> On Thu, 5 Jul 2007 15:05:07 +0200
> Quentin Garnier <cube@cubidou.net> wrote:
> 
> > What I don't understand here is why you don't want to have three
> > states there:  1.  envsys knows the value, it's on.  2.  envsys knows
> > the value, it's off.  3.  envsys doesn't know the value.
> > 
> > In this scenario N/A is a wholly different value than OFF.
> > 
> > I'm not saying that we can't have the two merged, but then I don't get
> > why the API offers a field for a value _and_ a field for the validity
> > of the said value.
> > 
> > And I'm pretty sure I've expressed that opinion to you before, albeit
> > not on a public list, but a lot before you asked for a final round of
> > comments.
> 
> If I understand it correctly, what you all want is to have the
> following?
> 
> value_cur set => ON
> value_cur not set => OFF
> invalid state = N/A
> 
> Is it what you all want? I want to finish this thread now.

yes, it's what i want.
(and make kernel to report "state=valid, value-cur=0" as it used to be,
if it doesn't currently.)

YAMAMOTO Takashi