[Om3] binary vs n-ary relations

[Professor James Davenport]

On Thu, 25 Sep 2008, Michael Kohlhase wrote:
> Professor James Davenport wrote:
> > On Wed, September 24, 2008 9:32 pm, Paul Libbrecht wrote:
> >> Let's be pragmatic, how much are we breaking if we claim that strict-
> >> MathML's relation symbols are binary only?
> >>     
> > I would hope nothing, but I worry about the cross-reference to 4.4.10
> > from, e.g. file:///D:/MML3-draft/appendixc.html#cds.eq (my version is
> > fairly old). 'eq' is NOT an associative operator.
> >   
> Well, MML3-draft is just that, a draft and it has language in the
> beginning that anything in it may change at any time, so the reference
> does not bother me.
It bothers me until such point as it changes, but I was merely tring to 
flag up a consequence.
> >> This can certainly be part of the pragmatic to strict translation right?
> >>     
> > This would be an excellent idea.
> >   
> So can we get a consensus here? We have a telcon this afternoon, where
> we can probably decide this. So if I interpret what I have seen, your
> preference would be:
> 
> 1. relation symbols are binary!!!!
Yes.
> 2. we still have pragmatic MathML of the form
>      <apply><eq/>a b c</apply>
>      but that will be translated to
>      <apply>
>        <csymbol cd="logic1">and</csymbol>
>        <apply><csymbol cd="relation1">eq</csmbol>a b</csmbol></apply>
>        <apply><csymbol cd="relation1">eq</csmbol>b c</csmbol></apply>
>      </apply>
>      in the content to strict translation.
and for the other transitive ones (> etc), but NOT for neq.
I don't know what to do about a<b>c etc.: as far as I can tell pragmatic 
did not have them, so we shouldn't add them.
Ditto a>b,c>d, which seems to have no clear semantics.
James