Dear Juergen,
That fixed the problem. Thanks. However, this scares me a little bit - random basic functionality failing. Perhaps I can suggest the following.
Create a test suite that, for now, just tests a bench of somewhat random stuff that you think are good tests. Then, every time someone reports a problem, such as this one, you can add it to the test suite. This way we'd be sure that there is some coverage going on and that old bugs don't reappear. Perhaps it can be run with "make test".
Please let me know if you think this is a good idea but you don't have time for it. Perhaps I, or someone else on the list, will do it.
Thanks.
Blake