RE: Floating points in MIBs

["David T. Perkins" <dperkins@dsperkins.com>]

HI,

No clean solution was found acceptable, so the SMI
is still being held hostage. The only universal
approach is to transfer Floating point values
in a displayable format (such as ASCII) and define
encodings for the special values such as "not a number",
"plus infinity", and "negative infinity".

In addition, the SMI has no
1) unsigned 64-bit integers (that are read-write)
2) signed 64-bit integers
3) discriminated union, or
4) sequence

There are solutions (which have been deployed), none
having enough political capital to result in the
SMI being updated.

So, the current universal solution is to "do the
encoding in ASCII".

On Tue, 24 Jan 2006, Wijnen, Bert (Bert) wrote:
> Brian,
> 
> I don't think we actually have one. I know we discussed it
> several times, but I do not recall we had consensus or
> well-defined TCs for it yet.
> 
> Where do you need them?
> 
> Might be good to float the question on the MIB doctors list
> (I copied that list on the reply)
> 
> Bert
> 
> > -----Original Message-----
> > From: Brian Haberman [mailto:brian@innovationslab.net]
> > Sent: Tuesday, January 24, 2006 17:05
> > To: Bert Wijnen
> > Subject: Floating points in MIBs
> > 
> > 
> > Hi Bert,
> >       Quick question for you.  Is there a standard way of 
> > representing 
> > floating
> > point numbers in a MIB?
> > 
> > Thanks,
> > Brian
> > 
> 
Regards,
/david t. perkins