Alex Shinn scripsit:

> On the other hand, it just seems surprising enough that w/nocase on one
> case isn't equivalent to the union of all cases.  And that (w/nocase
> upper) != (w/nocase lower).  Enough so that it might be worth making
> an exception for this.

Perhaps then the Right Thing would be to redefine upper and lower to
mean letters which are not only Lu or Ll as the case may be, but are
also part of a casing pair, when in the context of nocase.

On the other hand, PCRE can't do that.