Re: lack of model theory for errors in built-ins

Christian de Sainte Marie <csma@ilog.fr> wrote:
>
> because it is still not clear to me why we cannot just say that 
> RIF-BLD does not provide semantics for ExtTerms outside of their domains 
> of definition...

A layman's explanation:

   If you throw out, say, every 10th word from a story then the story is
   likely to look broken to the reader.

Technical explanation:

   You cannot just do handwaving, as you propose, and call it a semantics.
   You need to go through the motion of providing all the necessary definitions:

     - the definition of semantic structures
     - the definition of satisfaction of formulas by semantic structures
     - the definition of entailment

   Your proposal breaks hard already here. You probably did not realize
   this, but Jos' attempted formalization was a try at turning your
   handwaving into something rigorous.
   
   But even if you were to invent something truly new and the above things
   went through, you have to check the standard results from logic to see
   if your extension is compatible. I estimate this to be an intensive week
   or more with low chances of success.

   Bijan reported that he tried several possibilities, I spent hours, and
   we saw Jos' proposal. None worked.

please, pleaSE, plEASE, PLEASE: let's stop wasting time on this.
I have a LOT of work to do on the document, and we as a group as well.
This topic has become too distracting, and responding to all these
messages consumes too much time.

If one insists on continuing this discussion, please do so ONLY after
you have worked out the proposal in great detail, including the points
that I spelled out above.


	--michael

Received on Monday, 4 February 2008 18:29:22 UTC