Re: reference needed - versioned documents

(Sorry about the delayed response - been busy.)

On Apr 6, 2008, at 10:28 PM, Tim Berners-Lee wrote:
> I can't work out whether you are trying to attack HTTP-Range-14  
> because you don't like it, or really you are confused about how it  
> works and why it is important, or whether you were trolling to stir  
> things up.

The troll is likely to be the last person to know whether s/he is a  
troll, so I'm probably not the best person to ask.

Am I attacking httpRange-14? I like httpRange-14. I do my best to  
follow it. I use 303s and advise others to use them wherever 2XX is  
not clearly wanted. I put the idea under stress so that I can  
understand and explain it better. After considerable research and  
modeling effort I still don't know what an information resource is  
supposed to be; I can identify some instances (maybe even 10^11 of  
them), and I know many things that are not IRs, but I don't feel that  
as I describe information-related things in RDF I have much ability  
to reliably predict what's an IR or what constitutes an acceptable  
awww:representation. If someone can teach me the definition, and how  
applications are substantially benefiting from the exclusion of 2XX  
for URIs of non-IRs, then it will be easier for me to explain the  
2XX=>IR idea to the people I want to convert to the (semantic) web  
(see below). I am making a plea for particulars, whether implemented  
or not.

Am I confused about how it works and why it is important? I  
understand that httpRange-14 says something that was already stated  
less explicitly in RFC2616, and I see how the intent is to require  
people to name things and their descriptions separately, and give a  
way to say that one of them is a page. I would say I am ignorant of  
how it is (or will be) exploited *in practice* and which particular  
*applications* go wrong when it's not respected. I can't say I'm  
aware of any situation, so far, in which it has helped me. But that's  
not a strike against it as it's still quite new.

> There is a huge loss if a system is built which confuses an egg and  
> a page about an egg.    This resolution avoids the confusion, for  
> those who might be confused.

Of course these are different things, and the same URI shouldn't be  
used to name both, but httpRange-14 is only one way to encourage  
separate names and to indicate which URI names which thing.  You  
could, for example, use type assertions to distinguish the egg from  
its page, either in the retrieved content or out of band  (such as in  
a triple store or Link: header). I can see that giving ontological  
meaning to 200s is nice in a way since it attaches positive  
assertions to 10^11 things, and suddenly you seem to know something  
about them that without httpRange-14 you wouldn't have known. Except  
that *I* won't really have a good idea what logical conclusions to  
draw from something being an IR, or under what circumstances servers  
are allowed to respond with 2XX given what they want their URIs to  
denote, or where IR fits in to upper-level ontologies, because I'm  
not clever enough to understand the definition of IR. To you I appear  
to be obsessed with boundary cases; but I feel genuinely confused.

> Perhaps you move in circles less acquainted with the web  
> architecture, and inclined to invent new systems without reference  
> to existing conventions.

I do. I have occasion to talk to publishers, librarians, DBAs, and  
scientists, and to try to win them over to using URIs and web  
standards. Few of them know or care much about web architecture. They  
always think they know how to do it better, and are always  
reinventing wheels. The information they deal with is highly  
balkanized, and I want to change that, and weave their work into the  
web and semantic web. This is why it's important to me to have clear  
stories and high-quality specifications to back them up.

> Wow.  Just so that I gauge the way you use terms, would you say  
> that the statement that in the US the voltage between two flat  
> power pins is 110v alternating current is also a bit of pedantry  
> that provokes thought and helps to influence people to be honest?

If I were to get a shock when I used a 200 where I should have used a  
303, I'd agree that it's not just pedantry. I look forward to  
receiving such a shock!


Received on Monday, 14 April 2008 14:00:58 UTC