I haven't yet tried any of those. When it didn't support any of \w, \W, \s, \S, and \b, I just basically gave up on using character classes and wrote things out as character sets.
I'd also like to have a negative match (i.e. give me everything that matches X, but that does not match Y). That sort of construct can be useful to remove fp results, and can be more efficient than using negative look-ahead/behind.