Re: My action item on Moby Dec, issue 14, etc

On Thursday, September 19, 2002, at 08:38  AM, Norman Walsh wrote:
> / was heard to say:
> | When a server first supplies a representation, and marks it cacheable, I
> | think that is creating a contract that affects future accesses to
> | something.
> It's a pretty loose contract, I think. It says, "If you attempt
> another retrieval before the expiry date, it's OK if you get back a
> cached representation. If it's not exactly the same as the
> representation you'd get if there wasn't one in the cache, I don't
> care."

Yes.  It is remarkable how difficult that concept has been to get across
to people used to CPU-based memory caching, where transparency is required.

> Yes, it's useful to give names to things, even if the name alone
> doesn't tell you very much. Consider, for example, "Norman Walsh" (and
> assume for the moment that that is an unambiguous reference to "this
> one" :-). There's almost nothing about that resource that you can
> determine from the fact that you have an identifier for it. I could,
> in principle, change almost everything about myself: place of
> residence, employer, occupation, political and social views, legal
> name, SSN, spouse, even my gender, if I was determined and rich
> enough.
> It's still useful to be able to point to the (currently) tall guy with
> (currently) brown hair and say, "that one's Norm".

Likewise, it is reasonable for you or others to make assertions about
various aspects related to that identifier.  However, just like in real
life, we have to distinguish between assertions about individual
representations of Norman Walsh ("he was wearing shorts") and assertions
about all representations of Norman Walsh ("he always wears shorts").
Also, we have to keep in mind that just because someone has made an
assertion doesn't mean it is a true assertion, or even a testable one.


Received on Thursday, 19 September 2002 17:48:03 UTC