Re: ACTION-446 and Re: Deriving BLD PS from the XML syntax (Was: Re: escaping quotes)

On Tue, 13 May 2008 16:25:15 +0200
Christian de Sainte Marie <csma@ilog.fr> wrote:

> Michael Kifer wrote:
> 
> >>E.g. metadata everywhere does probably not make sense in the 
> >>PS; but why not allow it in the XML? If the only reason is that the
> >>XML is derived from the PS (in the sense of the EBNF), my point is
> >>that it does not have to.
> > 
> > 
> > Several rule systems use metadata at the presentation level.
> > [...]
> 
> Absolutely. And, as you pointed out already, some metadata may impact 
> the semantics in some rule languages (or something that is usually 
> considered metadata; e.g. certainty factors etc).
> 
> But not all the metadata need be at the presentation level (e.g. 
> metadata used by a specific implementation for round-tripping
> purposes comes to mind - consider that this sentence complete my
> action 446).
> 
> My point is that, while everything that is in the PS has to be in the 
> XML, the reverse is not true; 

Maybe. But XML is a superset not only of PS, but also of the *real syntax*.

> and, thus, that it might be better to 
> derive the PS from the XML than the reverse (any "positioning" 
> consideration apart).

That "thus" is a non-sequitur. Inclusion is not the main issue why PS
was introduced.  Since you are talking about PS being 
derived from XML, I gather that you are not talking about getting rid of
the PS (and understand its necessity for the formal part of the document,
which is 90% of it). It is PS that is defined *first* in all cases.
What you are then proposing is to introduce some kind of an ZML syntax for
"something". Then map it to PS, which at this point would be yet another
syntax for "something". And then define the meaning of PS formally.

But wait... XML cannot capture PS (signatures and other restrictions).
So, we start with something that is not a real syntax, but a superset of
it, then translate it to another superset, then define the real syntax in
terms of that superPS, then ...

> And Jos did not convince me yet that I did not demonstrate that it
> was rather easily feasible :-)

To demonstrate you have to convince others. So far you have demonstrated
only that you have managed to convince yourself only :-).


	--michael

Received on Tuesday, 13 May 2008 21:41:55 UTC