[Om] Content-dictionary notations draft

W Naylor wn at cs.bath.ac.uk
Thu Mar 22 14:26:53 CET 2007


On Thu, 22 Mar 2007, Paul Libbrecht wrote:

> There has been a discussion with David Carlisle that we might need to open.
> He indicated a few places where, indeed, the n-ary approach was too limited.
> Including a+-b and  d^2/dxdy f(x).
> This was in comparison to parts of ctop.xsl.
> 
> I concluded that some fine features of ctop will not be reachable by such a
> declarative infrastructure (which yields exchangeability).
> 
> However, I discovered that a simple extension of n-ary by allowing several
> arguments to its symbol in the prototype and parallelize this in the rendering
> could be interpreted as an alternative... that covers most of the cases above.
> 

In my opinion, the major problem that we have here is that OpenMath (by 
design) is extensible. Therefore, I believe that just by tagging extra 
things onto an 'n-ary' element will not be sufficient, unless the extra 
thing allows some extensibility to the presentation side. I believe that 
this could be achieved by the use of something equivalent to my suggestion 
of template functions. Though of course it would never be possible to 
achieve a complete (in the sense of all encompassing) implementation, nor 
(I suggest) would this be desirable!

> As for the quality of the specification, I know it is weak thus far and for
> good reasons... we discuss and add features to it...
> 
> paul
> 
> 
> Le 22 mars 07 à 12:08, W Naylor a écrit :
> > 2/ The second point is to do with nary functions, even-though these are
> > considered I believe that their treatment is oversimplified. What
> > happens,
> > for example, if there are parts of the presentation which have some
> > implicit nature, but also some dependant nature, e.g. the 0 above and
> > below the diagonal of a diagonal matrix, what about the below diagonal
> > elements of a skew matrix, again my solution was using template functions
> > (as indexing functions, and a combination of this and solution 1 would
> > work for the second example). These solutions are described in the second
> > paper, but again not normative.
> 
> 
> 

-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
-
-                  Dr. W.A. Naylor
-
-                  http://www.cs.bath.ac.uk/~wn
-
-                  work tel: +44 1225 386183
-
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*


More information about the Om mailing list