Xara
Xara

Reputation: 9118

HashCode giving negative values

I am converting the incoming string into hash code by doing the following function but some of the values are negative. I don't think hash values should be negative. Please tell me what I am doing wrong.

int combine = (srcadd + dstadd + sourceport + destinationport + protocol).hashCode();
System.out.println(combine);

Upvotes: 45

Views: 79507

Answers (3)

Pravat Panda
Pravat Panda

Reputation: 1090

sometimes the hashcode calculation itself goes beyond the Integer.MAX_VALUE, i.e 2147483647. what happens then is that we get a negative integer after the overflow. Negative hashcode is perfectly valid!

Upvotes: 45

Milky
Milky

Reputation: 391

It is perfectly legal to have negative hash codes, and if you are looking for hash values as used in hash-based collections you can use Math.abs(hash). This can also give you negative numbers when hash is bigger than 2^31, and the best way would be to use a shift mask (key.hashCode() & 0x7fffffff) % M, where M is the table size.

Upvotes: 26

Jon Skeet
Jon Skeet

Reputation: 1503429

I don't think hash values should be negative.

Why not? It's entirely valid to have negative hash codes. Most ways of coming up with a hash code naturally end up with negative values, and anything dealing with them should take account of this. However, I'd consider a different approach to coming up with your hash codes, e.g.

int hash = 17;
hash = hash * 31 + srcadd.hashCode();
hash = hash * 31 + dstadd.hashCode();
hash = hash * 31 + sourceport; // I'm assuming this is an int...
hash = hash * 31 + destinationport; // ditto
hash = hash * 31 + protocol.hashCode();
return hash;

It's not clear what the types of these expressions are, but I'm guessing you're ending up taking the hash code of a string... a string that you don't really need to create in the first place. While there are better approaches for getting hash codes for known domains, the above approach works well as a general-purpose hash generation technique.

Note that it would also help the readability of your code if you avoided abbreviations, and used camel casing, e.g. sourceAddress instead of srcadd.

Upvotes: 59

Related Questions