Re: differences

Mikito Harakiri wrote:
> I doublechecked diff source file, it's undoubtedly an heuristical
> comparison algorithm, so it can't be considered as a reference
> implementation.

Computation of (one of) the minimal sets of differences is an NP-complete problem. You won't be able to write it in SQL. I have seen some good papers on strategies for getting close without brute-force search though.

I don't think you answered the question anyhow. The way I read your initial question I would have answered "use an outer join on one field and an inequality test on the other", but now with talk of diff I'm not sure that's what you want. Instead of examples, can't you provide a propositional description/definition of what you want?

--
Clifford Heath