Closed agarciadom closed 2 months ago
On second thought, it may not be a good idea to raise a warning in this scenario: an Epsilon script running under the default EPackage registry in an Eclipse IDE that had Sirius installed could potentially raise this warning about multiple paths existing for a given type, without there really being any ambiguity.
It may be best to just reduce the EClass candidates to a set, and only report ambiguity if we really had different EClass instances.
If a module is configured with multiple EMF
EPackage
s, where some of them are subpackages of others, trying to access one of the subpackages' types (e.g. if we list the Sirius description package + one of its subpackages) will produce a warning of this form:While it's true that the reference is somewhat ambiguous due to the multiple paths in which the type can be accessible, there wouldn't actually be a risk of picking the wrong X as all those paths lead to the same exact EClass instance. This is more of a slight misconfiguration from the user, who should have only picked the main
EPackage
that contains the others. It would be better to change the warning to something like: