Sudoku Solver in Apt (note: this is pretty old you've probably all seen it I just felt it's cool and relevant)

http://algebraicthunk.net/~dburrows/blog/entry/package-manag...

This is really cool.

I thought one of the design decisions of SQL was that it would NOT be Turing complete, so hacks like this could not exist. It's funny that the nature of a database makes things like this possible. Can a turing machine be made only using SQL queries? Could a higher-level language compile to it?

In my humble opinion, there are so many proprietary and / or nonstandard extensions to SQL that it's turned into a right mess. I think that the language itself is a bit kludgey and it'd be much easier if it'd just correspond to relational algebra directly. (Eg if you do a select * from foo where x = y, you're actually doing two operations, a select and a filter, and the natural language like syntax of the language makes it very ambiguous and confusing when you're dealing with complicated structures like nested joins and such)

Nice. It's not nearly as awesome, but here's my Sudoku solver in Clojure:

http://github.com/vishsingh/lisp-toronto/blob/80ea008e80976d...

Of course this is just a silly hack so I can't be too hard on style here, but it would have been worth it to make a table for the input/output. An argument list of 81 values is painful to see.

I've been reading about that too:

http://www.celko.com/puzzles.htm (first part) vs http://sqlblog.com/blogs/hugo_kornelis/archive/2007/10/04/ho... (Celko replies in the comments)

My Postscript one.. http://chris-lamb.co.uk/2007/06/10/sudoku-solver-in-postscri...