While writing some randomized BinaryInteger/up(Shift) tests (#108), it became apparent that a negative data integer index extension would be convenient. DataInt.Body requires in-bounds accesses, but DataInt extends infinitely in the positive direction. It would be simple to extend it in the negative direction as well; all negative indices would return zero since the fractional part is always zero. I believe it would be consistent with the equivalent fixed-point representation, but I haven't thought much about that aspect of it yet.
While writing some randomized
BinaryInteger/up(Shift)tests (#108), it became apparent that a negative data integer index extension would be convenient.DataInt.Bodyrequires in-bounds accesses, butDataIntextends infinitely in the positive direction. It would be simple to extend it in the negative direction as well; all negative indices would return zero since the fractional part is always zero. I believe it would be consistent with the equivalent fixed-point representation, but I haven't thought much about that aspect of it yet.