Community
Participate
Working Groups
org.eclipse.test has compile warnings in the build and it is also not setup correctly in CVS: - wrong EE on build path - missing PDE nature and builder
Created attachment 179835 [details] Proposed patch I have also fixed other compiler warnings (deprecated, write access to enclosing field) and also some builder warnings...
Dani, I'm not granted to write changes into org.eclipse.test project :-( Could you please release the patch for me? TIA
Created attachment 179837 [details] Improved fix Frédéric, I think test= (Test)suiteMethod.invoke(null, (Object)new Class[0]); // static method must be test= (Test)suiteMethod.invoke(null, (Object[])new Class[0]); // static method I fixed that and also updated the bundle version.
Strange: I have commit rights to perf and also releng.tools but not to org.eclipse.test. Kim, please commit the patch.
(In reply to comment #3) > Created an attachment (id=179837) [details] > Improved fix > > Frédéric, > > I think > test= (Test)suiteMethod.invoke(null, (Object)new Class[0]); // static method > must be > test= (Test)suiteMethod.invoke(null, (Object[])new Class[0]); // static method > > I fixed that and also updated the bundle version. I do not think it makes any difference for the compiler as this a varargs parameter. But I agree this is a more accurate cast :-)
> I do not think it makes any difference for the compiler as this a varargs > parameter. I think it does: in one case it passes no arguments while in the other it passes 1 argument which is the empty array.
patch released
The compile warnings are gone in N20100929-2000. Hopefully, these changes aren't responsible that all tests DNFed.(In reply to comment #6) > > I do not think it makes any difference for the compiler as this a varargs > > parameter. > I think it does: in one case it passes no arguments while in the other it > passes 1 argument which is the empty array. Frédéric, do you agree?
(In reply to comment #8) >(In reply to comment #6) > > > > I do not think it makes any difference for the compiler as this a varargs > > > parameter. > > I think it does: in one case it passes no arguments while in the other it > > passes 1 argument which is the empty array. > Frédéric, do you agree? No, but I'm not a specialist of 1.5 syntax... IMO, casting either with (Object) or (Object[]) does not change the behavior at runtime. I'm pretty sure that it should be an empty array which is passed as an argument when calling the method in both cases... An empty array of Class can be considered as an array of Object or as an Object. It would just matter for the compiler if it referenced this object afterward, but as it is not the case I still think both cast can be used. Note that my assumption is confirmed by what the warning message is suggesting when the cast is not present... Anyway, as I said, I'm also happy by casting with (Object[]) which sounds more appropriate for an empty array of Class.
> No, but I'm not a specialist of 1.5 syntax... IMO, casting either with (Object) > or (Object[]) does not change the behavior at runtime. Well, it does ;-). Try this: public class Test { /** * @param args */ public static void main(String[] args) { foo("1", new String[0]); foo("2", (Object)new String[0]); foo("3", (Object[])new String[0]); foo("4", new String[] {"1", "2"}); foo("5", (Object)new String[] {"1", "2"}); foo("6", (Object[])new String[] {"1", "2"}); } static void foo(Object s, Object... args) { System.out.print(s + ": "); for (int i = 0; i < args.length; i++) { System.out.print(args[i]); System.out.print(" "); } System.out.println(); } }
(In reply to comment #10) Indeed! Thanks for the tip :-)
The test results are missing because the build scripts refer to a specific version of org.eclipse.test and the version changed in the manifest with the patches for this bug. This is very brittle and I will change this today.
Created attachment 179957 [details] patch patch for build scripts to load org.eclipse.test version from property file during test run
verified that the tests run with the patch in a test build I ran this morning.
