Re: sector alignment for DataOut PDUs?

[Paul Koning <ni1d@arrl.net>]

>>>>> "Sajay" == Sajay Selvaraj <sselvara@npd.hcltech.com> writes:

 Sajay> Hi, Now both the options look valid. How about mandating both
 Sajay> of them ?

 Sajay> - target should accept immediate data which is lesser than the
 Sajay> negotiated value of unsolicited data length - initiator should
 Sajay> not send "less than the negotiated unsolicited data length or
 Sajay> the total I/O length, whichever is smaller".

You could do that. 

Unfortunately, as the spec points out, flexibility comes at a cost.
If you want to make the receiver as simple and efficient as possible,
the best answer is to take away options from the sender.  Is it
important for the sender to be able to send short data?  It's not
clear to me why it would be.  

So if it isn't a problem to require the sender to send the full amount
when there is enough total data to do so, then this allows the
receiver to rely on that and be simpler and possibly more efficient.

The current spec points in that direction; the issue is that it
doesn't state it strongly enough to ensure interoperability.  My
proposal #1 keeps the intent as it was but makes the requirement
stronger so you do get interoperability.

	 paul