Open eclipse-qvtd-bot opened 2 weeks ago
By Ed Willink on Mar 23, 2017 06:01
(In reply to Ed Willink from comment #0)
It seems the root is still there and gets re-executed.
A few months and fixes later, execution is better. The failure is now that 6 invocations remain blocked at the end of execution.
By Ed Willink on Mar 23, 2017 07:31
(In reply to Ed Willink from comment #1)
The failure is now that 6 invocations remain blocked at the end of execution.
Debugging suggests that the complex/simple attributes speculation has got confused and so some unwanted AttributeToColumn traces are not finished off. Probably fixed by Bug 513375.
By Ed Willink on Aug 28, 2017 04:11
AbstractTestQVT.suppressFailureDiagnosis added so that the two testQVTcCompiler_SimpleUML2RDBMS_CG can run the old way but so that all other transformations have stronger checking.
| --- | --- | | Bugzilla Link | 511028 | | Status | NEW | | Importance | P3 normal | | Reported | Jan 25, 2017 07:52 EDT | | Modified | Aug 28, 2017 04:11 EDT | | Depends on | 513375 | | Reporter | Ed Willink |
Description
If the boolean return from running a test is observed, then testQVTcCompiler_SimpleUML2RDBMS_CG fails with an obscure NoSuchMethodException.
Debugging shows that the transformation runs and that the failure occurs during flushing. It seems the root is still there and gets re-executed.
Re-execution is clearly bad. Perhaps the obscure error is because the object no longer fully exists.