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

 


Help: OASIS Mailing Lists Help | MarkMail Help

sca-j message

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


Subject: Raw chat room log of Nov 12


Mark Combellack: 9:00 to 9:15 Welcome
Appointment of scribe for day 2
Agenda bashing
 
9:15 - 10:40 SCA Lifecycle
JAVA-65 There is no lifecycle defined for SCA Components
No proposal
 
JAVA-53 what happens if init() throws a runtime exception
Outline proposal in Jira
Proposal at http://lists.oasis-open.org/archives/sca-j/200811/msg00004.html
 
 
10:40 - 11:00 Break
 
11:00 - 12:45 General Issues
JAVA-20 annotations on parameters
Proposals in Jira and http://lists.oasis-open.org/archives/sca-j/200806/msg00017.html
 
JAVA-39 Incorrect example in Java Component Implementation Spec v1.00 - Sec 1.2.4
(Pending proposal from Anish)
 
JAVA-46 equals() method on ServiceReference and CallableReference
Proposal in Jira
 
JAVA-60 Sharing Java artifacts across contributions
Partial proposal outline in Jira
 
 
12:45 - 1:45 Lunch
 
1:45 - 3:40 Access to SCA from non-SCA clients
JAVA-1 Accessing SCA Services from non-SCA component code
(Pending proposal from Mike Edwards)
 
 
3:40 - 4:00 Break
 
4:00 - 6:00 Spring
JAVA-59 SCA Spring C & I specification does not state what happens when a Bean exposed as a service implements mutliple interfaces
Proposal in Jira
 
JAVA-63 SCA Spring C & I specification does not state whether Constructor Injection should be supported
No proposal
 
JAVA-64 SCA Spring C & I specification does not state the purpose of sca:composite element
No proposal
 
JAVA-58 SCA Spring C&I Specification does not have a normative definition of how to calculate the ComponentType of a Spring Application context
No proposal
 
JAVA-92 Spring C&I spec needs to define how effective CT is calculated
No proposal
 
6:00 Adjourn
Mark Combellack: Sorry - wrong agenda - that is the old one. The correct one is below:
Mark Combellack: 9:00 to 9:15 Welcome
Appointment of scribe for day 2
Agenda bashing
 
9:15 - 10:40 SCA Lifecycle
JAVA-65 There is no lifecycle defined for SCA Components
No proposal
 
JAVA-53 what happens if init() throws a runtime exception
Outline proposal in Jira
Proposal at http://lists.oasis-open.org/archives/sca-j/200811/msg00004.html
 
 
10:40 - 11:00 Break
 
11:00 - 12:45 General Issues
JAVA-20 annotations on parameters
Proposals in Jira and http://lists.oasis-open.org/archives/sca-j/200806/msg00017.html
 
JAVA-39 Incorrect example in Java Component Implementation Spec v1.00 - Sec 1.2.4
(Pending proposal from Anish)
 
JAVA-46 equals() method on ServiceReference and CallableReference
Proposal in Jira
 
JAVA-60 Sharing Java artifacts across contributions
Partial proposal outline in Jira
 
 
12:45 - 1:45 Lunch
 
1:45 - 3:40 Scopes
JAVA-21 Clarify Request Scope lifetime
Proposal in Jira and http://lists.oasis-open.org/archives/sca-j/200811/msg00005.html
Previous discussions at http://lists.oasis-open.org/archives/sca-j/200803/msg00034.html
 
JAVA-30 "Process" Scope
Proposal in Jira and at http://lists.oasis-open.org/archives/sca-j/200807/msg00068.html
 
 
 
3:40 - 4:00 Break
 
4:00 - 6:00 Spring
JAVA-59 SCA Spring C & I specification does not state what happens when a Bean exposed as a service implements mutliple interfaces
Proposal in Jira
 
JAVA-63 SCA Spring C & I specification does not state whether Constructor Injection should be supported
No proposal
 
JAVA-64 SCA Spring C & I specification does not state the purpose of sca:composite element
No proposal
 
JAVA-58 SCA Spring C&I Specification does not have a normative definition of how to calculate the ComponentType of a Spring Application context
No proposal
 
