[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [sca-j] Minutes for Thursday afternoon?
Hi Mike, Here is the whole chat log for Thursday. Thanks, Mark [8:49] Mark Combellack: Agenda: 9:00 to 9:15 Welcome Appointment of scribe for day 3 Agenda bashing Conference call on Monday 17th November 9:15 - 10:40 Concurrency JAVA-61 Describe the concurrency model for each scope No proposal JAVA-62 Clarify what a Component Implementation can do with threads Outline for proposal in Jira JAVA-8 Concurrency model for Service Reference instances Proposal in Jira JAVA-10 State sharing between ServiceReference and type-safereference Proposal in Jira 10:40 - 11:00 Break 11:00 - 12:45 Concurrency (continued) 12:45 - 1:45 Lunch 1:45 - 3:40 Component Type/unannotated POJOs JAVA-55 SCA Java Specifications do not Adequately Define the ComponentType of a Java implementation Proposal in Jira JAVA-17 A hole in the algorithm of introspecting property/reference from an unannotated impl class Proposal in Jira JAVA-87 Java-related C&I specs needs to define how effective CT is No proposal 3:40 - 4:00 Break 4:00 - 6:00 Bindings/Interfaces JAVA-77 A remotable service SHOULD be translatable into a generally accepted standard for a service, such as WSDL 1.1 or WSDL 2.0 Proposal in Jira JAVA-2 Determining the data binding to use (e.g. JAXB or SDO) No proposal JAVA-7 Support for JAX-WS style async methods No proposal JAVA-51 More examples on mapping to Java No proposal 6:00 Adjourn [9:14] Mark Combellack: We are about to get started and will set up the phone call [9:20] anonymous morphed into anish [9:20] Mike Edwards: Agenda Bashing: [9:20] Mike Edwards: replace items currently from 16:00 - 18:00 with a consideration of Testing [9:26] Mike Edwards: Administrivia [9:26] Mike Edwards: Motion to cancel Java TC call on Monday 17th [9:26] Mike Edwards: Proposed Martin, Ashok seconds [9:27] Mike Edwards: Accepted w/o [9:27] Mike Edwards: Motion to extend Monday TC calls to start 30 minutes earlier [9:27] Mike Edwards: Proposed Dave B, seconded Mike E [9:27] Mike Edwards: Accepted w/o [9:28] Mike Edwards: "Concurrency" Section [9:28] Mike Edwards: JAVA-61 Describe the concurrency model for each scope [9:29] Mike Edwards: http://www.osoa.org/jira/browse/JAVA-61 [9:29] Mike Edwards: No proposal at present [9:30] Mike Edwards: 1) Stateless Scope [9:31] Mike Edwards: Single threaded - the SCA runtime MUST ensure that any implementation instance object only ever has ONE thread runnoing within it at any one time [9:31] Mike Edwards: 2) Composite Scope [9:32] Mike Edwards: Multi threaded - the SCA runtime MAY run multiple threads within a single implementation instance object and it DOES NOT perform any synchronization [9:33] Mike Edwards: 3) Conversation scope [9:35] Mike Edwards: Multi-threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it DOES NOT perform any synchronization [9:37] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST only dispatch any implementation instance object on ONE thread at a time [9:38] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST only use one thread to dispatch any implementation instance object at a time [9:48] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST only dispatch any implementation instance object on ONE thread within the SCA lifecycle of the instance. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [9:54] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST only dispatch any implementation instance object ONCE on ONE thread within the SCA lifecycle of the instance. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [9:55] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST only dispatch any implementation instance object ONCE for one business method on ONE thread within the SCA lifecycle of the instance. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:00] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object only ever has ONE thread runnoing within it at any one time and in addition, the SCA runtime MUST only dispatch any implementation instance object ONCE for one business method on ONE thread within the SCA lifecycle of the instance. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:02] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time and in addition, the SCA runtime MUST only dispatch any implementation instance object ONCE for one business method on ONE thread within the SCA lifecycle of the instance. