[Om3] Binding Integral symbols (was [Fwd: MathML CDs])

[James Davenport]

On Sun, 8 Jul 2007, Michael Kohlhase wrote:

> James, David,
>
> > > I'm not sure why we can't keep the OM symbols for integration  as
> > > they
> > >
> > I was certainly not proposing any CHANGES to the OM symbols, unless
any
> > missing FMPs etc crop up in the process. I was proposing to add ones.
> >
> I think you are totally right here, I also do not want to change them.
OK - no change to calculus1.
> > > which doesn't bind variables in the mapping between this "legacy"
> > > syntax
> > > and the canonical syntax.
> > >
> > I could easily add 'binder' version of the MathML definite integrator
> > to the
> > CD I circulated earlier. Indeed, it could be argued that OpenMath's
> > definit 'ought' to be a binder anyway.
> > What worries me is Michael's claim that MathML will have an indefinite
> > integration which is a binder: this I do not understand.
> >
> Sorry for the confusion, I do not really want a binding indefinite
integral
> operator. I had been concentrating on the mechanism and not on the
> specific instance.
Excellent: then my int in calcmml1 solves the 'indefinite' case, except
for the bvar issue. Since this IS NOT actually a bound variable, does
MathML want to keep the name? Anyway, we can add a rule/FMP to calcmml1: I
might do that later tonight.
> > One of the other problems is that Math2.0 talks of "the
> > uplimit,lowlimit
> > pair", and makes no mention of what happens if there is only one. The
3.0
> > draft says that they default to \pm \infty. I had always assumed that
> > they would default to the variable of integration.
> >
> I think that we should not have a default, let's make sure that both are
> specified. I think that we can do that, and handle any defaulting
mechanism in
> MathML via the legacy conversion.
I would rather that 3.0 retained the 2.0 wording, and the converter only
handles (as at present in my suggestion)  the case where both are
specified.
James