Thanks for the reply. 'Guessing' was a poor choice of words. I was referreing to the try/contradict/backtrack... approach. I appreciate the feedback. It was a good exercise nonetheless.

The medved test apparently requires "guessing".

It still has only one solution, so it's not so much guessing as finding the contradictions. But pure deduction seems to be not enough. Though I feel it should be ( or is ), for appropriate values of deduction.

Hello. May I ask whether your samples and tests require constraint programming (guessing)? I had gotten the impression somewhere that the Katas could be solved solely through logic. My deductive approach has passed all the 4x4's, 6x6's, and the 7's up to the medved tests. Just trying to figure out if I have a bug or have been off-base all along. Thanks.

*************************
*  Manual notifiaction  *
*************************

Please read the message above... x)

Your example has a single solution. I'm not statistically sure, I'm mathematically sure: I generated all possible solutions, and there was one of 'em.

I have added it to tha JavaScript tests; we'll see how many solutions break ( ETA: quite a lot actually, and not just the cheating ones ).

Translators are kindly invited to add it to their language as well.

( Publishing this change actually succeeded. :] This means it may be possible to add random tests as well. )

If anyone has a random generator that will generate single solution puzzles within a reasonable time, I'm open to suggestions! My own one is a bit of a kludge. Otherwise, I may go with ekaruk's 100 pre-generated puzzles.