add hex literal, binary literal, float literal, hex float literal about zig HOT 2 CLOSED

here's a test for trying out hex float literals in C99

#include <stdio.h>
#include <stdint.h>

int main() {
    float f = 0x1f.0fp-4f;
    uint32_t bits = *(int *)&f;
    for (int i = 31; i >= 0; i--) {
        if (i == 30 || i == 22) printf(" ");
        printf("%d", (bits >> i) & 1);
    }
    printf("\n");
    return 0;
}

$ gcc -std=c99 -Wall -Werror -Wpedantic a.c && ./a.out && rm a.out 
0 01111111 11110000111100000000000

from zig.

[<radix>]<whole_number>[.<fraction>][<exponent_designtator>[<sign>]<exponent>]

radix can be 0x for hex, 0o for octal, or 0b for binary. In hex, the digits a-f are case insensitive.

if either the fraction or exponent are given, then it is a float literal. otherwise, it's an integer literal.

the whole_number and the fraction are interpreted in the radix specified. the exponent is always specified in base 10.

exponent_designator is e or E for base 2, 8, and 10, and p or P for base 16.

sign is + or -.

the significance of the exponent is 10^exp for base 10 floats and 2^exp for base 2, 8, and 16 floats. This means that 0x1p4 is 0x10 not 0x10000.

This is very similar to the C99 syntax. It is more flexible in these ways:

• octal literals start with 0o instead of just 0.
• binary literals exist.
• octal and binary float literals exist.
• the exponent is not required for hex float literals.

I don't know about compliance with the IEEE 754-2008 standard; I can't find a copy of it on the internet.

from zig.