std.hash.RapidHash incorrectly assumes usize is 64 bits

[cryptocode]

Zig Version

0.15.0-dev.822+dd75e7bcb

Steps to Reproduce and Observed Behavior

Test case:

test "RapidHash" {
    const bytes: []const u8 = "abcdefgh" ** 128;
    const hash = std.hash.RapidHash.hash(0, bytes);
    std.debug.print("sizeof usize is {}, hash is {x}\n", .{ @sizeOf(usize), hash });
}

Build on 32-bit target:

zig build -Dtarget=riscv32-linux test

Compilation error:

std/hash/RapidHash.zig:26:25: error: expected type 'usize', found 'u64'
            b = ((r32(k[d..]) << 32) | r32(k[(e - d)..]));
                        ^
std/hash/RapidHash.zig:26:25: note: unsigned 32-bit int cannot represent all possible unsigned 64-bit values

Expected Behavior

The test case should compile and run

[jedisct1]

Hot take: I don't think RapidHash should be included in the standard library.

The main reason is that its design keeps evolving. See: https://github.com/Nicoshev/rapidhash/releases

It's great to see the design improving, and we can implement each version as it's released. But over time, this will lead to code rot; versions that aren't widely used but would still have to live in the standard library forever and be maintained.

I’d rather see it maintained as an external dependency that applications can opt into. Then, in a few years, if a version proves to be stable and widely adopted, it could be considered for inclusion in the standard library.

[alexrp]

Hot take: I don't think RapidHash should be included in the standard library.

I don't have a strong opinion either way, but this should probably be filed as a separate issue for discussion.

[andrewrk]

a0d1682