--- Dan Sugalski <[EMAIL PROTECTED]> wrote: > At 10:52 AM -0800 1/13/03, Austin Hastings wrote: > >--- "Mr. Nobody" <[EMAIL PROTECTED]> wrote: > >> --- Buddha Buck <[EMAIL PROTECTED]> wrote: > >> > Mr. Nobody wrote: > >> > > >> > > > >> > > Unicode operators in the core are a very, very, very, very, > very, > >> very, > >> > very, > >> > > very, very, very, very, very, very bad idea. > >> > > >> > We've already had this discussion. We wouldn't be bringing up > >> using > >> > unicode operators for this function if we hadn't already talked > >> about > >> > unicode operators for other things -- like vector ops. > >> > >> So if we already talked about why they're such a terrible idea, > why > >> are people still proposing them for other things? > > > >Think it through... > > > >Perhaps not everyone feels they're a bad idea...? > > The question, then, is "Does Larry?" (I already do, but that's not > necessarily a showstopper) > > Requiring things outside the ASCII 7-bit range is problematic, as it > requires a Unicode-capable system. That's somewhat troublesome if > you're already dealing with an extended ASCII system that's not > Unicode. (Full Latin-1, Cyrillic, Japanese, Chinese, or Korean > systems, for example, though you can potentially get by)
In this case, that's not even a consideration -- the original proposal was the "ascii7" flavor grin/dragon ops (~> & <~). Only later did we say "Wait, there's a uniglyph that does this in one (very wide) character!". So the real question should be "What kind of upgrade path are we providing for converting these tired old multigraphs into single uniglyphs?" To which I add: What's the right behavior / syntax for a "I want all multimethods (even ones that haven't instantiated yet, a la AUTOLOAD) with this name here: ~>(...) to now be visible as this name here: \u21CD(...) ?" (This is only confusing to me because I don't understand the interrelationships between multimethod dispatch and autoload and type conversion/promotion, etc.) =Austin
