Details
-
Type:
Bug
-
Status:
Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: None
-
Fix Version/s: 1.0-beta-1
-
Component/s: None
-
Labels:None
-
Number of attachments :
Description
I've attached a bunch of unit test cases for using RegistrationPicoContainer in a registry style - using a mixture of interface classes or String keys.
There seems to be 2 bugs outstanding right now. The main one is - changing DefaultPicoContainer's getComponents() method to this...
public Object[] getComponents()
{ return componentKeyToInstanceMap.values().toArray(); }which fixes it - though it breaks some other tests.
The other thing that doesn't seem to work is
pico.registerComponent("foo", SomeClass.class):
pico.hasComponent("foo");
pico.getComponent("foo");
So we're close. I hope to provide patches to fix this soon if noone else jumps in first.
Activity
james strachan
made changes -
| Field | Original Value | New Value |
|---|---|---|
| Attachment | patch.txt [ 10465 ] |
james strachan
made changes -
I haven't looked at your tests, but there is a reason why the current getComponents() implementation is like it is:
See the big comment section in DefaultPicoContainer that I just submitted:
I haven't checked thoroughly, but ypur proposed change to a trivial implementation of getComponents() will break the HierarchicalPicoContainer. The present getComponents() code on CVS uses getComponentTypes() method. This method is overridden in HPC, but your more shorthand implementation bypasses it, thereby breaking some tests.
Aslak
Aslak Hellesøy
added a comment - I haven't looked at your tests, but there is a reason why the current getComponents() implementation is like it is:
See the big comment section in DefaultPicoContainer that I just submitted:
http://picocontainer.cvs.codehaus.org/pico/src/java/picocontainer/defaults/DefaultPicoContainer.java.diff?r1=1.24&r2=1.25&cvsroot=picocontainer
I haven't checked thoroughly, but ypur proposed change to a trivial implementation of getComponents() will break the HierarchicalPicoContainer. The present getComponents() code on CVS uses getComponentTypes() method. This method is overridden in HPC, but your more shorthand implementation bypasses it, thereby breaking some tests.
Aslak Thanks for that Aslak.
I'll have a look into it all at tonights GeekNight and see if I can figure out whats going on.
james strachan
added a comment - Thanks for that Aslak.
I'll have a look into it all at tonights GeekNight and see if I can figure out whats going on. Okay, fixed the tests.
The first bug: of course, that
Second problem: Slight misunderstanding, the tests work after you do instantiateComponents():
pico.registerComponent(WilmaImpl.class);
assertFalse(pico.hasComponent(WilmaImpl.class)); // doesn't work yet
pico.instantiateComponents(); // <--- don't forget!!
assertTrue(pico.hasComponent(WilmaImpl.class)); // now it works!
Jon Tirsen
added a comment - Okay, fixed the tests.
The first bug: of course, that
Second problem: Slight misunderstanding, the tests work after you do instantiateComponents():
pico.registerComponent(WilmaImpl.class);
assertFalse(pico.hasComponent(WilmaImpl.class)); // doesn't work yet
pico.instantiateComponents(); // <--- don't forget!!
assertTrue(pico.hasComponent(WilmaImpl.class)); // now it works! Jon Tirsen
made changes -
| Status | Unassigned [ 1 ] | Closed [ 6 ] |
| Resolution | Fixed [ 1 ] | |
| Fix Version/s | 1.0-beta-1 [ 10144 ] |
DOH! Thanks Jon!
james strachan
added a comment - DOH! Thanks Jon!
Here's the test case. doesn't quite pass yet.