On Mon, Dec 16, 2002 at 08:26:25PM +0000, Piers Cawley wrote: > Dave Storrs <[EMAIL PROTECTED]> writes: > > On Mon, Dec 16, 2002 at 06:47:39PM +0000, Piers Cawley wrote: > >> Michael Lazzaro <[EMAIL PROTECTED]> writes:
> I haven't been arguing against his syntax for adding L to R > pipelines, but against the damage he proposes doing to R to L syntax. Fair enough. I'd like to find a way for neither of them to go away, or get damaged. > > However, I think that L2R is valuable enough that it should make it > > into the language, and I don't have a better suggestion. > > Well, L2R is really easy: > > @ary.map({...}).grep(rx/.../).whatever(...); > > For ugly values of 'really easy' of course. Yick. I'll definitely agree on the "ugly" part. However, I'm curious--and I know this has been hashed over, I'm just not clear on where we stand at this point--are you proposing that map, grep, and whatever would be methods on Array? Because that seems unnecessarily restrictive. And yet, having them be methods of Object seems a bit TOO generous. Perhaps the answer is to have an inheritance tree that goes Object | v Collection | +-Array | +-Hash | +-etc (maybe Set, or maybe junction) ...and map, grep, etc, would be elements of Collection, overriden in sensible ways by the derived classes? This is an off-the-cuff idea and I may well be full of it. --Dks