Thanks for some interesting insight.

I'm not insulted- merely impressed that you figured out I haven't systematically measured a thing, except by staring at my watch. Hey, I inherited this system!

You are correct also in that loading this hash is a pig: but that happens only once, generally when users are not waiting, so I'm not greatly concerned about it. However, then various files are read and checked against the hash, and that would be nice to speed up.

I am intrigued by your statement that I will incur no lookup penalty whether or not hash collisions occur. That's contrary to my understanding, and I suppose that is something I should understand.