in reply to Unique Sequence

merlyn prefers a simple MD5-based scheme, which he demonstrates here. I've adopted this technique, and am reasonably happy with it.

(Note: this is very close to the scheme that Ovid posted seconds before above.)