Question

When generating hashcodes for a class, is it ok to use the hashcodes of that class’s members? Here is a sample class:

class Sample
{
    private readonly string _strA, _strB;
    public Sample(string strA, string strB)
    {
        this._strA = strA;
        this._strB = strB;
    }
    public override int GetHashCode()
    {
        return (this._strA + "###" + this._strB).GetHashCode();
    }
}

I think this will work as long as neither _strA nor _strB contain the string “###”. I’m not totally sure though as I don’t know the specifics of how hashcodes are generated on strings.

I saw a solution in the post at Create a hashcode of two numbers that I could tailor for my purposes, but I think that my solution is more simple (as long as neither string contains “###”).

Answer 1

A better approach would be to combine the hash codes mathematically, using something like the Times 33 hash. In your current code you create a temporary string each time GetHashCode is called, which could suffer from poor performance.

public override int GetHashCode()
{
    // omit null-coalesce if we know them to be non-null
    return (33 * (this._strA ?? "").GetHashCode())
         + (this._strB ?? "").GetHashCode();
}

If your class is truly immutable, calculation of the hashcode up front may be worth the 4-bytes:

private readonly int _hash;

public Sample(string strA, string strB)
{
    this._strA = strA;
    this._strB = strB;
    this._hash = (33 * (this._strA ?? "").GetHashCode())
               + (this._strB ?? "").GetHashCode();
}

public override int GetHashCode()
{
    return this._hash;
}