order of the input impacts the generated layout - why ?

[yonatanm]

when I change the order of the objects {answer: 'xxx', clue: 'yyy} within the input array to the main generateLayout() function I get a different layout.

considering the fitness function - could be that a different order of the input will yield a better layout ?

[MichaelWehar]

Thanks so much for your message!

I just tested it out. Yes, order does appear to matter!

Here are a few ideas I quickly came up with that might be a little bit helpful:

• We could randomize the order of the words first if you want different outcomes each time
• We could sort the words alphabetically first if you want the same outcome each time
• Or, we could make a new fitness function that doesn't care about order

I think that having many different fitness functions that one could choose from would be really neat! Maybe this is something to be done in the future.

[yonatanm]

Thanks for your suggestion, here is my POV

Random ? I can do that on the app level
Sort? Yes, but I won't get the best possible crossword (as determined by your fitness function)
The third option is best. I can do it at the app level, and run all the permutations of order (n!). That way I can either use my owner fitness function or (preferred) have generateLayout() return its rank.

BTW, I'm using it in a side project. I run generateLayout() for all permutations I can imagine (explained below) and then I use my own fitness function to pick the best.

If I permute each SAPCE in an answer, I can decide where to delete it (and make it a long word without spaces) or to split it to a "sub" clue-answer pair. which gives me 2^[number-of-spaces] permutations (which is not ideal)

for example:
[{answer: "hello world", clue: "whatever"}] --->
p1:
[{answer: "helloworld", clue: "whatever", originalAnswer: "hello world"}]
p2:

 [{answer: "hello", clue: "whatever", originalAnswer: "hello world"}, 
  {answer: "world", clue: "whatever", originalAnswer: "hello world"}]