JAVA-92 Spring C&I spec needs to define how effective CT is calculated
No proposal
 
6:00 Adjourn
Ashok: Scribe: Ashok
Ashok: Mark explains revised agenda ... replaced Java 1 with Java 21 and 30
Ashok: Topic?:  Java 55 and Java 53 ... re. lifecycle for components and failure modes
Ashok: Anish:  Is this not an Assembly issue?
Ashok: Mark:  Do we need lifecycle model
Ashok: Mike:  What wd a lifecycle model look like?
Ashok: Anish:  We need some stages defined not just for Java ... independent of language
Ashok: That should have been 65 and 53
Ashok: Mike:  One concerm is potential variations in runtimes
Ashok: Some are static and some very dynamic
Ashok: Martin:  We need to know the state ... such as activated ... and what you can do in each state
Ashok: Mike explores situation wrt stateless components
Ashok: Anish:  If I am in destroy method what can I assume?
Ashok: Mike:  If there is no proposal how do we get one?
Ashok: Mike: 3 scopes to worry about.
Ashok: Mike:  Describe timing of init and destroy methods and whet we can do within them.
Ashok: Dave:  I want to talk abt when injection occurs
Ashok: Mike:  We need to describe state of pooled instances
Ashok: Dave:  Shd act like they came out of constructor brand new
Ashok: Mark:  Order in which injections occurs
Ashok: All properties first then all references
Ashok: Mark brings up  constructor injection
Ashok: Anish:  If componets reference each other whose init method get run first
Ashok: When init returns the object is ready
Ashok: Runtime decides if methods can be run on object i.e. init finished
Ashok: Anish:  MUST NOT assume other component is ready when you are in your init
Ashok: Simon:  Cannot guarantee
Ashok: Mark:  split init into 2 phases ... init and postInit
Ashok: Business methods can run before postInit is finished
Ashok: Runtime must start new thread to run postInit
Mark Combellack lowered your hand
Ashok: Simon:  call it threadInit
Ashok: Martin:  call it businessInit
Ashok: sca:Run
Ashok: Mark:  Do we need mirror image for destroy?
Ashok: preDestroy method?
Ashok: Pradeep:  There is a pattern in windowing systems ... close and then destroy
Ashok: Pradeep:  Shd preDestroy return boolean saying you can or cannot destroy me
Ashok: Mark:  is there guidance that you shd not call finalize?
Ashok: Mark:  Java 53 ... what happend if init throws a runtime exception
Ashok: In this case postInit would not be called and object would not transition to running state
Ashok: s/happend/happen/
Ashok: Created, Initializing, Running, Terminating, Terminated
Ashok: What is state when injection occurring?
Ashok: New state: Injected
Ashok: Need state diagram
Ashok: Mike: Can I dispatch to you if in preDestroy state?
Ashok: Mike argues shd not dispatch because runtime knows he is going away
Ashok: Will throw exception
Ashok: Anish:  destroy wd be called only is you are no longer in scope or preDestroyed
Ashok: Mike:  Asks whey we need preDestroy?
Mark Combellack lowered your hand
Ashok: s/whey/why/
Ashok: Mark: Explains graceful shutdown
Ashok: Simon:  predestroy is a courtesy notification from the runtime ... method while in running state
