So, Sleepy...

I should have gone to bed hours ago, but there was one 'little' job that I needed to do first. I had been putting off some EBay accounting, and I need to ship stuff tomorrow, so I decided to sit down and sort it out. 4 hours later, I'm finally finished, but YEESH.  I keep thinking that I need to automate the whole dang thing, but I have too many computer projects right now as is.

Anyway, I'm done now, and I can at least sleep with the knowledge that I got a signifigant amount accomplished today.

Trying to turn off my snooze alarm.

So, I is the trying-to-be-awake now. Urph. Its not a working. Quick the coffee IV!

Encrypted Computation.

This post is mainly just a reminder to self. I was in a thread (in a minor way) ages ago in algorithms about the problems of doing encrypted computation. Its a fascinating subject that I currently know far too little about. I'll have to look into it sometime.

Fusion Sort?

Years ago (probably over 10), I heard about a new sort algorithm that was supposed to be faster than quicksort, on the sort of data that quicksort handles best, but only for very large sorts (10s of millions of items or more).

The algorithm was called, IIRC, "Fusion Sort" and was supposed to do three-way tests, with different behavious for less, greater and equal. This gave a marginally better sort for very long runs, and a marginally worse sort for anything smaller. At the time of the announcement, the algorithm had been accepted by some journal or other, and had not yet been published, so they couldn't give any more details.

Other than that one news article, I have never heard another thing about this sort, and searches for "Fusion Sort" turn up nothing. Does this ring any bells with anyone? I would like to know if this algorithm exists, what it is currently called if it does, or was it discredited soon after publication?