* chromatic <chromatic@wgz.org> [2009-12-28 19:45]: > ... and calling can() as a function, not a method is still > a hack that obscures the intent of the code. Agreed. I just had to write several paragraphs to John because of this. I’m not promoting this as a good way, merely the best available way. > I'd rather see a way to test that a given non-reference scalar > is a valid class name without relying on side effects of > implementation. It’s OK to use them as an implementation detail in an intention- revealing routine, which I’ve been arguing we should have. But it sucks that application code currently has to use such side effects directly, or at least has to contain wrappers, reinvented thousands of times over. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>Thread Previous | Thread Next