RE: Some comments on 27 June 2003 Web Arch WD

There are a number of aspects of "on the web".  The one that I'm focusing on
is whether the dereference operation requires a representation or not, and
hence the relationship between the retrieved representation and the URI.  In
particular, a HTML FORM POST sends a representation to a URI, and retrieves
a representation from a URI based upon the input representation.  But the
representation that is returned is not directly related to a URI, so that
resource is NOT "on the web".  And that's not a bad thing, just the way it
is.

You're focusing on the dynamic availability of the resource.  I would say
that there is some expectation, and I don't know how to qualify it, that the
resource will be available.  If cnn has an up time of 99.9999% of the time,
we probably consider it "on the web".  If cnn is only available 00.0001%, is
it on the web?  I don't know where the dividing line is.

I'd prefer to avoid the "availability" aspect as I don't think we can say
much.  But we certainly can talk about whether it's even possible for the
resource to be available.  In the case that I mentioned, the form post
result is never ever on the web.

Cheers,
Dave
> -----Original Message-----
> From: Bullard, Claude L (Len) [mailto:clbullar@ingr.com]
> Sent: Friday, July 11, 2003 10:31 AM
> To: 'David Orchard'
> Cc: www-tag@w3.org
> Subject: RE: Some comments on 27 June 2003 Web Arch WD
>
>
> I hope so.  I leave the classification by methods
> in more knowledgeable hands.
>
> It occurs to me that if a method returns an error,
> say 404, the representation is not on the web.  It might
> be on the server, but if it can't be retrieved, it
> isn't on the web.  Does that make sense?
>
> Paring identification and retrieval methods may
> narrow the definition but because the results are
> in some context, observable/provable, that should
> be acceptable.  It would be saying, perhaps, that
> identity alone is not enough to say 'something'
> is on the web.  The queasy feeling returns.
>
> The obvious question:  can a resource be on the
> web if a representation is not?  Can that question
> be answered without dropping into the familiar
> rat hole?
>
> len
>
>
> From: David Orchard [mailto:dorchard@bea.com]
>
> That makes sense.
>
> I want to identify which of the web's means of identification
> and retrieval
> classify a resource as being on the web.  In particular, the
> difference
> between POST and other methods.
>
> > Unless the web is understood to be a system, an architecture
> > makes very little sense or difference.  It shouldn't be that
> > hard to state that a representation of a resource is 'on the web'
> > if the web's means of identification and retrieval can be
> > used and it can be observed that they successfully do retrieve
> > the representation.  The property 'on the web' can only be proved
> > by testing.  It can be defined in terms of the test.
> >
> > len
> >
>

Received on Friday, 11 July 2003 13:57:34 UTC