Details
-
Type:
Bug
-
Status:
Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: 1.1
-
Fix Version/s: 1.2
-
Component/s: PicoContainer (Java)
-
Labels:None
-
Number of attachments :
Description
The ImplementationHidingPC does not respect the special functionality of the used ComponentAdapterFactory, since it is caching its components in certain cases, that even might break the functionality (e.g. ThreadLocalCAF).
Test case:
public void testUsageOfADifferentComponentAdapterFactory()
{ MutablePicoContainer parent = new DefaultPicoContainer(); ImplementationHidingPicoContainer pico = new ImplementationHidingPicoContainer(new ConstructorInjectionComponentAdapterFactory(), parent); pico.registerComponentImplementation(List.class, ArrayList.class); List list1 = (List) pico.getComponentInstanceOfType(List.class); List list2 = (List) pico.getComponentInstanceOfType(List.class); assertNotNull(list1); assertNotNull(list2); assertFalse(list1 == list2); }Activity
Jörg Schaible
made changes -
| Field | Original Value | New Value |
|---|---|---|
| Assignee | Paul Hammant [ paul ] |
I didn;t add it persoanlly. It was refactored to its current position.
I'll take this and fix as per intent.
However, the reason I /got/ /involved/ /with/ PicoContainer et al was to see a simple modern-day CDI version of Phoenix that was embeddable. Phoenix had impl hiding built in. And component caching.
Pico delivers new things in that it is very usable for object instantiation (with aspects of dependency resolution), not just component management in the server sense that I am questing for.
Paul Hammant
added a comment - I didn;t add it persoanlly. It was refactored to its current position.
I'll take this and fix as per intent.
However, the reason I /got/ /involved/ /with/ PicoContainer et al was to see a simple modern-day CDI version of Phoenix that was embeddable. Phoenix had impl hiding built in. And component caching.
Pico delivers new things in that it is very usable for object instantiation (with aspects of dependency resolution), not just component management in the server sense that I am questing for. Jörg Schaible
made changes -
| Fix Version/s | 1.2 [ 11330 ] | |
| Environment |
Jörg Schaible
made changes -
| Assignee | Paul Hammant [ paul ] | Joerg Schaible [ joehni ] |
IMHO it is not valid for a DelegatingPC to have its own CAF at all. The components have been registered with a special intent and possibly even with a special CA. This must be respected. The current design will break ThreadLocalCA, EJBClientCA, PoolingCA, HotSwappingCA, ...
Jörg Schaible
added a comment - IMHO it is not valid for a DelegatingPC to have its own CAF at all. The components have been registered with a special intent and possibly even with a special CA. This must be respected. The current design will break ThreadLocalCA, EJBClientCA, PoolingCA, HotSwappingCA, ... I see that this test has already been put in IHPCTestCase, and it passes.
Joerg, do you believe that the IHPC cxtor's that do not take a CAF are invalid?
peter royal
added a comment - I see that this test has already been put in IHPCTestCase, and it passes.
Joerg, do you believe that the IHPC cxtor's that do not take a CAF are invalid? After discussion, we may inherit directly from DPC instead of using delegates.
Jörg Schaible
added a comment - After discussion, we may inherit directly from DPC instead of using delegates. I think this is OK now. As in the two instances are not the same.
It still does a delegation rather than inheritance, tho, but the code has been simplified and a bug eliminated.
Paul Hammant
added a comment - I think this is OK now. As in the two instances are not the same.
It still does a delegation rather than inheritance, tho, but the code has been simplified and a bug eliminated. Since the bug has been fixed can we then postpone refactor to future release 1.3 or 2.0?
Mauro Talevi
added a comment - Since the bug has been fixed can we then postpone refactor to future release 1.3 or 2.0?
All done bar inheritance over composition See PICO-256
Paul Hammant
added a comment - All done bar inheritance over composition See PICO-256 Paul Hammant
made changes -
| Status | Open [ 1 ] | Closed [ 6 ] |
| Resolution | Fixed [ 1 ] |
Paul, you know, why you've added the special code ...