RE: sector alignment for DataOut PDUs?

["Martin, Nick" <Nick.Martin@compaq.com>]

Paul,

Yes, it makes sense to select other values to have the same alignment, and most implementations would do so anyway.  I think all of the currently specified values limiting data size are likely (not required) to be powers of 2 from 2**9 (512) to 2**18 (256K).  I don't think anything breaks if they are not multiples of DataPDUAlignment which is likely (not required) to be a power of 2 from 2**0 (1) to 16 (64K).  If we make that a requirement and an something does not follow it, then what?  Do we want to get into specifying those kind of error cases?  I think this falls under "be conservative in what you send and liberal in what you accept" common sense rules.  If it does not break anything, why enforce it?  The benefits of doing it are clear enough.  

I thought about having DataPDUAlignment just specify the power of 2 (range 0 to 23), but I don't think restricting it in this way is called for.  It could make the length calculations slightly more efficient, but that just doesn't seem very important to me.

We will get into trouble of some of the other limits are less than DataPDULength because the only integer multiple allowed would be 0, potentially causing an infinite sequence of 0 length PDU payloads.  I will study how to correct this.

Thanks,
Nick
-----Original Message-----
From: Paul Koning [mailto:ni1d@arrl.net]
Sent: Friday, March 01, 2002 11:13 AM
To: Martin, Nick
Cc: ips@ece.cmu.edu
Subject: RE: sector alignment for DataOut PDUs?


I like the way you've filled in the details on this.

It would make sense for targets to negotiate burst size limits, and
send requested length in R2T PDUs, that are multiples of the specified
DataPDUAlignment.  Is it necessary to specify that explicitly?  I
think it would make sense to do it, even though the current (rev 10 at
least) rules don't make it necessary).

       paul