[
Lists Home |
Date Index |
Thread Index
]
On Mon, Apr 15, 2002 at 09:45:35AM -0500, Bullard, Claude L (Len) wrote:
> Does item 1 imply that all URIs should use http:
> in the URI string for namespaces?
For existing information spaces that haven't yet been engulfed by
the Web, no. New URI schemes can be defined for that purpose.
e.g. TFTP, SMS.
But for new URIs being created that don't identify existing resources,
IMO yes, the HTTP URI scheme is sufficient.
> Does item 2 imply that only HTTP-defined semantics
> as defined in the Fielding REST architecture
> must be implemented in any other protocol that
> claims to enable web-access?
Depends what you mean by "HTTP-defined". I'd say that any semantic
that is generic to any resource is a semantic that is Web architecture
friendly. If implemented in a protocol, alongside similarly constrained
semantics, then that protocol is Web architecture friendly.
> RDF and the SemWeb concepts are interesting in themselves.
> They aren't currently required in items 1 and 2 above.
> However, they are defined in terms of these, yes? So
> aren't they web-application only? IOW, one only gets
> these if one is working within the web architecture
> per items one and two?
In theory, but not in practice. There are a lot of Semantic Web people
who don't grok REST.
My epiphany moment" about the Semantic Web actually came during a REST
discussion, because it addressed the two most prevalent outstanding
issues with REST that I saw;
1. if HTTP POST doesn't permit me to know what happened to the data I
just sent it, how do I know to use POST in the first place?
Answer; expectation of POST side effects is formed by assertions
communicated through previous GETs, the same way I expect that
dragging a file onto the trash can will result in the deletion of the
file, because I see the trash can icon and as a human know what that
means.
2. how do I prevent my "generic XML/HTTP client" from expanding in size
while trying to process all these different forms of representational
state (i.e. the multitude of schema).
Answer; build a generalized descriptive model that enables resource
state to be asserted, rather than hiding the assertions in schema such
that machines can't extract them without explicit knowledge of those
schema.
MB
--
Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA. mbaker@planetfred.com
http://www.markbaker.ca http://www.planetfred.com
|
