On Thu, 7 Oct 1999, Robert Elz wrote:

> Rather than negative, would it perhaps make more sense (well, be a little
> easier to comprehend) to just have a dividing line at value 32.
> 
> That is, values < 32 are shift counts (power of two block sizes up to
> 2GB blocks) and values >= 32 are block sizes (block sizes 32 bytes or
> bigger).   This means no non-power-of-2 blocks smaller than 32 bytes,
> but is that a loss?   It also leaves negative values to be able to
> indicate unset, or error, or whatever else might be useful (still allowing
> conditional compilation to drop out the non-power-of-2 handling if
> desired).

The one thing I do like about negative values is that it rolls the
unconfigured value in nicely - it's -1. So all we have to do is have one
test against shift >= 0.

If we go with the dividing line == 32, I'd say that the unconfigured value
should either be 0, or 32. :-)

Take care,

Bill