Lists Home |
Date Index |
Simon St.Laurent <email@example.com> writes:
> Peter.Hunsberger@stjude.org (Hunsberger, Peter) writes:
> >However, it seems to me that like it or not, you've already
> >to standardized semantics just by defining syntax...?
> That strikes me as a completely perverse interpretation.
Well, I only mean "standardized" to the extent that you share a common
understanding with whomever you defining the syntax with....
> My general rule at this point is that standardization should
> be performed as little as possible, and only in cases where
> there is clear immediate benefit.
Since discussions on this list can't help but invoke visions of W3C and
OASIS et al, I understand your reluctance. My point, is that the only
debate is about what degree of standardization is required.
> >Well, there's the issue; locally you have to standardize at least to
> >the extent that you and your direct partners agree on the semantics.
> >However, with the Web it's never just local: when you
> exchange XML with
> >someone you are exchanging semantics with everyone your partner has
> >ever exchanged semantics with...
> That doesn't strike me as cause to set up an Academie
> Francaise for XML vocabularies. Communications is always
> about negotiation and sorting out who will take what in what
> form. Computers have never been particularly good at such
> negotiation, but it may be fair to suggest that continuing
> human involvement in such things is a better idea than
> standardizing and automating as much of it away as possible.
Yes, but the fact that in the past computers have not been good at this does
not mean that people shouldn't be playing with finding someway of making
computers good at this! Once upon a time computers weren't very good at
understanding anything not written in machine language...
> I don't believe the Web changes that.
Again, it's only a matter of degree: we've moved from private networks to
public networks. Computerized communications no longer happen in isolation;
locally implied semantics are no longer sufficient for robust