On Friday, 14 March 2014 at 17:10:53 UTC, John Colvin wrote:
On Friday, 14 March 2014 at 16:38:00 UTC, bearophile wrote:
This comes from a Rosettacode entry:
http://rosettacode.org/wiki/First-class_functions#D
import std.stdio, std.math, std.typetuple, std.functional;
enum static sin = (in real x) pure nothrow => std.math.sin(x),
asin = (in real x) pure nothrow => std.math.asin(x),
cos = (in real x) pure nothrow => std.math.cos(x),
acos = (in real x) pure nothrow => std.math.acos(x),
cube = (in real x) pure nothrow => x ^^ 3,
cbrt = (in real x) /*pure*/ nothrow =>
std.math.cbrt(x);
void main() {
alias dir = TypeTuple!(sin, cos, cube);
alias inv = TypeTuple!(asin, acos, cbrt);
foreach (immutable i, f; dir) {
writefln("%6.3f %6.3f",
compose!(f, inv[i])(0.5),
compose!(dir[i], inv[i])(0.5));
}
}
It prints:
0.500 0.500
0.866 0.500
0.713 0.500
Do you know why there's such difference in the results?
It's a bug, and quite a wierd one at that. See the output from
this: http://dpaste.dzfl.pl/1962d61bb185
It's calling sin every time instead of cos or cube
