Floats and compile time ints #12873
Description
Zig Version
0.9.1
Steps to Reproduce
const std = @import("std");
test {
// Case 1: Pass
var i: i32 = 1 / 3;
try std.testing.expectEqual(@as(i32, 0), i);
// Case 2: Fail
var f: f32 = 1 / 3;
try std.testing.expectEqual(@as(f32, 1.0 / 3.0), f);
// Case 3: Compile Error
var one: i32 = 1;
var three: i32 = 3;
_ = one / three;
// Case 4: Compile Error
var f2 = 1.0 / 3;
_ = f2;
}Expected Behavior
I expect case 1 to be a compile error.
I expect either a compile error or f == 1.0 / 3.0 for case 2.
The current behavior allows for errors to be added easily and can it be hard to track down the cause. I was getting a chain of NaN.
Actual Behavior
Case 2
Test [0/1] test ""... expected 3.33333343e-01, found 0.0e+00
Test [1/1] test ""... FAIL (TestExpectedEqual)
.../zig-linux-x86_64-0.9.1/lib/std/testing.zig:79:17: 0x215b61 in std.testing.expectEqual (test)
return error.TestExpectedEqual;
^
.../src/test.zig:8:5: 0x21550b in test "" (test)
try std.testing.expectEqual(@as(f32, 1.0 / 3.0), f);
^
0 passed; 0 skipped; 1 failed.
Case 3
./src/test.zig:12:13: error: division with 'i32' and 'i32': signed integers must use @divTrunc, @divFloor, or @divExact
_ = one / three;
Case 4
./src/test.zig:13:19: error: float value 1.000000 cannot be coerced to type 'comptime_int'
var f2: f32 = 1.0 / 3;