[M3devel] [M3commit] CVS Update: cm3

[hendrik at topoi.pooq.com]

On Tue, Jul 06, 2010 at 03:18:26PM -0400, Tony Hosking wrote:
> I was proposing a 0 value in the common case (where 0 bits are valid for the type).

Yes, I understood that.  I'm pointing out that a nonzero value (also 
where valid) is likely to find more errors than a zero value, and that 
a consistent nonzero value will make it easier to actually find the 
errors, instead of just detect them.

And it's less likely to lead to programs that run successfully only when 
you specify the option.

-- hendrik