Kevin Wortman scripsit: > Well, IIRC Alex Shinn has already proposed a procedure-based API and > reiterated it several times. And the only complaint lodged against using a > parameter object seems to be that referencing it costs one procedure call, > which seems perfectly acceptable for an infrequent operation in a > functional-leaning language. The procedure-based API is ideal, except for its lack of backward compatibility. Because a Scheme procedure with a functional argument cannot tell how that functional argument can be safely called, it must call it blindly according to some fixed protocol. That means, for example, that if you have written a hash function that accepts one argument (the dead minimum) and you pass it to a hash table creator that expects to pass three arguments to a hash function, or vice versa, you are hosed. If Scheme implemented arity inspection akin to SRFI 102, this would not be a problem: the hash table constructor could inspect the hash function and safely decide how many arguments to call it with. (A plausible candidate would be Dybvig's counter-proposal as simplified by Van Horn: (procedure-arity proc) returns a exact integer such that the b'th bit of the result is 0 if b arguments are not acceptable, and 1 if b arguments are acceptable. The bits are numbered from 0 in little-endian order, and the arguments are also numbered from 0. So the result 0 means that there is no number of arguments that is acceptable: 1, 2, 4, 8 ... means that exactly 1, 2, 3, 4 ... arguments are acceptable, and -1, -2, -4, -8 means that 0, 1, 2, 3 or more arguments are acceptable.) -- John Cowan http://www.ccil.org/~cowan xxxxxx@ccil.org Monday we watch-a Firefly's house, but he no come out. He wasn't home. Tuesday we go to the ball game, but he fool us. He no show up. Wednesday he go to the ball game, and we fool him. We no show up. Thursday was a double-header. Nobody show up. Friday it rained all day. There was no ball game, so we stayed home and we listened to it on-a the radio. --Chicolini
