06.12.2010 14:30, Karen Coyle wrote
And the reason for using http://-based URIs is that if they change you do not need to change the data -- you can implement a redirect, just as we do today with URLs. It is a fairly simple machine action that takes place gazillions of times a day on the web, and in the end it will save us a great deal of time.
Of course. And another redirect after the next change and so on. Personally, I'm against unnecessary complication, and this, for my taste, is one. The attitude of "Let machines take care of that" can be just too seductive and slowly lead into ever more interdependence, not to say addiction. Avoid it where possible or at least where it costs nothing. [And a redirect is something the data provider needs to do, you can't do that for your catalog. So you make your catalog dependent on providers' willingness to keep redirects for old URLs indefinitely.] Maybe that's my particular obsession, however. But I happen to think that attitude also gives us unwieldy, bloated metadata where you have difficulty finding some data inside the thorny thicket of tagging. Compromise: Let machines do the work, ok, but think hard where and in what way to involve them. What I was suggesting is not really a different approach: Don't store http://www.something.xyz/abc/IdNumber but just IdNumber and have presentation/service software add the rest according to current fashion. B.Eversberg
