Lists Home |
Date Index |
On Thursday 26 September 2002 16:35, Mark Baker wrote:
> On Thu, Sep 26, 2002 at 10:25:45AM +0100, Alaric B. Snell wrote:
> > fnp://foo.com:AlaricsIM/getUserDetails(POSTAL)
> That's not an identifier, it's a serialized method invocation. You are
> requiring that the consumer of it know what "getUserDetails" means.
> Lots of Web developers may use HTTP URIs that include what looks like
> a method name, but the key difference from your example is that they
> don't require the user of that URI to know what the method means,
> because they return the goods on a GET.
In what way does the user of fnp://foo.com:AlaricsIM/getUserDetails(POSTAL)
need to know what getUserDetails means? If they issue the call, they get back
a response. You could gateway the protocol onto HTTP by mapping that to
http://foo.com:<some non-80 port>/AlaricsIM/getUserDetails?type=POSTAL and do
a GET to it, and the result would be identical. So where does any knowledge
of the semantics of getUserDetails have to come in, huh?
A city is like a large, complex, rabbit