Lists Home |
Date Index |
On Fri, 17 Dec 2004 18:19:40 -0500, John Cowan <email@example.com> wrote:
> Peter Hunsberger scripsit:
> > On Fri, 17 Dec 2004 21:41:29 +0100, J.Pietschmann <firstname.lastname@example.org> wrote:
> > > Peter Hunsberger wrote:
> > > > I need to make this a little clearer: syntax doesn't matter __if
> > > > you're careful__.
> > >
> > > Perhaps in other words: Syntax doesn't matter unless you try
> > > to express semantics by a certain syntax (i.e. syntax implies
> > > semantics). Most of the EDI-formats I've seen attempt the latter.
> > Certainly that's part of the issue: if you commit to a Schema up front
> > then syntax matters since you've started to describe the semantics for
> > a given syntax (element).
> Every syntax implies some semantics, or there would be no point in having
> syntax: we could just concatenate the content-bits in alphabetical order
> and try to guess what it was all about: "alice, bob, charlie, hates,
> loves, peggy".
Sorry if I didn't give enough detail for you to see where I was going
with this (it's a little hectic around here). Perhaps I should have
saidn "then the syntax matters more", and gone into more detail? The
issue is: certain syntaxes carry with them more semantics than others;
if you're modeling you want to make sure the semantics are correct
before you commit to them....
> > The other thing to figure out is how to do generic processing on
> > everything. For example in an XSLT you don't look for element names,
> > you look for other distinguishing attributes (type or such).
> All that means is that you are generic about element names but specific
> about types. Now types are either specified by name (xsi:type), or are
> determined by examining the child elements and attribute values in ways
> that involve names. There's no such thing as generic-processing-all-the-way-down,
> any more than the world can really be held up by turtles all the way down.
Would you argue that RDF can'd be used to describe RDF?
Again, perhaps I should have been clearer: given that many of these
patterns are recursive they more-or-less can be "generic all the way
down" (I don't believe your analogy has any relevance). For parts of
this you don't really need to assume any type up front.