Re: Relative URI or relative URI reference

At 11:50 PM -0400 8/18/04, John Cowan wrote:


>Do you think that stone lions, like the ones outside the New York
>Public Library, are lions?  They certainly don't belong to the
>species _Panthera leo_.

Well, now I suppose we're going to start arguing about Platonic 
forms, but yes; I think the stone lions do participate to some extent 
in the nature of "lion-ness".

Of course, this occurs in the fuzzy world of human language where 
context is critical for sorting these things out so it's a bit of a 
canard. In the domain of technical specifications we're operating in 
here, it's possible to be much more precise, unambiguous, and clear. 
Indeed it is critical to do so. The current draft fails to achieve 
this. It is unclear, ambiguous, and imprecise. It cannot be properly 
understood without being familiar with the discussions that went into 
it, and the intentions of its authors.

I am in the interesting position of straddling the line between the 
precise technical world of spec writing and the more imprecise world 
of human language. I often find myself trying to decide (and explain 
to copy editors) exactly when one should say "URI", "URL", "URL 
reference", "URI reference", "absolute URL", "absolute URI", etc. and 
explain the distinction to readers.

It is simply confusing to readers to say that "absolute URI" is a 
synonym for "URI" and that a "relative URI" is not in fact a "URI". 
It is very difficult to write about this in a precise fashion, and in 
fact when it comes to helping readers to get practical work done 
myself and at least one other major author/editor in the XML space 
have come to the conclusion it's best to ignore URIs completely and 
simply talk about absolute and relative URLs, maybe with a footnote 
on URIs noting that they're a theoretical construct unlikely to be 
encountered in the real world readers are operating inn.

When it comes to actually teaching developers and content creators 
how to use URLs for web site development and other purposes, it's 
been my repeated experience that they learn faster and more 
effectively if I simply say, "There are  two kinds of URLs, relative 
and absolute" and go from there. Maybe on a bad day I'll mention 
URIs. If I try and explain the whole notion of URI references vs 
URIs, the class nods off. It's simply not a practical or useful 
distinction.

I thought 2396bis might make this clearer (it does make some things 
clearer) but it's becoming apparent it's not going to. We seem likely 
to remain stuck in a division between the rarified realm of the spec 
writers and the practical world of developers and content creators; 
and the terminology mismatch between the two worlds is likely to 
continue to cause confusion for anyone moving between the two domains.

Oh well. I suppose it's not that big a deal. People will continue to 
write web pages and design new URI schemes and APIs based on existing 
practice while blithely ignoring the actual specifications, just as 
they have for the last ten years; and the world will continue to spin 
on its axis.  In the large scheme of things, the minor 
inconsistencies and noninteroperabilities that will arise because the 
spec is too unclear for non-language-lawyers to correctly interpret 
will be at worst minor warts on the face of the Web. Still, it does 
seem like a missed opportunity to finally remove these warts once and 
for all.
-- 

   Elliotte Rusty Harold
   elharo@metalab.unc.edu
   Effective XML (Addison-Wesley, 2003)
   http://www.cafeconleche.org/books/effectivexml
   http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA

Received on Thursday, 19 August 2004 13:34:40 UTC