Re: scripted operators

Perhaps I don't understand this clearly enough.  Bruce says:

> The point is that there is no well-defined (unambiguous) way to
> use precedences, if we simply directly use operators in place of
> terms.

We're currently using localized tests of precedence.  Is Bruce's
remark still true when precedence has been globally tested?  For the
"simple" case (identifiers and operators online with embellishments),
doesn't global testing reduce the token stream to onbaseline
constellations first, then allow the online objects to argue among
themselves?  I may well be missing something.  Or is this just too
simple?  Or are the global tests of precedence something we definitely
don't want to do?

If Bruce or someone else could give me an explicit `odd example' which
goes beyond the local/global problem, this would help me.