Ashok: Martin:  I think undeploy is another state
Ashok: Discussion on whether postInit and preDestroy are separate states
Ashok: Dave:  preDestroy is like quiesce
Ashok: Break for 15 minutes
Mark Combellack: Resume at 11:10
Ashok: Resuming after break
Ashok: ACTION:  Mark to write up state diagram on blackboard
Ashok: Topic: Java 20 -- Annotations on parameters
Ashok: Proposal on JIRA and in email
Mark Combellack: Java-20 proposal at http://lists.oasis-open.org/archives/sca-j/200806/msg00017.html
Ashok: Simon:  2 weeks ago this was discussed under issue 55 or 43 but no resolution
Ashok: Dave:  This is about references, your's is about propoerties.
Ashok: Simon moves to resolve Jave 20 with text in above email
Ashok: Dave Booz seconds
Ashok: Mike:  The @reference aspect already resolved by resolution to Java 43
Ashok: Seconded by Simon
Ashok: Mike:  Apply to constructor or constructor parameter
Ashok: Simon:  constructor parameter
Ashok: Amendement accepted unanimously
Ashok: Simon moves to resolve Java 20 with above amendment.
Ashok: Anish makes another amendment
anish: The @Property annotation is used to denote a Java class field, a setter method, or a constructor parameter
that is used to inject an SCA property value. The type of the property
injected, which can be a simple Java type or a complex Java type,
is defined by the type of the Java class field or the type of the input
parameter of the setter method or constructor.
Ashok: Simon seconds
Ashok: Amendment accepted unanimously
Ashok: Dave:  If you put @reference on a method parameter, what happens?
Ashok: Error or ignored?
Ashok: Not allowed to do on a setter method, not allowed on a constructor parameter
Ashok: Simon:  We may want to add words start of Chap 8.
Ashok: ACTION: dave Booz to create issue on unsupported use of annotations.
Ashok: Motion on table is text from Anish above noting the @reference aspect already resolved by Java 43
Ashok: Motion accepted unanimously.  Java 20 is now resolved.
Ashok: Topic: Java 39 -- incorrect example in Java C & I spec
Ashok: Anish:  The example was not a good example for that section
Ashok: Chapter 5 around line 308
Ashok: Simon: split into 2 examples illustrating the 2 points
Ashok: Mike:  Does this fall afoul of the issue on Component Type ?
Ashok: Mike:  Mess is caused by uncertainty about what it means to be an unannotated class
Ashok: Mark:  There are many problems with this.  Do we need a new proposal?
Ashok: Simon:  There are much bigger problems than issue 39 discusses.
Ashok: ACTION: Mike to write new proposal for issue 39.  Anish's action is discharged.
Ashok: Topic: Java 46 ... equals method on service reference and callable reference
Ashok: Mike:  I read equals as identical
Ashok: Simon: Do we have an issue on conversations
Ashok: Mark thinks we do not
Ashok: Question abt instance equality vs. type equality
Ashok: Anish:  Add methods to get info about services references
Ashok: Simon:  We still need to define equals
Ashok: Mike:  ... equal if metadata is equal
Ashok: DaveB:  We had a getMetadata method which returned the metadata serialized in XML in SDO form
Mike Edwards: In sca-javacaa-1.1-spec-cd01.doc, 
 
