Re: [BUG]0/1=?

[KHMan <keinhong@...>]

Rob Kendrick wrote:
> On Fri, 23 Jan 2009 02:25:46 +0800
> KHMan <keinhong@gmail.com> wrote:
>
> > > I'm trying to work out what causes this, and if it's intentional,
> > > why it is desirable.  So far, I've failed.  
> > The specification for strtod in IEEE Std 1003.1, 2004 says, among 
> > other things:
> >
> >    * One of INF or INFINITY, ignoring case
> >
> > But MSVCRT is not entirely POSIX-compatible, so no big deal, we 
> > learn to live with its quirks.
>
> Actually, this isn't a POSIX requirement, it's a ANSI/ISO C requirement.
> (ISO/IEC 9899:1999 7.10.1.3.3).

I guess you can call msvcrt as deprecated, see the reasons why 
MinGW still uses it. There are a number of more recent C runtimes, 
but since the redistributable has a license and it is not 
guaranteed to be in a system, many FLOSS projects for instance 
will stick to msvcrt.

> But surely we're talking about the output of printf() and friends, not
> the input accepted by strtod()?

Sorry for being unclear, in the Unix Spec V3, printf points to 
strtod for the details.

--
Cheers,
khman
KL, MY