[interfacekit] Re: BHandler::fToken, int32 or... uint32!??!?

Axel D=F6rfler wrote:

> "Adi Oanca" <e2joseph@xxxxxxxxxx> wrote:
> > You guys were talking about BHandler::fToken taking values that reach
> > 4.294.967.295, but it is just an int32! That means that it's value 
> > range is
> > form -2.147.483.647 to +2.147.483.648!
> > 
> > I ask you that, because by looking in TokenSpace sources I'm not 
> > finding the
> > location where you initialize fTokenCount with -2.147.483.647 !
> 
> 2 billion, 4 billion, who cares=3D3F ;-))
> Seriously, I don't know if we have to reserve the range smaller than 
> zero to express failure codes - if we have to we should stick with 
> int32, if not, we could just switch to uint32 and nobody would see any 
> difference (that's the theory :).

This really doesn't matter. If you have a int32 and continue to 
atomic=5Fadd(&var, 1)
this will work for the full range, of cause, the first 2^31 will be positive, 
and the
other 2^31 will be negative values. But the wrap around to zero will happen
after 2^32.

I don't think tokens need to have any special meaning, they are just a 32 bit
identifier, no matter if you treat it as signed or unsigned (exept if you want 
to
reserve negative values as failure codes, but I don't think they are used in 
this
case).

Marcus

Other related posts: