The New Sudoku Players' Forum

[PIsaacson]

A RCC only works... If it can be true in both (as in overlapping ALS ...

I agree except in the case where the overlapping cell RCC can be proven to always be false, thus effectively eliminating it from consideration. Base on the haphazard results from incorrectly allowing overlap, I assume that in those cases where it functioned correctly, it was due to the fact that the RCC candidate in the overlap could be safely ignored. But I wouldn't bet the farm on my supposition since the results were so unpredictable. Plus how would you prove that an RCC candidate in an overlapping cell was always false, except by some prior logical step??? In which case the RCC would not exist (have been eliminated hopefully) in the overlap and the standard rules would prevail...

[hobiwan]

I agree except in the case where the overlapping cell RCC can be proven to always be false, thus effectively eliminating it from consideration. Base on the haphazard results from incorrectly allowing overlap, I assume that in those cases where it functioned correctly, it was due to the fact that the RCC candidate in the overlap could be safely ignored.

I think you are correct here: If the candidate in the overlap area is not part of the solution, it should work. But as you said yourself the only way one could be sure about that would be to eliminate the candidate before building the chain...

I still think that RCCs should be forbidden in overlap areas.