add after line 768:
 
equals( Object anObject ) - returns true if and only if anObject is:
 
a) a CallableReference object
b) anObject relates to identically the same reference metadata as the Callable Reference.
 
otherwise returns false.
 
Note that "identically the same reference metadata" means that the metadata for the reference
must be the same set of metadata in the domain and not merely the same values for the metadata.
In particular, where a reference has 0..n or 1..n multiplicity, metadata for the nth reference is never
identical to the metadata for the mth reference (where m <> n) even if it has the same values.
 
In addition, for a callback reference, since the metadata also involves dynamic information ("EPR")
which cannot be compared equals() can only return true if anObject is identically the same as 
the CallableReference object.
 
add after line 799:
 
equals( Object anObject ) - returns true if and only if anObject is:
 
a) a ServiceReference object
b) anObject relates to identically the same reference metadata as the ServiceReference.
 
otherwise returns false.
 
Note that "identically the same reference metadata" means that the metadata for the reference
must be the same set of metadata in the domain and not merely the same values for the metadata.
In particular, where a reference has 0..n or 1..n multiplicity, metadata for the nth reference is never
identical to the metadata for the mth reference (where m <> n) even if it has the same values.
 
In addition, for a callback reference, since the metadata also involves dynamic information ("EPR")
which cannot be compared equals() can only return true if anObject is identically the same as 
the ServiceReference object.
Ashok: DaveB: Is there anything other than the structural URI in the metadata
Ashok: Mark:  If they had same binding and same structural URI and were considered equal then I would be happy
Ashok: Simon:  I thik we shd remove last para
Ashok: s/thik/think/
Ashok: ACTION:  Mark to write proposal fro Java 46 drawing inspiration from above text.
Ashok: s/fro/for/
Ashok: No further discussion on Java 46
Ashok: Break for lunch
Ashok: Resume at 1:50
Mark Combellack: Dinner venue for tonight:http://www.junction28.com/
Mark Combellack: table booked for 6:30
Pradeep: Scribe: Pradeep
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-60
Pradeep: Mike: We need Import.Java
Pradeep: Discussion about using OSGi
Pradeep: Action item: Dave to work on proposal for JAVA-60
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-21
Mark Combellack: Proposal at http://lists.oasis-open.org/archives/sca-j/200811/msg00005.html
Pradeep: Request scope looks like a conversation.
Pradeep: Mike: We need to understand the use case.
Pradeep: Dave: This is about shared context.
Pradeep: Mike: A session type activity stuff can include conversation stuff as well.
Pradeep: Simon: How about removing request type scope and leaving conversations between 2 parties and detailing about callbacks.
Pradeep: Mike: conversations is an assembly thing.
Pradeep: Anish: Do a really good job of conversations (involving assembly, bindings) or get rid of it.
Pradeep: Mark: A newcomer can look at the spec and say that SCA is stateless and no help for maintaining state.
Pradeep: Mike: Remove conversational interfaces but keep conversation scope.
Pradeep: We should annotate implementations and not interfaces for conversations.
Pradeep: Martin: borrow the BPEL model of correlation sets
Pradeep: Action item: Simon to raise new issue to simplified stateful implementation mechanism to replace conversations.
Pradeep: Action item: Simon to write proposal for the same.
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-59
Pradeep: Discussion about naming the services.
Pradeep: Action item: Mike to raise a new issue for providing names with @Service annotaion
Pradeep: Action item: Mike to work on a new proposal for JAVA-59
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-63
Pradeep: Mike: we should support constructor injection.
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-64
Pradeep: Mike makes a motion to resolve JAVA-64 by removing the SCA composite element from the spring specification
Pradeep: Dave seconds
Pradeep: Motion passes w/o
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-58
Pradeep: Action item: Mike to write up with proposal for JAVA-58
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-92
Pradeep: Mike: Spring already has context file. We do not need another file.
Pradeep: Mike: Leave JAVA-92 open for now. JAVA-58 needs to be done before JAVA-92.
Pradeep: Break for 20 minutes: Resume at 15:35 GMT
Pradeep: Meetting resumes.
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-21
Mark Combellack: http://lists.oasis-open.org/archives/sca-j/200811/msg00005.html
Pradeep: Jim: Remove request scope and consider multi party conversation as a separate issue.
Pradeep: Mike moves to resolve JAVA-21 by remove Request scope from the specification.
Pradeep: Anish seconds
Pradeep: Motion passed w/o
Pradeep: JAVA-21 is now resolved
Pradeep: Topic: http://www.osoa.org/jira/browse/JAVA-30
Mark Combellack: http://lists.oasis-open.org/archives/sca-j/200807/msg00068.html
Pradeep: Dave: user of a composite scoped component deployed in a cluster should see it as just one component.
Pradeep: Dave: Composite scoped components are not singleton in the domain. For example, in the case of component in a cluster.
Jim Marino: brb
Pradeep: Discusssion about clusters, instances, state, @local=true
Jim Marino: ok
Pradeep: Dave: issue with stating that there is a single instance. in a cluster there are many instances. we need to describe this in terms of the callers of the component.
Pradeep: Mark: can we use the term "logical instance"
Pradeep: Mark: Do we need composite scope?
Pradeep: Action item: Dave to come up with proper words for JAVA-30 based on Mike's proposal.
Pradeep: Continuing discussions on Mikes proposal 2.2.3 b
Pradeep: Action item: Anish to raise an issue on @init annotation on a static method.
Pradeep: Discussion about scopes for implementation.composite.
Jim Marino: sorry need to redial
Jim Marino: phone dead one sec
Pradeep: Mark: We need to figure out use cases for process scope.
Pradeep: Jim: Composite scope with multiple instances (because of clustering) not replicating state is what is needed.
Pradeep: Action item: Mark, Jim and Mike to describe use their cases for JAVA-30
Pradeep: Meeting recessed. Will resume tomorrow Nov 13 9:00 GMT

 



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