Feedback and Questions

Thanks for the feedback. I'm still not completely happy with the KenKen solver. I can solve all bar 40% of the diabolicals from KenKen itself (obviously 100% of my own but that doesn't count). It is partly related to Killers where the solver still struggles on a few very hard ones and I get often get sent examples. So when that improves I'll release a 9x9 version.

I don't think it's a bug as such but perhaps it's not informative enough. The solution count is quite separate from the solver and the solver doesn't check the number of solutions before trying to step through. With multiple solutions it usually it 'runs out of known strategies' because there is not completely logical solution. In this case, which is unusual, it takes a logical path and runs it to the end, despite there being multiple solutions.



But your example does inspire me to change the grader to check for solutions since the grade should be meaningless with more than one solution.



On the grade drop - I have found that the number of clues has no relevance to the grade apart from making the puzzle longer - which can accumulate scoring points and boost the grade slightly. It is easy to make 17 clue puzzles that are trivial and 32 clue puzzles that are diabolical, so it it not indicative of anything that adding or removing a clue changes the grade.

I quite agree, it's totally not designed for the iPad. I have built a Sudoku player for iPad which is available here. For the solver we are nearly finished the iPhone version and intend to release an iPad version soon after. More info here.

This is something I'd love to do. Especially with chaining strategies there are usually lots of alternatives. I have had to design a special rule set to pick out the 'best' chain based on a lot of factors - length, type of link components, how many eliminations it makes, etc. It would be more elegant to allow the user to flick through them and choose. The main problem is User Interface and how to present this information/feature without over complicating the page. Definitely on my wish list

Thanks for the grammar check; that change will go in the next release, appreciated.



On multiple solutions - The solver uses logical strategies to take one through the solution path, so a single solution is required. Anything with multiple solutions is not a proper puzzle and will not solve logically - since it is ambiguous what the solution will be. The Solution Count uses a rapid brute force method to calculate the solutions - a very different approach - and can provide all those solutions, however many there are. Some publishers don't know how to check for multiple solutions so they are publishing garbage, to be frank, because a user expects a single solutions. But if you miss off a clue, or put the wrong number in, zero or many solutions are also possible and a logical outcome is not possible.

Yes the Jigsaw solver thought finned x-wings were a valid strategy. Pure coincidence if it worked, but I've taken it out now and the puzzle solves in the way it should (AICs in this case).
Thanks for the alert, I hadn't caught the bug. Glad you are enjoying the puzzles!

The 1 in E6 is green because the 6 in E6 is blue. E6 is a bi-value cell which means it can be a link in a chain. The chain (or net) rather got to the 6 in E6 first and turned it blue. Then it continued to the 1 turning it green. You are not taking into account bi-value cells basically, you are just looking in units (which is valid for Singles Chain, the 2D version of Medusa)