Precision and CD's

[Bruce R Miller]

David Carlisle wrote:
> 
> > Are you suggesting the mantissa OMI should be required to be in the same
> > base as the radix?
> 
> The OM Object for integers does not have a concept of the base in which
> the integer was entered (the x10 / 16 distinction is just a feature of
> the xml encoding) so this requirement wouldn't really be possible to
> enforce even if you wanted it.

Yes, exactly what I was saying in the deleted portion: bigfloat sees
only 3 mathematical integers.

> One of the list of jobs to do is to provide, for each of the core CDs,
> translations to presentation MathML (probably expressed in XSL). 

Good.  This is, I think, essential.
XSL (if possible) is particularly good, because browsers (in principle)
will be able to accept and render OM directly.

This
> will provide a standard presentation of every openmath object using the
> core CDs (individual systems may choose a different presentation of
> course). A related activity is to provide, for the CDs in the `MathML
> CDGroup' XSL transformations that specify exactly what is the mapping
> between OM and Content MathML. Mapping <OMS cd="transc1" name="sin"/> to
> <sin/> is fairly obvious but the correspondence for some things,
> especially anything involving bound variables, is more involved and so
> we ought to really have the transformations explicit. So the final
> release of the core openmath CDs (including one for floats, if that's
> the way this goes) should include a specification answering the two
> questions that you ask above. (So only they still need to be answered,
> but only once:-)

I'd be happy with only one answer :>

I'm not insisting on seeing the code, or even the CDs, now, but
_before_ we decide to take the bigfloat CD approach, I'd just like some
assurance that such a bigfloat CAN be converted to presentational 
MathML by XSL.

I'd be quite content with bigfloat as a cd symbol, IFF that CD & symbol
are `sufficiently First Class', ie. in CORE cd set, in MATHML cd set,
AND supported (or supportable) by most applications : converted to
simple float, when appropriate, AND tastefully (if not perfectly)
presented by 
reasonable software.

--
bruce.miller at nist.gov
http://math.nist.gov/~BMiller/