Lists Home |
Date Index |
firstname.lastname@example.org (Tim Bray) writes:
>> I think the main thrust of this thread has that URIs can effectively
>> identify multiple "resources", whatever those are, depending on
>> context. "The resource" as a monolithic being disappears.
>I might agree if you drop the word "effectively".
Sure thing. I'm just surprised to find myself saying things like "URIs
can identify multiple resources," though at this point I'm happy about
>I think the whole
>system, both the everyday Web and the Semantic Web, has a design
>assumption that a URI identifies something. When this is not the case
>(stupid publishers, DNS ownership changes, etc etc etc), it is damaging
>to both Webs. However, it does happen. It should be discouraged and
>avoided, but it does happen and both humans and software have to deal
>with it when it does.
After spending what I feel are wasted years trying to argue that such
things should be "discouraged and avoided", at least in part by
requiring developers to pay attention to how RFC 2396 describes the
functionality of URIs, I'm sure that I'm willing to fight the tide any
"Cool URIs don't change" is a bad joke, even at the W3C (which tries far
harder than anyone else I've seen), and pretending that XML Namespace
URIs identify the same thing as the same URI used in an HTML href just
doesn't make sense after a while, even given the valuable work put into
things like RDDL. RDF has a similar set of differences.
>> That lets us give up on the notion of consistent (URI+time->result)
>> expectations that Jeff seems to be picking on, and get on with our
>I think it's highly reasonable to demand that all the different
>representations of a resource be at some level *consistent*. I think
>the word "equivalent" has been justly criticized, but inconsistency
>seems like a real problem to me. Unlike some, I'm not trying to claim
>it doesn't happen or that we can make it go away by writing specs.
Inconsistency is real, yes. Whether it is a problem or not depends on
how you view it. I used to expect consistency of URIs, as all the
various contending URI orthodoxies at least agreed on that. At this,
point, though, it seems wiser to accept chaotic reality over a set of
abstractions that make me long for more training in Platonic
The specs would do better to acknowledge inconsistency than to try to
hide the hodge-podge of URI practices under a single set of claims.
That seems to avoid most of the problems that nearly every URI
discussion in the history of the concept has faced.
I may ask for consistency - it's convenient for both people and
computers - but at this point "demanding" consistency seems laughable.
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org