Quick attempt at replicating my barrel-shifter, on freedom.is/svg/barrel-roll.htm
Numerous mysterious contention errors, often varying between runs. At first, I thought I had got my MUXes and DEMUXes confused (they are often implemented as the same circuit, after all), and double-checked those. But that only fixed one error.
Most of those errors, turned out to be from copying the format of my inputs, to serve as outputs: The simulator doesn't like inputs being used as outputs! (This actually is reasonable.) That, and my wiring of the final stage turned out to be wrong (rewired).
I realised my I/O copying error, while editing my simplest barrel shifter, which I forked from this one.
No errors generated anymore, but tests show the circuit isn't working as intended yet (apart from simple shifting), probably due to a copying error: The MUXes & DEMUXes can't be flipped/rotated to work as expected, so I inverted their control inputs for the banked ones, and swapped their outputs for the ones used as logic. This sort of thing tends to complicate transcribing. (I'll have to re-check the copying accuracy another day, as I've run out of time for this now.)
Quick attempt at replicating my simplest barrel-shifter, on freedom.is/svg/roll-left.htm
This is forked from my sophisticated shifter, rather than the other way around... This actually aided the debugging of the more-complex circuit. This one appears to work now: It's always nice to run additional validation.
