It works - here, the exact source line so the original problem can be considered solved.
You’re the hakyll author right ? Nice to see you here, and thanks for the strict version advice.
While everyone has switched to the performance problem of IORefs vs MVars vs outer thread sync. How would you measure performance of your program ?
The crosswords generator I’m writing is the typical recursive Depth First Search. So:
- Algorithm complexity is of course paramount
- Memoization came next : precompute the maximum amount of material beforehand
- Assuming all this is done , would you just go the usual profiling path ? another advice on performance of a critical loop in general ?