Hi, Martin: Sorry that I have not respond to you on your comments earlier, since I was run out of time to explain what I meant.
The frame that I have in mind is in IDNA now as toASCII and toUnicode and is consistent with what I have called idn-label earlier in idn-map I-D. Now that someone else have defind the frame, and my question is still why not let the user group define their own mapping as TSconv-2 does? Why do we trying so hard to exclude them? As you have pointed out, a lot of people have been working hard to come up with a 100% solution, and there isn't one. Is it the time we take a limited solution say, only 1-1 mappings of TC/SC is allowed and trading off n-1, 1-n cases (like the SC standard did) and telling the user, "sorry this is an identifier system, not a word processor". Or a layered approach, to hide its imperfection? Liana On Fri, 26 Oct 2001 16:08:28 +0900 Martin Duerst <[EMAIL PROTECTED]> writes: > At 02:57 01/10/25 -0700, liana Ye wrote: > > >What if we define a frame and make it possible to let any > >script to be mapped by their own user group, as long as > >the user group can come up with their agreement on how > >to do the mapping. The [nameprep] is severing as example > >of such mapping. BTW, is "PROFILE" have such meaning? > > Well, there is only one DNS for the whole world. That's > why profiles or frames (or whatever you call it) are not > a good idea. > > Also, DNS server have heavy loads. That's why some kinds > of solutions designed by user groups may not be workable, > and why we have DNS experts in this group. > > Regards, Martin. > > > > >Liana > > > > > >On Thu, 25 Oct 2001 10:08:48 +0200 > =?ISO-8859-1?Q?Patrik_F=E4ltstr=F6m?= > ><[EMAIL PROTECTED]> writes: > > > --On 01-10-25 12.54 +0900 Martin Duerst <[EMAIL PROTECTED]> wrote: > > > > > > > At 19:41 01/10/23 +0200, Patrik F舁tstr仁 wrote: > > > > > > > >> o Nameprep mappings is defined by UTC, not IETF. > > > >> -> I am nervous when we in IETF talk about these things > > > > > > > > The main mappings used by nameprep (casemapping and > normalization) > > > > are defined by the UTC. But it is IETF that chooses which > mappings > > > > to select (e.g. NFC vs. NFKC), it is IETF that may request > other > > > > mappings to be created (NFC was created based on a request by > W3C, > > > > an additional file was added to help IETF with case mapping, > > > > a (tentative, implicit) request for TC/SC mappings was turned > > > down). > > > > > > > > Also, nameprep contains a few additional mappings and > prohibitions > > > > that are motivated by the specific situation of domain names, > > > > without taking these from anywhere. > > > > > > You are correct, but I am still nervous. > > > > > > paf > > > > > > > >
