[ANSI-Smalltalk] Agenda items...

[Andres Valloud]

Paolo,

However, I do not think 2^1024 is a double.  Is this just a representation
of mantissa issue, as opposed to "what is the actual value in the binary
representation of the double"?  I'd think that the protocol in
Float/Double/Etc should match the binary representation interpretation, but
that's just me.

Thanks,
Andres.

On Dec 2, 2007 5:39 AM, Paolo Bonzini <paolo.bonzini at lu.unisi.ch> wrote:

> > I am not sure that the exponents chosen are putting the mantissa in [1,
> > 2)... why would that be?
> >
> >     Andres Valloud wrote:
> >      > Paolo,
> >      >
> >      > Yes, the implementation here uses 1023 and -1022.
> >
> >     Then you put the mantissa in [1,2), not [0.5,1).
>
> Because, for example, 1 * 2^1023 is a valid IEEE number, but it is not
> valid if expressed with a mantissa in [0.5,1).  If you want to express
> with the right mantissa range, you have to write it as 0.5 * 2^1024
> (i.e. use an exponent of 1024).
>
> I don't have time to check now, but it *might* be that IEEE used a
> mantissa in [1,2).  Smalltalk-98 however uses LIA-1 and I'm sure that
> LIA-1 uses [0.5,1).
>
> Paolo
>
> _______________________________________________
> ANSI-Smalltalk mailing list
> ANSI-Smalltalk at lists.openskills.org
> http://lists.openskills.org/cgi-bin/mailman/listinfo/ansi-smalltalk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.openskills.org/pipermail/ansi-smalltalk/attachments/20071202/fbe02d1e/attachment.html