Re: Proposed disposition of Stuart Williams' comments on Metadata in URI 31

On 9/18/06, Schleiff, Marty <marty.schleiff@boeing.com> wrote:
>  While I think this idea might be a
> start, it doesn't go far enough. It doesn't answer the following:
>
> 1) It relies too much on tribal knowledge. How's an application supposed
> to know that "<newSchemeOrganization>.org" is intended to convey
> scheme-type information, while "<otherOrganization>.org" does not convey
> scheme-type information?

I'm not sure what you mean by "scheme-type information", but I'd say
that a wide variety of information can be learned by dereferencing the
identifier.

A problem with assigning semantics to the names themselves, is that
semantics change over time, whereas names shouldn't.

> 4) TAG members frequently justify the use of a single scheme by claiming
> it's expensive to introduce new schemes, and difficult for applications
> to be taught how to process new schemes. I claim it would be just as
> expensive and difficult to teach applications how to recognize the
> various URI characteristics and semantics with a
> "<newSchemeOrganization>.spec" approach, and more expensive with a
> nebulous "<newSchemeOrganization>.org" approach, and even more expensive
> with no approach at all.

I believe that managing it using the data returned from dereferencing
URIs would be less expensive than all of those options.

Cheers,

Mark.

Received on Thursday, 28 September 2006 00:28:08 UTC