Floats, precision and probability distributions

[David Carlisle]

Mike Richardson (who's in the next office to me, but I didn't know was
   reading this om list:-) writes:


> I would, therefore, suggest that, if a representation of
> reals with a built-in accuracy is added to OM, it should
> be done in a manner which is sufficiently open-ended to
> allow higher moments of the probability also to be
> specified, should that be desired.


In the OM context, being `open-ended' means using an encoding using
symbols from Content Dictionaries rather than fixing a particular
model for these objects into the language itself. As I said when I
started this thread, I think that that is the more natural way to deal
with these problems. We should however at least consider the
extended OMF proposal, even if OMF is extended, the option of encoding
further semantics via some CD will still be there. I don't really
have a good feeling myself whether the extended version of OMF as
specified by Bruce gives enough benefits to phrasebook implementers
that it is worth making an incompatible change at this point.


David