Looks like using ;:, 5:!1 and the class_trace_ function (trace.ijs) could be used in a not too complicated way to find out how all the components of a sentence evaluate to conj adv dyad monad leftnoun and rightnoun, except that's not quite enough for a preprocessor.
I'd also need to know whether specific conjunctions/adverbs evaluate to a verb or noun (after their arguments processed). I'm guessing there's only 2 ways to find out: 1. execute the conjunction (not safe obviously), or 2. Would need J support for classifying user-defined conjunctions by their possible return types. The compiler doesnt need to know for conjunction c in (m c n) f y whether it will result in a verb or noun because it evaluates those before the rest of the sentence. But is there a way to get that info out of the system? Or has anyone made a code analysis routine to guess that result in most cases? --- "Miller, Raul D" <[EMAIL PROTECTED]> wrote: > p j wrote: > > Regarding a preprocessor, could it still be > practical > > for private use, even if no one else understood > the > > benefits of the syntax, and was ruled out of being > > incorporated into the language? > > This depends. > > For a proposal as complex and pervasive as yours, a > preprocessor is certainly a worthy endeavor. A > preprocessor would eliminate ambiguity in your spec, > and would allow serious testing of the idea. > > Note also that you could use a modified version of > the define adjective which incorporates the > preprocessor. > Since this would mess up some uses of noun define, > you > might instead name it Define. > > -- > Raul > ---------------------------------------------------------------------- > For information about J forums see > http://www.jsoftware.com/forums.htm > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
