Thanks for looking into this. There were two problems. First the TestMethodsAsObjects did not clean up correctly as you noted. Second, TestCase>>#methodChanged: did not expect an object as a method---neither did the PCC tests. I committed the following two changes to the Inbox to fix this: Name: Tests-al.40 Clean up TestObjectsAsMethods (do a proper tearDown) Name: SUnit-al.59 Make edge case where method is an object work. ---- Now we are back to 13 failures and 4 errors. That's already much better. An interesting test failure is [Wide]StringTest>>#testAsInteger '-94' asInteger --> 94. I could trace this back to a change in PositionableStream>>#back by Damien Cassous. Damien, could you have a look? Cheers, Adrian BTW: I'm just the author of #methodChanged: because I had fixed another problem in this method ;) On Feb 9, 2009, at 19:33 , Lukas Renggli wrote:
Many of the errors and failures are from PCCByLiteralsTest. Was this introduced recently? Are these tests supposed to run green?
I already thought that this was because of my ByteArray patch ;-)
However, if you individually run the PCC tests in a clean image they all pass. It is the side-effect of another test, presumably TestObjectsAsMethods that fails and does not clean-up after itself.
Now why does TestObjectsAsMethods fail? It looks like this is related to a change in TestCase>>#methodChanged: that hooks into the change notification system. Looks like you did that change yourself ;-)
Lukas
-- Lukas Renggli http://www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project