eclipse-qvtd-bot
commented
14 hours ago By Ed Willink on Mar 06, 2017 14:03
(In reply to Ed Willink from comment #0)
Unfortunately the *.qvtcas cannot be viewwed to see if it shows a null navigation hazard, since there is some kind of editor load failure on a model leading to too many proxies.
If (Bug 509404) the catch of null navigation errors is removed, testQVTcCompiler_Forward2Reverse_CG fails with a bad navigation from forwardList.headElement which was not in the original QVTc.
The QVTs diagram shows that forwardList.headElement is the first of a three element chain of optional elements.
QVTs2QVTi should synthesizing these using safe navigations since null is a valid value.
The test currently passes since the required null assignment is the default and so the premature exit via an erroneous null source guard doesn't matter so long as the erroneous exit is also (erroneously) ignored.
| --- | --- | | Bugzilla Link | 511026 | | Status | NEW | | Importance | P3 normal | | Reported | Jan 25, 2017 07:27 EDT | | Modified | Mar 06, 2017 14:03 EDT | | Reporter | Ed Willink |
Description
If (Bug 509404) the catch of null navigation errors is removed, testQVTcCompiler_Forward2Reverse_CG fails with a bad navigation from forwardList.headElement which was not in the original QVTc.
Unfortunately the *.qvtcas cannot be viewwed to see if it shows a null navigation hazard, since there is some kind of editor load failure on a model leading to too many proxies.