Based on your description, it sounds like there are significant gains to be made in efficiency and speed by rewriting. I'm interested in a follow-up -- can you give us some simple benchmarks when you are done, such as old size, new size, prior system calls, new system calls, and best of all, performance on a moderate data set?

Not only would such a magnanimous post generate cascading blankets of appreciative warmth from your fellow monks, the excercise might give you some impressive "bonus" information to pass along to the client in question, information that might make you stand out when they cast around for future contracts.

Uh, that of course assumes that the decision-maker at the client shop is not the original author of the script with a fragile ego to boot. ;-)

Matt