On thinking it over some more, I realize that a=0 is a valid outcome of
the GMA. It represents the case of a single block where in which all of
the ports for each CE are consecutive. The text should note this. I will
make specific text suggestions in a more extensive E-mail once I deal
with the individual issues I see in the draft.
On 24/12/2012 1:57 PM, Tom Taylor wrote:
I hate to raise an old topic, but based on the explanation in the text I
proposed in my previous note, a is the number of bits required to
represent the value 65536/(M*R) - 1. Even if this value comes out to
zero (implying just one block and all ports for a given CE are in a
single consecutive range), surely one bit is required to represent that
value. Thus a=0 may be a possible partitioning of the binary
representation of the port number P, but it falls outside the scope of
the Generalized Modulus Algorithm.
We are left with the choice of discarding the GMA as an explanation for
the partitioning of P (leaving a simple procedural declaration in its
place) or excluding a=0 from the set of possible values of a.
Tom Taylor
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires
