🕸Fergus Duniho wrote on Sat, Oct 21, 2017 02:38 PM UTC:
That's now fixed. It initially looked like I could fix it by getting rid of the not before onfile, but that didn't make sense, and my investigation showed that onfile worked correctly. Looking at the arguments used with onfile, I saw it using R instead of P. Since the Rook was in the only file without a Pawn, and it was actually looking for files without the Rook in them, this gave the appearance of onfile giving the opposite result, but that is not what was happening. The value of #piece was being passed to onfile, and at this stage in the subroutine, it was simply whatever it last was when the previous foreach loop finished. Changing this to #pawn fixed it.
That's now fixed. It initially looked like I could fix it by getting rid of the not before onfile, but that didn't make sense, and my investigation showed that onfile worked correctly. Looking at the arguments used with onfile, I saw it using R instead of P. Since the Rook was in the only file without a Pawn, and it was actually looking for files without the Rook in them, this gave the appearance of onfile giving the opposite result, but that is not what was happening. The value of #piece was being passed to onfile, and at this stage in the subroutine, it was simply whatever it last was when the previous foreach loop finished. Changing this to #pawn fixed it.