[Anpheus]

Edit: I changed this page for arithmatic based modulo. Should be a bit faster. And I must note that as with all floating point operations, accuracy may vary.

/* forbidden operations */
forward operator%(Float:oper1, Float:oper2);
forward operator%(Float:oper1, oper2);
forward operator%(oper1, Float:oper2);

This is incorrect... Assuming you only want to return unsigned values in the range [0, oper2)...

stock Float:operator%(Float:num, Float:modulo)
    return num - modulo * floatround(num / modulo);
stock Float:operator%(Float:num, modulo)
    return num % float(modulo);
stock Float:operator%(num, Float:modulo)
    return float(num) % modulo;

Now, if you want 'simple' support for signed values, use this.

To get a signed byte, a range of [-128 to 12*: a % -256.0

stock Float:operator%(Float:num, Float:modulo)
{
    new Float:x = num / modulo;
    if (modulo < 0) {   //Act as though it's a wrapping value (signed)
        return (x - floatround(x+0.5,floatround_floor)) * modulo
    } //else:
    return num - modulo * floatround(-x,floatround_floor);
}
stock Float:operator%(Float:num, modulo)
    return num % float(modulo);
stock Float:operator%(num, Float:modulo)
    return float(num) % modulo;

*: This is using mathematical notation for ranges, as in:
[a, b] denotes that the value can be anywhere between a and b, and also equal to a or b.
(a, b) denotes that the value can only be between a and b, not equal to either.
Thus, [-128, 12 means the value can be equal to -128, and anywhere between -128 and 128. It cannot be equal to 128, however. Also, floatround(127.5 % 256) would be equivalent to floatround(127.5) % 128.

[Anpheus]

No matter what I do, I can't get AMX to compile a plugin with mixed-tag modulo operations without crashing.

Is this something coded in the compiler?

Oh, here's an example:

#include <amxmodx>
#include <engine>

public plugin_init() {
    new Float:test = 12345.67890 % 0.123456789;
    new Float:test_two = test % test;
    client_print(0, print_chat, "%f^n", test);
    client_print(0, print_chat, "%f^n", test_two);
    return PLUGIN_HANDLED;
}

That compiles... this does not:

#include <amxmodx>
#include <engine>

public plugin_init() {
    new Float:test = 12345.67890 % 0.123456789;
    new Float:test_two = test % test;
    new Float:test_three = test % floatround(test);
//  new Float:test_four = floatround(test) % test;
    client_print(0, print_chat, "%f^n", test);
    client_print(0, print_chat, "%f^n", test_two);
    client_print(0, print_chat, "%f^n", test_three);
//  client_print(0, print_chat, "%f^n", test_four);
    return PLUGIN_HANDLED;
}

Or...

#include <amxmodx>
#include <engine>

public plugin_init() {
    new Float:test = 12345.67890 % 0.123456789;
    new Float:test_two = test % test;
//  new Float:test_three = test % floatround(test);
    new Float:test_four = floatround(test) % test;
    client_print(0, print_chat, "%f^n", test);
    client_print(0, print_chat, "%f^n", test_two);
//  client_print(0, print_chat, "%f^n", test_three);
    client_print(0, print_chat, "%f^n", test_four);
    return PLUGIN_HANDLED;
}

[Johnny got his gun]

You should never do modulus operations using decimals.