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:06] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only ONE business method ONCE on ONE thread. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:06] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only dispatch ONE business method ONCE on ONE thread. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:19] Simon Nash: .. make a single invocation of one business method [10:19] Simon Nash: Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only make a single invocation of one business method. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) [10:20] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only make a single invocation of one business method. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) 2) Composite Scope Multi threaded - the SCA runtime MAY run multiple threads within a single implementation instance object and it DOES NOT perform any synchronization 3) Conversation scope Multi-threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it DOES NOT perform any synchronization [10:22] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that any implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only make a single invocation of one business method. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) 2) Composite Scope Multi threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it MUST NOT perform any synchronization 3) Conversation scope Multi-threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it MUST NOT perform any synchronization [10:27] Mike Edwards: 1) Stateless Scope Single threaded - the SCA runtime MUST ensure that an implementation instance object is only ever dispatched on ONE thread at any one time. In addition, within the SCA lifecycle of an instance, the SCA runtime MUST only make a single invocation of one business method. (NB the SCA lifecycle might not correspond to the Java object lifecycle due to runtime techniques such as pooling) 2) Composite Scope Multi threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it MUST NOT perform any synchronization 3) Conversation scope Multi-threaded - the SCA runtime MAY run multiple threads in a single implementation instance object and it MUST NOT perform any synchronization [10:28] Mike Edwards: Mike E moves to resolve Issue 61 with the wording as above [10:29] Mike Edwards: Martin seconds [10:30] Mike Edwards: (Noting that editorial work will be required) [10:31] Mike Edwards: Accepted w/o - Issue 61 is resolved. [10:31] Mark Combellack: Breaking - resuming at 10:50 GMT [10:50] Mike Edwards: JAVA-62 Clarify what a Component Implementation can do with threads Outline for proposal in Jira [10:50] Mike Edwards: http://www.osoa.org/jira/browse/JAVA-62 [11:22] Simon Nash: http://gee.cs.oswego.edu/dl/concurrencyee-interest/index.html [11:26] Mike Edwards: Direction #1 [11:26] Mike Edwards: Need to create a series of statements relating to the implementation code containing "SHOULD" and "SHOULD NOT" clauses [11:27] Mike Edwards: Need to create a corresponding series of statements relating to the SCA runtime containing "MAY" clauses concerning what the runtime might do if an implementation chooses to use/abuse threading capabilities [11:28] Mike Edwards: eg: [11:28] Mike Edwards: A stateless scope implementation SHOULD NOT attempt to start, stop, suspend, or resume a thread, or to change a thread's priority or name. [11:33] Mike Edwards: Direction #2 [11:34] Mike Edwards: Need to create a controlled set of facilities for SCA implementations to use in the areas of Work management and Timers, parallel to the specification of JSR 236 [11:36] Mike Edwards: eg: An SCA runtime MAY refuse to provide any runtime facilities for a thread that it does not create/control (eg refuse to allow invocation of service reference methods) [11:38] Mike Edwards: Action: (Mark C) Draw up some wording for Direction 1 and publish to the list [11:39] Mike Edwards: Action: (Mark C + others prepared to help) Investigate the Workmanager JEE spec and determine its applicability to SCA [11:40] Mike Edwards: JAVA-8 Concurrency model for Service Reference instances Proposal in Jira [11:40] Mike Edwards: http://www.osoa.org/jira/browse/JAVA-8 [11:43] Mike Edwards: The real problem here is mutability of service references - associated with Callbacks and Conversations [11:43] Mike Edwards: If the mutability is removed, then Issue 8 gets resolved very simply [11:44] Mike Edwards: This depends on the resolution of some other