value type
ArthurR: domainNamespaceUI is really rdf:type
----------------------------------------------------------------------------
OSLC Core spec final review session #3/4
TackT - X
SteveS - X
JamesC - X
DavidB - X
PaulM - X
NickC - X
IanGreen
RobertElves - X
JamesC - so, what do we do when we need ordering? Use RDF collection to represent this, RDF sequence?
DaveJ - Introduce Ordered Node?
PaulM - re: xml:base. Difference between creation time and update?
JameC - when resource is created, does POSTed representation have to be valid? Server may fill in some required values.
PaulM - some fields are "required but read-only" e.g. the AtomPub ID example
AI: don't require absolute URIs on creation - e.g. XML base URI may be missing
TackT: what is value-type URI for, URI as string?
AI: URI for string, Resource is for true URI. Make this more clear in Defined Resources section.
PaulM: what about starting with RDF/XML then proving mechanism for generating JSON, Atom, etc.
TackT: reporting work-group really needs form-based authentication, would prefer it in core
AI: get Form Based Auth on WG agenda for guidance
RobertE: form based will be loose recommendation
AI: DaveJ to follow-up on need for "those three OAuth URLs"
AI: clarify what two-legged OAuth means
----------------------------------------------------------------------------
OSLC Core spec final review session #4/4
SteveS
ArthurR - X
TackT - X
DavidB
PaulM - X
JamesC - X
NickC - X
re: Resource Shapes
SteveS: allowed values are often computed, allow way to get them via URI
SteveS: allowed/default values must be of same type as property that defines
ArthurR: re: alternative translations, specify use of xml:lang attribute where
appropriate - most appropriate with XML/XHTML value
AI: get xml:lang into spec, RDF/XML section
re: what happens when oslc.properties omitted?
AI: omitted: all "reasonable" properties, e.g. those specifed in resource definition
AI: * means dump of everything that can be safety , system values, etc.
Look for SteveS's ResourceRepresentation
NickC: don't return 403 when missing properties specified in oslc.properties
AI: remove link from core spec
re: POST for query
AI: yes, allow this.
AI: Need predefined namespace prefixes so every query URI does not have to define them
- as properties associated with Service Provider?
re: query item #15: no change
re: query item #16: no change
AI: Arthur to add examples to semantics document
JamesC: re: query item #19 need to be able to query for all things that link to a URI