Vincent BĂ©ron vberon@mecano.gme.usherb.ca writes:
So if we put them in a simple ok test, we'll get failures for some chars, and if we put them in a todo block, then we'll get failures for all the other cases which work fine.
We could get on with a todo_table listing for which chars we fail, but I think (Dmitry can correct me as I'm likely wrong here) these tests are locale-dependant, so we're in for more problems using that method.
The tests should be fixed to not hardcode character values (or maybe only test a-z), we don't use the exact same Unicode tables as Microsoft so the test will never succeed 100%, and this won't be fixed unless we find an app that requires it. We don't want to provide our own Unicode tables just to satisfy the tests.