> Date: Sat, 19 Jul 1997 00:04:04 -0700 (PDT)
> From: "John A. Kunze" <[log in to unmask]>
>
> I think it's very illuminating to ask the question,
>
> "if we could only define 2 Resource Types, what would they be?"
>
> This really forces the question, "what's the single most important
> Type discrimination needed for resource discovery?"
[I accidentally left out the rest of this message, which follows.]
It seems like we're having some trouble limiting the number of resource
types to 7 or 20 or whatever. What if we approach it from the other side,
and ask what's the smallest number of of types we can get away with?
To me this takes the issue to a more fundamental level and reveals a
likely gap in our planning. I think we've rushed directly into listing
what I would call secondary types (eg, poem, dissertation, essay), over-
looking primary types that are probably more important to users, such as
+ whether the described resource is online (one click away)
+ whether the described resource is interactive (versus "fetchable")
+ whether the described resource is offline (significant work to access)
Some of these types can be deduced from the current crop of URLs, but
they won't be when the internet switches to URNs. Some of these types
might overload the DC Format element (as they currently overload the URL),
but only if the Format element is present and the deducer (eg, browser)
has hardwired knowledge of which formats -- present and future -- belong
to which types.
To summarize, I recommend a split-level Type:
a primary type (one of 3 choices) and
a secondary type (one of N choices, where N is small but extensible).
-John
|