issues [11:44] Mike Edwards: As a result, cannot progress this issue at this time [11:47] Mike Edwards: JAVA-10 State sharing between ServiceReference and type-safereference Proposal in Jira [11:47] Mike Edwards: http://www.osoa.org/jira/browse/JAVA-10 [11:48] Mike Edwards: Again, this depoends on the question of mutability of service references [12:10] Mike Edwards: Direction for resolution: [12:11] Mike Edwards: IF mutability of references is removed (Issue 95) THEN the need for ServiceReference is removed - and by making reference proxies guaranteed to be serializable, all necessary function is present. [12:12] Mike Edwards: Make Issue 10 dependent on the resolution of Issue 95 [12:13] Mark Combellack: Breaking for lunch - resuming at 1:10 GMT [13:12] Simon Nash: scribe: SimonN [13:13] Simon Nash: next issue: JAVA-55 [13:13] Simon Nash: JAVA-55 SCA Java Specifications do not Adequately Define the ComponentType of a Java implementation [13:13] Simon Nash: related issue is JAVA-87 [13:14] Simon Nash: maybe one or other of these is a duplicate [13:20] Simon Nash: document attached to email from Mike Edwards sent Aug 4 [13:20] Simon Nash: http://www.oasis-open.org/apps/org/workgroup/sca-j/email/archives/200808/msg00006.html [13:26] Simon Nash: SimonN: it seems we have no way to attach different combinations of intents to different services provided by an implementation [13:26] Simon Nash: same issue applies to policySets [13:29] Simon Nash: what should happen if the interface identified by @Service is not implemented by the Java class? [13:31] Simon Nash: Anish: would like @Service to be inheritable [13:34] Simon Nash: action Anish: raise new issue for inheritability of annotations [13:43] Simon Nash: Anish/Simon: the error statement should not be here, because it contradicts section 8.18 of JavaCAA [13:43] Simon Nash: need to remove operation child [13:48] Simon Nash: action MikeE: raise issue to fully specify interface.java definition in section 3.1 [13:50] Simon Nash: question under discussion is whether the callbackInterface attribute is needed if the interface.java is bidirectional [13:53] Simon Nash: for references, need to include the name attribute from @Reference if it was present [13:57] Simon Nash: multiplicity of references should be 0..1 or 0..n if required=false, and 1..1 or 1..n if required=true (the default) [13:59] Simon Nash: Mark: multiplicity of 1..1 is stated for type is an interface. This could include java.util.Collection, which leads to ambiguity with the next case. [14:01] Simon Nash: need to say how the type is derived for an array or collection (it is somewhere else in the spec, needs a cross0ref) [14:01] Simon Nash: s/cross0ref/cross-ref/ [14:01] Simon Nash: need to add constructor parameter explicitly here [14:05] Simon Nash: can't find the names of parameters unless compiled with debug info.. Sun bug 6444738 [14:05] Mark Combellack: http://bugs.sun.com/view_bug.do?bug_id=6444738 talks about inability to get parameter name using reflection [14:05] Simon Nash: therefore we can't make use of that name in the property and reference algorithm [14:08] Simon Nash: properties.... [14:08] Simon Nash: the rule for value doen't deal with the case of a computed initializer [14:09] Simon Nash: s/doen't/doesn't/ [14:12] Simon Nash: Anish: do we need to introspect the default value, even for a simple literal value [14:18] Simon Nash: consensus is that we will identify these values by introspection (need precise words to say what can be introspected) [14:19] Simon Nash: value attribute is included when we can statically determine its value [14:19] Simon Nash: s/its value/the initializer's value/ [14:22] Simon Nash: for property type, need to mention constructor parameter [14:27] Simon Nash: Simon: can't put a Java type on the type attribute of a property [14:29] Simon Nash: action Simon: figure out how JAXB converts Java types into either XML schema types or global elements [14:30] Simon Nash: or possibly both [14:37] Simon Nash: debate about what is an "unannotated" implementation [14:55] Simon Nash: objections to the terminology used here. The word "unannotated" is confusing given that this case does include annotations. [14:56] Simon Nash: The rules for considering references/properties should be changed. Remove point 2, and change point 3 to public only. [15:02] Simon Nash: Anish is raising an issue to change the default for required on the @Reference annotation. [15:04] Simon Nash: