Javascript function rewritten in Java gives different results

Question

There is this Javascript function that I'm trying to rewrite in Java:

function normalizeHash(encondindRound2) {
    if (encondindRound2 < 0) {
        encondindRound2 = (encondindRound2 & 0x7fffffff) + 0x80000000;
    }
    return encondindRound2 % 1E6;
}

My Java adaptation:

public long normalizeHash(long encondindRound2) {
        if (encondindRound2 < 0) {
            encondindRound2 = (((int) encondindRound2) & 0x7fffffff) + 0x80000000;
        }
        return (((int) encondindRound2) % 1_000_000);
    }

When I pass -1954896768, Javascript version returns 70528, while Java returns -896768. I'm not sure why. The difference seems to start inside the if condition: in Javascript function after the if encodingRound2 = 2340070528, while in Java: encodingRound2 = -1954896768.

I made these repls to show it online:

Javascript: https://repl.it/repls/NumbGuiltyHack

Java: https://repl.it/repls/ClumsyQualifiedProblem

EDIT: Changing Java function to this

public long normalizeHash(long encondindRound2) {
        if (encondindRound2 < 0) {
            encondindRound2 = (encondindRound2 & 0x7fffffff) + 0x80000000;
        }
        return (encondindRound2 % 1_000_000);
    }

doesn't seem to affect the result - it's still -896768

Answer 1

In Java, 0x80000000 is outside the range of a 32bit int, so it wraps around to -2147483648.

In JavaScript, 0x80000000 is well inside the range of a 64bit double, so it remains 2147483648.

Obviously, adding -2147483648 vs adding 2147483648 results in a very large discrepancy.

You can either use a long 0x80000000L in Java, or coerce your JS number into a 32bit int with (0x80000000|0), depending on which you want.