Comment #0 by matti.niemenmaa+dbugzilla — 2006-09-16T12:31:41Z
void main() {
assert (-1 % 10 == 9); // fails
}
ANSI C leaves modulo undefined for negative values, which is fine, but D doesn't mention anything like that in the spec. If this is intentional, it deserves a mention.
Comment #1 by smjg — 2006-09-16T17:33:43Z
To claim simply that it "doesn't work" is nonsense. Did you actually try experimenting to see what it does?
But see also
http://tinyurl.com/m4rms
Comment #2 by matti.niemenmaa+dbugzilla — 2006-09-17T01:35:50Z
import std.stdio : writefln;
import std.stdio : writefln;
void main() {
writefln(-1 % 10);
writefln(-1 % 5);
writefln(-1 % 3);
writefln(-1 % 2);
writefln(-2 % 8);
writefln(-28 % 7);
writefln(-1 % -1);
writefln(-2 % -2);
writefln(-9 % -2);
writefln(8 % -2);
writefln(16 % -2);
writefln(9 % -2);
writefln(5 % -4);
}
--
Where both values are negative, it gives the correct answer.
Where only the dividend or divisor is negative, it appears to return (remainder-divisor) instead of just the remainder. Following your link, it seems that this is the so-called "trunc-mod", common in many other programming languages.
I'm sorry, I was in a bit of a hurry when I posted this report and couldn't be bothered to take a detailed look at the issue.
Anyhoo, a mention in the spec would be nice.