for properties, the comments on properties in the previous section also apply [15:24] Simon Nash: Pradeep: the unannotated algorithm only recognizes interfaces marked with @Remotable. This is not useful, so the unannotated capability is not useful. [15:25] Simon Nash: action Simon: raise issue on brain-damaged defintion of @Service annotation. [15:25] Simon Nash: s/defintion/definition/ [15:26] Simon Nash: specifically: 1. default to Void.class contradicts default to empty array of classes. [15:27] Simon Nash: 2. meaning of Void.class if specified explicitly or by default is not defined. [15:27] Simon Nash: Claim that @Service on its own is meaningless is not useful. [15:27] Simon Nash: s/Claim/3. Claim/ [15:51] Simon Nash: describe case of service introspection in one subsection of 8.1, and the case of reference/property introspection in another subsection. These descriptions are orthogonal. In both of these we say what the default algorithm is for no annotations of that kind. [15:51] Simon Nash: action Mike to produce revised proposal [15:51] Simon Nash: for issue 55 [15:52] Mark Combellack: Break until 4:10 [16:13] Simon Nash: Testing... is a testing subject [16:15] Simon Nash: Testing... is a serious subject [16:16] Simon Nash: Assembly will have tests... Mike suggests Java is a suitable impl type to use for testing, and maybe Tuscany can be used as the testing runtime [16:16] Simon Nash: Other tests will be needed for Java that go beyond assembly, e.g. for scopes [16:17] Simon Nash: need to decide which are the important features that we need to test for sca-j [16:17] Simon Nash: won't have time to do everything [16:19] Jim Marino: I can also describe how we do it in Fabric3 [16:21] Simon Nash: Jim: the Fabric3 tests are components [16:22] Simon Nash: also they have an implementation.junit [16:47] Simon Nash: discussion about how to create a script that runs all the tests... will be proprietary [16:57] Simon Nash: question from Plamen about delegating JEE testing to the JEE subcommittee... nods round the table [17:08] Simon Nash: will be hard to test for errors n deployment [17:09] Simon Nash: s/n/in/ [17:09] Simon Nash: action Dave: will produce a slide recording the proposed approach [17:09] Simon Nash: Mark takes a photo of the whiteboard [17:10] Simon Nash: Mike suggests listing what we need to test in order of priority [17:10] Simon Nash: so that we can start at the top and work down [17:20] Simon Nash: action chairs: raise issue for each spec saying that we need RFC 2119 language [17:24] Simon Nash: action Martin: consider what does compliance mean (to the sca-j specs) [17:26] Simon Nash: actoin Martin: raiase above as a discussion with the liaison committee [17:27] Simon Nash: s/actoin/action/ [17:27] Simon Nash: s/raiase/raise/ [17:28] Simon Nash: issue 87 [17:49] Simon Nash: discussion about whether component type side file is used ot override what's in an annotated implementation or completely override it [17:49] Simon Nash: examples are to add a binding or an intent [17:50] Simon Nash: should it be possible to do that as a delta override without totally replacing the whole definition [17:51] Simon Nash: The JEE spec provides a switch to specify one or the other [17:51] Simon Nash: discussion about next meetng [17:51] Simon Nash: s/meetng/meeting/ [17:52] Simon Nash: suggestion February on West Coast [17:52] Simon Nash: or March 17th in Dublin [17:54] Simon Nash: motion Ashok second Martin thanks to host [17:54] Simon Nash: unanimous acclaim for that [17:54] Simon Nash: meeting is adjourned [17:55] Mark Combellack: Reminder: Call on Monday is Cancelled [17:55] Mark Combellack: Reminder 2: The call afterwards will start 30 minutes earlier and run for 1.5 hours -----Original Message----- From: Mike Edwards [mailto:mike_edwards@uk.ibm.com] Sent: Fri 14/11/2008 13:54 To: OASIS Java Subject: [sca-j] Minutes for Thursday afternoon? Folks, Did the minutes for Thursday afternoon of the F2F meeting get published anywhere please? I'd like to check out the several action items that got tossed my way.... Yours, Mike. Strategist - Emerging Technologies, SCA & SDO. Co Chair OASIS SCA Assembly TC. IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain. Phone & FAX: +44-1962-818014 Mobile: +44-7802-467431 Email: mike_edwards@uk.ibm.com Unless stated otherwise above: IBM United Kingdom Limited - Registered in England and Wales with number 741598. Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]