| Summary: | CDOLegacyWrapper has problem with feature maps | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | [Modeling] EMF | Reporter: | Eike Stepper <stepper> | ||||||||
| Component: | cdo.legacy | Assignee: | Martin Fluegge <martin.fluegge> | ||||||||
| Status: | CLOSED FIXED | QA Contact: | Eike Stepper <stepper> | ||||||||
| Severity: | normal | ||||||||||
| Priority: | P3 | Flags: | stepper:
review+
|
||||||||
| Version: | 4.0 | ||||||||||
| Target Milestone: | --- | ||||||||||
| Hardware: | All | ||||||||||
| OS: | All | ||||||||||
| Whiteboard: | |||||||||||
| Attachments: |
|
||||||||||
|
Description
Eike Stepper
Created attachment 185418 [details]
Fix v1
Martin please check if you agree with this...
With Fix v1 two other legacy tests fail ;-( I'll have a look at it as soon as possible. Could it be that this one is dublicate of bug 302807? At least, it seems, that they are somehow related. Created attachment 185484 [details]
Patch v2
I applied a patch to fix the problem.
The problem was that we have more than one representation for NIL. In fact the difference between CDORevisionData.NIL and EStoreEObjectImpl.NIL caused the problem.
I am not quite sure whether we could join both to one? Or is this separation needed because revision and store a two separate concepts?
Would it make sense to create something like CDOUitl.isNIL which chekcs for the different NIL objects?
(In reply to comment #4) > I am not quite sure whether we could join both to one? Or is this separation > needed because revision and store a two separate concepts? I guess this stems from the time where cdo.common had no EMF dependencies >I guess this stems from the time where cdo.common had no EMF dependencies
This means I should file a bugzilla to join the different NIL values to one object? ;)
You could try to change it with this bug. All right. I'll have a look how conplex the story is and if it is quite easy to solve I'll come up with a new patch :) Created attachment 185495 [details]
Patch v3
I added the changes. All tests are passing and I was even able to optimize the store a bit. O.k. the improvement might not be countable, but it feels good ;)
See whether it is o.k. for you to remove the NIL class.
Committed to HEAD. Available in R20110608-1407 |