Puzzles don't necessarily have a unique solution about sdd-sudoku-teacher HOT 5 CLOSED

The latest fix I made should resolve this issue. The algorithm used to generate puzzles guarantees uniqueness; there was simply a bug in the implementation.

For the record, the algorithm works as follows:
While count is less than 81:

1. On the current puzzle, run all difficulty-appropriate tutor algorithms. If any are successful (meaning a square's value is forced), put that number into the internal puzzle, and increment count. Return to top of loop.
2. If no algorithm succeeds, generate a random number between zero and forty, call it n. If the squares n and 80 - n are unfilled, fill them in in the internal puzzle and the puzzle to be displayed, and increment count. If n != 40, increment count again (as two squares would be filled). Return to top of loop.

The algorithm guarantees a unique solution because every square must either be filled at the start, or be deducible using one of the difficulty-appropriate algorithms (it may not be deducible immediately, but it must eventually be). The bug that was causing the problem was in step 2 of the algorithm, which did not originally check if the square n and 80 - n were filled before incrementing count. Thus, if the same squares were selected, count would be incremented without changes being made to the puzzle, and thus, ambiguity could arise.

from sdd-sudoku-teacher.

After this fix, I hit another puzzle without a unique solution. See screenshot below.

from sdd-sudoku-teacher.

Fixed with revision 1bc2ce9

from sdd-sudoku-teacher.

Not totally fixed yet... This is a moderate puzzle.

from sdd-sudoku-teacher.

I haven't seen this for a while, closing.

from sdd-sudoku-teacher.