MVALID (was Re: M-Label or MVALID, and dangers with mappings?)

Pete Resnick presnick at qualcomm.com
Sun Apr 12 21:34:24 CEST 2009


On 4/11/09 at 8:26 AM +0200, Patrik Fältström wrote:

>I have seen the discussions regarding M-Label, and my say that as an 
>editor of the tables document, I think it might be "more 
>interesting" to define MVALID as a property that is calculated in 
>the tables document.
>
>Suggestion:
>
>MVALID would be a codepoint that is mapped, according to the 
>standardized mapping function, to something that is not DISALLOWED.

I really hated M-label. The suggestion of MVALID (which I also don't 
like so much) finally crystallized in my head why I thought this was 
such a bad idea: MVALID is only definable in terms of characters that 
are "mapable" to "not DISALLOWED". But *anything* is mapable: A 
mapping function could choose to map U+2022 (bullet) to U+002D 
(hyphen-minus), and for some input methods in some contexts, that 
might make perfect sense; in others, it might be best rejected as 
"illegal input". But, as I tried to say before (and was summarily 
ignored), a mapping function of this sort is an input method/user 
interface issue, not a protocol issue. In all cases, a mapping is a 
best guess at what U-label string the user intended. We are already 
in the land of "let's make a mapping function that preserves 
IDNA2003, except for a particular set of characters". Mappings are 
going to change, and for some contexts (whether different locales or 
using them for web browsers vs. an application for registration) 
mappings might reasonably be different.

If we're going to deal with mappings, I want it to be a separate 
document. I am fine with that document saying, "Use this mapping for 
maximal compatibility with IDNA2003". I am reluctantly OK with that 
mapping document laying out a full mapping instead of being 
rule-based or having a reference to an external algorithm. But I 
strongly object to any such mapping (or the definition of M-label or 
MVALID) appearing in the current documents.

pr
-- 
Pete Resnick <http://www.qualcomm.com/~presnick/>
Qualcomm Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102


More information about the Idna-update mailing list