Dan Sugalski writes: : >Strict, but doesn't really matter. Nobody sane will use anything other : >than $^a and $^b. : : Well.... Are we allowing non-latin characters in identifiers? There : may be potential interesting ramifications with those. Kanji : specifically, though I don't have details for them yet.
Yes, you can use anything with the letter or number property in identifiers, plus you can use ideographs. As it happens, the Kanji for "one" and "two" come in the right order, but don't try to extend that to "three". I am, of course, overstating the case when I say that anyone who uses anything other than $^a and $^b is insane. Any conventional pair will do, as long as they're known to order correctly, and serve as "good" identifiers. That's the only reason we didn't restrict curried parameters to single letters. (Damian can vouch for the fact that I seriously proposed doing so somewhere out in the middle of the Atlantic.) Interestingly, the Kanji for "yin" and "yang" sort in that order, even though the anglicized words don't. So a program or module not intended for circulation beyond Japan could reasonably use those two ideographs for "a" and "b" if it did so consistently. Indeed, it's almost beneficial to use them to represent natural pairs, because people would not then be tempted to extend the sequence to a third item, unless there's something I don't know about that comes after "yang". :-) But anyone reading this should be using $^a and $^b most of the time. Larry
