[
Lists Home |
Date Index |
Thread Index
]
ktl@ktlim.com wrote:
> As I see it, we have these camps (leaving aside the possibly-important
> issue of fragids in URIrefs):
>
> 1) If you do an HTTP GET on an http: URI, not only may you get nothing
> because something is broken, but you may get nothing because the authority
> creating the URI purposely does not make anything available. The URI is
> just a string that may identify anything, document/not, concrete/abstract.
> The authority creating the URI decides what that thing is.
>
> 2) If you do an HTTP GET on an http: URI, you should expect to get something
> back (though you may not because something is broken). What the URI
> identifies is exactly the thing you get back, or a slight abstraction of
> that to account for conneg and modification over time.
>
> 3) If you do an HTTP GET on an http: URI, you should expect to get something
> back (though you may not because something is broken). The thing you get
> back is just a representation of the actual thing that the URI identifies;
> what that actual thing is is decided by the authority creating the URI.
Roy Fielding's been writing good stuff on this over in www-dev. The
correct answer is half 1, half 3:
If you do an HTTP GET on an http: URI, not only may you get nothing
because something is broken, but you may get nothing because the
authority creating the URI purposely does not make anything available.
The thing you get back (if any) is a representation of the resource;
what that resource is is decided by the authority creating the URI. -Tim
|
