Lists Home |
Date Index |
As long as a representation is returned, you are right.
A representation of a physical singularity might be any
of the things you cite; ie, documents.
A definition of identity where the words "all", "any",
etc. are used will collapse into a singularity.
At that point, we lose access to details. If however,
the Web is defined as an abstract "system", it is
likely to be defined as a set of components
any of which is boundable, but which when
combined, the notion of boundary is described
in terms of that assembly (the system, not
the information it can represent). These boundaries
are useful because the thing named is the
assembly, and the capabilities of that assembly
can be specified and named. Then and only
then does the concept of identity as bound
to location become useful because that property
limits the choice of choices. Basic Shannon.
We can cite abstract definitions of information space,
and even very large information space(s), but
the web is a system that enables us to identify
(and I use the verb deliberately) and access
representations of information items in the
space. That the space is abstract is fine.
It needs to be. One who cannot program to
abstractions should not use XML on the Web.
But ultimately the architecture of the Web system
is concerned with the specifications of the
components and the ways in which these can
be combined to meet a given requirement in
the context of a network: The Internet.
The Web is an abstraction. The Internet
components are not. Information space
is an abstraction; representations or
resources are not. We do not simply
enumerate components; we define a context
of use in accordance with the requirements.
We can usefully say that SOAP/RPC is a Web
system and that information it accesses is on
the Web where it uses these Internet components
in accordance with these requirements (which
Fielding, et al have brilliantly enumerated).
We cannot be as picky about how extensively
that is applied: that is, if a URI identifies
a WSDL, that is on the web. Anything that is
returns as a representation is on the web. If
the implementor chooses to hide information
behind that, it is not on the web and that
is a strictly local and private decision, and
not warrantied by the Web system definitions.
If they use a Web service that hides these,
then as TimBL and Paul have pointed out, the
user is not held accountable for side effects
Beating it out of them just makes the pig mad.
From: Mark Baker [mailto:firstname.lastname@example.org]
On Fri, Apr 26, 2002 at 09:25:20AM -0500, Bullard, Claude L (Len) wrote:
> A black hole.
Could have lots of different representations;
- a picture (xray)
- its location
- its Schwarzschild radius
It obviously can't identify itself, but anybody can identify it for
themselves. For example;
A GET could return any of the above.