- From: (wrong string) é <alb@riq.qc.ca>
- Date: Tue, 01 Sep 1998 16:19:06 -0400
- To: Leslie Daigle <leslie@Bunyip.Com>, Tripp Lilley <tlilley@perspex.com>
- Cc: Larry Masinter <masinter@parc.xerox.com>, Harald Tveit Alvestrand <Harald.Alvestrand@maxware.no>, Jon Davis <jdavis@inetinit.org>, ietf@ietf.org, URI distribution list <uri@Bunyip.Com>
A 10:32 98-09-01 -0400, Leslie Daigle a écrit : >Howdy, > >On Tue, 1 Sep 1998, Tripp Lilley wrote: >> Subject: Mailing List for iDNR (was RE: iDNR, an alternative name resolution , protocol)B >> >> On Mon, 31 Aug 1998, Larry Masinter wrote: >> >> > I suggest 'uri@bunyip.com' as the mailing list most appropriate for >> > this discussion. >> >> If we're just discussing alternate URI schemes, then I agree. However, I >> think we're getting into far more than that. Is there already a list for [Leslie] : >I think the uri@bunyip.com mailing list is a suitable place for discussing >the document Larry has proposed, insofar as it does deal specifically >with handling i18n in URI. > >However, the document does touch on issues of handling non-ascii characters >(specifically, matching, transitioning between 7-bit and 8-bit systems) >that are clearly more general than URI. If there is to be a serious >discussion of THOSE issues in the broader context, a new discussion forum >(or an existing one, more appropriate than uri@bunyip.com) would be excellent. > >I can see the need for an IETF-standard way of handling things like >relaxed character-matching (e.g., case-insensitivity, in 7-bit world), >because we are at the point where different protocols are going to take >different approaches (== mess). In my opinion, even with my limited >understanding of the issues, this will have to be an engineering compromise >solution, because I don't know if there is a globally-acceptable >answer to: > > should é match E in a case-insensitive-style match > >(Hint: ask a native French speaker from Québec, and ask one from France). [Alain] : Speaking as a native speaker of French for both Québec and France (I believe I am competent in both cases), I should say that the issue of matching is not the same at all as the debate over putting accents on upper case or not. It makes consensus in all French-speaking communities. So the answer to: "should é match E in a case-insensitive-style match?" is (both for Québec, France and all the Francophonie) : YES *and* NO. (; If you search for "cle" you sometimes want to retrieve clé and clef, or CLÉ and CLEF... and CLE... (why not, in the last case, it is sometimes a "fait accompli"). On the otehr hand, sometimes you want to do precise searches... You then want that the specific key "dû" only retrieve "dû" or "DÛ" (which means "DUE"), *but not* "du" or "DU" (which means "of the", maculin form, the 22nd most used French word, in other words, noise). Altavista, designed by a Frenchman (Louis Monier), does both, and it is nice. The rule in Altavista is that if you search with accents, you'll retrieve precise objects, as much as if you search with uppercase, and if you search without accents or with lower case only you will retrive all occurrences. Not bad as an implementation. There are more sophisticated ways to do whatever you want (which require setting some parameters on a Boeing 747 dash board), but this one is simple and improved. Alain LaBonté Chevalier, Ordre des Arts et des Lettres (France) Québec
Received on Tuesday, 1 September 1998 16:27:33 UTC