This message is from the T13 list server.
This is a potentially useful clarification, copied from the Scsi-over-1394 reflector (aka 1394 Sbp2 normative Annex B) ... x4402 Pat LaVarre [EMAIL PROTECTED] http://members.aol.com/plscsi/ >>> ... 04/12/02 07:03PM >>> > Consider a random access command set, such as RBC. If a disk READ stored the information in initiator memory out of order and didn't transfer all the data requested, what good would it do you to know how many bytes were transferred? This is fun to think about, thank you. This could explain why email threads on this issue so often go down ratholes. When is redundancy worthwhile - rather than a pointless waste of resources - is hard to judge, agreed. Maybe the reason reporting residue redundant with pass/fail of the Cdb helps out in the real world of Scsi-over-whatever is that __Scsi__ Cdb's don't __clearly__ encode how many bytes to copy which way. It's too easy for a host and a device to disagree over block size or whether a header may be copied in part or whether a block descriptor should be present, or what a vendor-specific op means, or ...
