For me, the difference tends to fall in terms of whether or not the IRI is intended to be dereferenceable or not. If the IRI is intended to be used solely as an indentifier (e.g. atom:id, XML Namespaces, etc), then character-comparison (at a maximum) is appropriate. If the IRI is intended to be a navigatable reference, then some degree of normalization is appropriate.
In the case of the link/@rel and category/@scheme, Atom does not say whether or not the IRI's should be considered to be dereferenceable. @rel and @scheme are clearly intended to be used primarily as identifiers. - James Martin Duerst wrote: >[snip] > Well, RFC 3987 (and 3986) don't *exactly* say "use whatever you prefer", > but for the case at hand, what they say comes pretty close. > What they try to do is give advice on what kind of normalization > or comparison function(s) is/are adequate in which case. > Given that it's difficult to immagine every possible case in > advace, the advice they give unfortuanetly has to be somewhat > vague. > > BTW, as a co-author of RFC 3987, I of course appreciate any > suggestion on how to make it clearer, and any suggestions for > fixes or changes (should they be necessary). These should be > directed to [EMAIL PROTECTED] > > Regards, Martin. > > > > #-#-# Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University > #-#-# http://www.sw.it.aoyama.ac.jp mailto:[EMAIL PROTECTED] > >
