[om] A Proposal for extending OpenMath with structure sharing

[jhd at cs.bath.ac.uk]

On Thu, 4 Apr 2002, Andreas Strotmann wrote:
> Hmm, the way I understood Michael's proposal, he says there is actually no
> need to change the OM standard in the sense of changing the definition of
> OM Objects *even* for OMR (and I agree with him provided we choose
> interpretation B instead, or OMR:=OMRSYN in your suggestion).
But he IS changing the XML encoding of OpenMath objects, which is 
specified in the standard. If that is all, and the actual OpenMath objects 
is merely a compressed tree, that's fine, but he doesn't say that. He 
actually said:
MK>OpenMath standard. The main benefits would be that OM objects can be
MK>encoded as directed acyclic graphs (DAG), as opposed to trees,  

> An OM application that understands Michael's encoding with OMRs then has
> the choice when decoding an OMR whether to just retrieve a pointer to the
No - if the application supports semantic sharing, it HAS to take a copy, 
because in its semantics, the two are different.
> referenced OM Object (making it a DAG, which is licensed by the OM
> standard) or whether to generate a copy of that OM Object (making it a
> tree, which is just as legal).  When it chooses the latter way of handling
> 
> By contrast, OMRSEM requires changes to the OM Object defs because they
> are currently (by design, I think) entirely mute on the question of
> semantic sharing except via variables and bindings.  Only because it does
> keep quiet on this can the current XML OM encoding (without OMRSEM) be a
> valid one in the sense we defined it in the OM Objectives (lossless
> conversion going down and up again must be possible).
Entirely mute, and therefore we did not meet that objective - see the 
discussion I quoted earlier in the thread.
> 
James
--
om at openmath.org  -  general discussion on OpenMath
Post public announcements to om-announce at openmath.org
Automatic list maintenance software at majordomo at openmath.org
Mail om-owner at openmath.org for assistance with any problems