This could never work as-is, since the decoding algorithm as implemented
in SPD has the property that after decoding, port1 > port0 always. If
port0 > port1 initially, then the decoder will compute the final port0
as 63 - port0 and port1 as 63 - port1, making the final port1 larger
than port0. The new encoding algorithm here is the one that I had in
mind initially after RE'ing this.