Community
Participate
Working Groups
Matching of not unordered containment references should use Set-semantics. Currently the ProximityMatcher looks at the position of an EObject in its containment feature to determine the match. In case of unordered containment features the position should be ignored. At least the feature-filter and measureDifferences should be fixed.
(In reply to Rolf Theunissen from comment #0) > Matching of not unordered containment references should use Set-semantics. > Currently the ProximityMatcher looks at the position of an EObject in its > containment feature to determine the match. In case of unordered containment > features the position should be ignored. > > At least the feature-filter and measureDifferences should be fixed. Hi Rolf, I also have this problem, since the containment references are disordered, it always compare the wrong EObject. Any solution to fix this bug yet? Regards, Fangyi
Sorry, this ticket kinda went under the radar. It sounds like there is indeed a bug in there if we use the positions when checking for diffs in unordered references. We'll have to check the code, but the proximity-based matching is not on our priority list for now. Feel free to look at the issue if you have a reproduction case and some idea of what you expect. We'll gladly look at patches, and we'll be able to provide you with some pointers as to what to change if you have specific questions.