Re: Collect Proposed wordings (Was: Can everyone be happy?)

At 3:02 PM -0700 6/22/00, Henrik Frystyk Nielsen wrote:

>I assume you refer to [1] (it took me less than 30 secs to find it so the
>fact that you didn't look it up and read what it said takes away almost
>anything you say). However, I should point out that I can't see any
>conflict between what Larry and I say although I would let Larry speak for
>himself.


No, I was not referring to [1]. I was referring to the message that I 
extracted and quoted literally at the bottom of my mail. Since I find 
my email program easier to search than the web-archive, I tend to 
quote rather than reference. This may be primitive, but it seemed 
obvious to me why there was a long quote of Larry's prose at the end 
of my mail. Perhaps you had best read the quote I was actually 
referring to, before commenting on my mail (or dismissing me out of 
hand for "not looking things up").

Whether or not it is good practice to use URIs as you prefer, it's 
not currently a requirement of the namespace specification, and there 
are good reasons why, that have to do with the need for a 
deterministic comparison algorithm. I won't rehearse these arguments, 
they're adequately represented in the archive.

Other parts of the IETF/W3C in frastructure have adopted the same 
policy for comparison of URIs (allowing the bad practice of making 
htpp://www.w3.org/ to be  treated as differing from 
http://WWW.w3.org/). At least we have Larry's word for this, 
supported by a number of quotations from the relevant RFCs.

>
>My point is that somebody defining a namespace MUST NOT assign different
>semantics to two names that according to the properties of the URI space
>are the same name. This doesn't mean that an application consuming a
>document with namespace identifiers have to now any normalization rules at
>all - and what I think Larry points out is that an application generating
>a document using namespace identifiers MUST NOT rely on those
>normalization rules.

Your MUST NOT is an interesting note as to good practice, but as it's 
entirely unenforceable by code, it's only going to be honored insofar 
as it meets the needs of all the users of namespaces. Currently, they 
don't seem to find the issue very important. It will become important 
exactly when there are interoperable implementations of standards for 
the retrieval of namespace-related information based on the namespace 
URI.

I'd also note that in practice, no-one is using URIs in the way you 
denigrate, because it's confusing and useless to do so.

What would be harmful is to imply that the comparison algorithm 
should take scheme-specific URI equivalence into account: this would 
break the interoperability of namespace references, which only need a 
well-defined comparison algorithm to meet their goals.

Larry's example contradicts your case for the use of a more flexible 
flexible comparison algorithm in another way, by showing that the 
only exceptions to literal comparisons of absolute URIs are limited 
to extremely specific, well-known, and limited contexts. That is to 
say that literal comparison doesn't seem to have been thought a 
serious problem in the definition of the overarching URI framework.

   -- David
-- 
_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
http://cs-people.bu.edu//dgd/             \  Chief Technical Officer
     Graduate Student no more!              \  Dynamic Diagrams
--------------------------------------------\  http://www.dynamicDiagrams.com/
                                              \__________________________

Received on Friday, 23 June 2000 09:39:14 UTC