Hi, Jim.
I'll commit the fix this afternoon.
Thanks,
Raymond
----- Original Message -----
From: "Jim Marino" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, October 02, 2006 12:07 PM
Subject: Re: component hierarchy refactorings and system service support
Hi Raymond,
Thanks. Do you want to go ahead and fix it or leave it to me tonight?
Jim
On Oct 2, 2006, at 11:06 AM, Raymond Feng wrote:
Hi,
From the debugger, I see it tries to resolve an autowire by
navigating the composoite chain which finally reaches the top
composite whose parent is null. I think we should check and retun
null because the target of the autowire is not found (ServletHost
in my case).
Thanks,
Raymond
----- Original Message ----- From: "Raymond Feng"
<[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, October 02, 2006 11:00 AM
Subject: Re: component hierarchy refactorings and system service
support
Hi, Jim.
Your changes seem to be causing some regressions. I got the
following exception in a web srevice sample now.
Thread [main] (Suspended (exception NullPointerException))
DefaultRuntime(CompositeComponentExtension).resolveSystemInstance
(Class<T>) line: 390
CompositeComponentImpl
(CompositeComponentExtension).resolveSystemInstance(Class<T>)line:
390 CompositeComponentImpl
(CompositeComponentExtension).resolveSystemInstance(Class<T>)
line: 390 CompositeComponentImpl
(CompositeComponentExtension).resolveSystemInstance(Class<T>)
line: 390 SystemOutboundAutowire.getTargetService() line: 84
SystemWireObjectFactory.getInstance() line: 40
MethodInjector<T>.inject(T) line: 46 SystemAtomicComponentImpl
(PojoAtomicComponent).createInstance() line: 123
ModuleScopeContainer.eagerInitComponents() line: 144
ModuleScopeContainer.onEvent(Event) line: 72 CompositeComponentImpl
(AbstractSCAObject).publish(Event) line: 94 CompositeComponentImpl
(AbstractCompositeComponent).publish(Event) line: 136
CompositeComponentImpl(AbstractCompositeComponent).start() line:
106 HelloWorldWSClientTestCase(SCATestCase).deployExtension
(CompositeComponent,String, URL) line: 98
HelloWorldWSClientTestCase(SCATestCase).setUp() line: 58
HelloWorldWSClientTestCase.setUp() line: 29
HelloWorldWSClientTestCase(TestCase).runBare() line: 125 TestResult
$1.protect() line: 106 TestResult.runProtected(Test, Protectable)
line: 124 TestResult.run(TestCase) line: 109
HelloWorldWSClientTestCase(TestCase).run(TestResult) line: 118
TestSuite.runTest(Test, TestResult) line: 208 TestSuite.run
(TestResult) line: 203 JUnit3TestReference.run(TestExecution)
line: 128 TestExecution.run(ITestReference[]) line: 38
RemoteTestRunner.runTests(String[], String, TestExecution) line:
460 RemoteTestRunner.runTests(TestExecution) line: 673
RemoteTestRunner.run() line: 386 RemoteTestRunner.main(String[])
line: 196Any clue?Thanks,Raymond----- Original Message -----From:
"Jim Marino" <[EMAIL PROTECTED]>To: <tuscany-
[EMAIL PROTECTED]>Sent: Monday, October 02, 2006 12:10 AMSubject:
component hierarchy refactorings and system service support> Hi,>>
I checked some changes in to support system services scoped to
acomposite hierarchy in the Java SCA runtime. This will allow us
to addsupport for things such as type systems specific to an
application andprogramming model extensions such as OpenJPA and
Hibernate.>> As part of this, the SystemCompositeComponent and
AutowireComponentspecializations have been removed, leaving
CompositeComponentImpl as thecomposite implementation that forms
both the application and system runtimetrees. Previously, each
composite contained one set of children. This hasbeen changed so
that each composite will contain a collection ofapplication and
system children in separate namespaces. Each composite onthe
"application" side of the runtime composite hierarchy may contain
bothapplication and system children. For example, normal user-
providedapplication artifacts and system services such as a type
registry or EJB3entity manager factory. Composites on the
"system" side of the runtimewill only contain system artifacts as
children and will not have anyapplication children (the
collections will be empty).>> The application and system sides of
the CompositeComponentImpl aredistinct. Fro example, autowiring on
the system side will traverse up thesystem side of the runtime
hierarchy while application components may notautowire to system
targets. Since system and application children arekept separate
to avoid collisions, additional APIs have been added
toCompositeComponent for navigating the hierarchy (e.g.
getSystemChild).>> Also, in line with these changes, I removed
AutowireComponent and movedthe methods up to CompositeComponent.
In addition to removing casts, thiswill allow us to add support
for autowiring application components, whichwill hopefully be
included in the SCA specs soon.>> Finally, I moved
SystemAtomicComponent to SPI so that people can createcustom
system extensions. This will allow us to add support for OpenJPA
andHibernate where an EntityManagerFactory and SessionFactory
respectively aremanaged by a custom system atomic component
contained in an applicationcomposite. An injector can be used to
access that system component andinject a JPA EntityManager or
Hibernate Session onto a component instance.Similarly, a type
system helper could be stored in a system component foruse by an
application.>> Jim>>
---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]>
For additional commands, e-mail: [EMAIL PROTECTED]>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
