H. G. Muller wrote on Wed, Jun 24, 2020 06:40 AM UTC:
I implemented 'repeated shuffling' in the diagram now, where the shuffle parameter accepts a comma-separated list of shuffles, which are then performed in order. I also added a new shuffle limitation: a piece prefixed with ':' (colon) will not be shuffled to different shades (like a '!' prefix would do), but to symmetric locations. (Which, on boards with an even number of files, should also have opposit shades.) Better make sure to apply it only to an even number of pieces on a set of symmetrically distributed locations!
I think this should allow you to do the shuffling you want, through a trick: When you shuffle (say) Knights and Bishops with :BN specification, it can only leave them as they were, or swap the N and B on both wings. Otherwise the Bishops would no longer be symmetrically positioned.
So what you can do is start the Bishops in the d- and g-file, and (say) A and C on the b- and i-file, and then shuffle A:BC. This will make the B end up either on d/g or b/i, with 50-50 chance. After that you shuffle B and N (which started on c- and h-file), as :BN, to randomize those while preserving symmetry. Finally you shuffle ACQ without limitations. So the specification would be shuffle=A:BC,:BN,ACQ .
I implemented 'repeated shuffling' in the diagram now, where the shuffle parameter accepts a comma-separated list of shuffles, which are then performed in order. I also added a new shuffle limitation: a piece prefixed with ':' (colon) will not be shuffled to different shades (like a '!' prefix would do), but to symmetric locations. (Which, on boards with an even number of files, should also have opposit shades.) Better make sure to apply it only to an even number of pieces on a set of symmetrically distributed locations!
I think this should allow you to do the shuffling you want, through a trick: When you shuffle (say) Knights and Bishops with :BN specification, it can only leave them as they were, or swap the N and B on both wings. Otherwise the Bishops would no longer be symmetrically positioned.
So what you can do is start the Bishops in the d- and g-file, and (say) A and C on the b- and i-file, and then shuffle A:BC. This will make the B end up either on d/g or b/i, with 50-50 chance. After that you shuffle B and N (which started on c- and h-file), as :BN, to randomize those while preserving symmetry. Finally you shuffle ACQ without limitations. So the specification would be shuffle=A:BC,:BN,ACQ .