Lists Home |
Date Index |
On Tuesday 29 January 2002 11:32 am, Al Snell wrote:
> On Tue, 29 Jan 2002, Gavin Thomas Nicol wrote:
> > The point really is that for data interchange, you can do things.
> > Data formats can be standardized well (GIF et al. as you point
> > out). For information interchange (i.e. data that has "meaning")
> > it's much harder.
> I'm interested in how you define the difference; the meaning of an
> image is a pattern of light, the meaning of a purchase order is a
> request for a purchase.
Yes, I wasn't very precise. I guess I should say that you can easily
define data formats with a static, non-extensible set of components.
A good data format standardizes the extension mechanism (as you note
with PNG) to allow the format to evolve. The use and support of those
extensions causes problems (as you also note). EDI IMPDEFS are a good
case here, as are DTD's in general. DTD's/Schemas evolve over time.
In theory, you can support an evolving set of keywords using
combinations of underlying primitives (XEXPR does this with XML tags,
and FORTH does it with words).