well this is great.

Two or three years ago the New Yorker ran an article about digitizing a large tapestry. As I remember the tapestry was laid out on a floor, and a high-resolution scanner was moved over it on a framework of some sort.

This took a while (days or weeks) and the fabric, responding to changes in temperature and moisture, would slightly moved between the times when different sections were digitized. Reconstructing the original appearance of the tapestry in the digitization became quite a problem

This seems to resemble your problem in several aspects.

The article describes how two mathematicians solved the problem.

Though it concentrated more on the human side of the issue than the technical, it still contained a few hints as to how they did it.

I'd suggest reading that article to see what you can glean. At the very least it can provide with some names to use either for a literature search or to contact directly.

A hint is that the tapestry featured a unicorn, and that word was probably in the title of the article.

If you have trouble locating it, try writing me.

"...sends the message that the status quo is the highest goal."

What a non-surprise.

It's the guys who own the status quo that are sending the message.

....are made up on the spot.

Wow... did we just have a reasonable discussion on slashdot? That's creepy. :)

But yeah, modifying production code on the fly was something that took me a while to get used to. I had come from a tradition "IT Environment" in which you had programmers coding in a dev environment, and test environments, and staging, and deployment... etc... And suddenly, I find myself in an environment where you *can't* really have a test environment. How do you test, for example, a mile long conveyor that diverts packages at 100 feet per second accross 250 diverts. Particularly when one of the primary things that goes wrong is that the electricians wire up motors backwards 1/2 the time :)

So yeah, it's been a bizarre transition, but even though I now have to routinely be *at work* by 5 a.m., I love ever second of it.

To be completely honest, I wish I could program (and have my staff program) in Common LISP or some similar language, but the fact of the matter it's that it's hard enough to find dependable programmers, much less dependable programmers willing to be at work at 5 a.m.... and even less dependable programmers willing to be up at 5 a.m. and hack functional paradigm :)