Community
Participate
Working Groups
Build Identifier: 20110615-0604 I am using the 3.1.1.v20110830-1428 OCLinEcore build that should fix the bug https://bugs.eclipse.org/bugs/show_bug.cgi?id=353793. While the build does fix the 353793 bug, it still does not solve all the problems with the imports. Declarations in imported packages are not "visible" in the ecore model that perform the import. Moreover, when saving, closing and re-opening the ecore models that use imports, the imported packages are replaced with the default ecore 'http://www.eclipse.org/emf/2002/Ecore#/', and the referenced types are replaced by ecore::EObject Reproducible: Always Steps to Reproduce: 1. Create two simple ecore models, A and B, and have B import A 2. Create a class in A, say AClass, and one in B, say BClass 3. Have BClass extend AClass 4. Save the B.ecore file and close the editor. 5. Reopen the editor: All the information on A.ecore is gone, and the references to types declared in A are replaced by ecore::EObject
Sorry somehow I missed this one. I don't think it's critical; at most major since the problem is on the 'load' path. Using Indigo SR1, I cannot reproduce the problem on an extends, but I can reproduce the type reference problem: module _'A.ecore' package a : aaa = 'platform:/resource/Activity/A.ecore' { class A; class A2; } ----------------------------------------------------------------- module _'B.ecore' import ecore : 'http://www.eclipse.org/emf/2002/Ecore#/'; import aaa : 'A.ecore#/'; package b : bbb = 'platform:/resource/Activity/B.ecore' { class B extends aaa::A { property a2 : aaa::A2; } } reloads as module _'B.ecore' import ecore : 'http://www.eclipse.org/emf/2002/Ecore#/'; import aaa : 'A.ecore#/'; package b : bbb = 'platform:/resource/Activity/B.ecore' { class B extends aaa::A { property a2 : ecore::EObject[1]; } }
I cannot reproduce this again. It would seem that I got lucky/unlucky with the exact ordering of editorial errors and saves. One possibility is that a save was performed while the source text contained errors, which either were not shown to the user or were not diagnosed so as to inhibit a save. There is certainly no inhibition on saving when errors are shown so deliberately corrupting a reference emulates the repro. Bug 363273 raised for the save hazard.
CLOSED after a year in the RESOLVED state.