You are correct. The case lambda will accept 1 2 or 3 args. The construction process may not accept 1 arg tomorrow
On Jun 6, 2014, at 2:18 PM, Asumu Takikawa <as...@ccs.neu.edu> wrote: > On 2014-06-06 13:31:17 -0400, Matthias Felleisen wrote: >> I think we can get them in this sense: >> >>>> (arity-includes? (procedure-arity (case-lambda [(x) x][(x y) x][(x y z) >>>> x])) 1) >>> #t >>>> (arity-includes? (procedure-arity (case-lambda [(x) x][(x y) x][(x y z) >>>> x])) 3) >>> #t >>>> (arity-includes? (procedure-arity (case-lambda [(x) x][(x y) x][(x y z) >>>> x])) 4) >>> #f >> >> but I agree that it is of questionable value. -- Matthias > > I'm not sure this predicate, if implemented, can return useful results. > To make this more concrete, consider the following interaction: > > -> (require racket/date) > -> (define base-class% > (class object% > (super-new) > (init x y z))) > -> (define weird-class% > (class base-class% > (if (= (date-week-day (current-date)) 5) > (super-new) > (super-new [x 1])))) > > Today (since it's Friday), the `weird-class%` class will take three init > arguments. If you try this tomorrow, it should take two init arguments. > > Even worse, the class could change the number of init arguments it takes > based on the parity of the millisecond field, a random coin-flip, or by > doing face recognition with your webcam. > > It might be that the arity actually will have changed from the time of > the predicate call to the actual use of the constructor. > > Cheers, > Asumu ____________________ Racket Users list: http://lists.racket-lang.org/users
