Per Bothner writes:
I just noticed one unresolvable incompatibility with R6RS:
(bytevector-copy! bv1 i bv2 j k)
In R6RS this copies bv1[i..i+k-1] into bv2[j..j+k-1]. In the draft, it copies bv2[j..k-1] into bv1[i..i+(k-j)-1]. Ouch. OTOH, I think R6RS got this one wrong: Mutable operations in Scheme have the target ("left-hand") value first, and the source ("right-hand") value last.