M-label definition

[Vint Cerf]

sounds about right to me, Alex.

if M (label) is the mapping function, then we are seeking to define M  
and M-label such that M (M-label) = U-Label

implying that for all U-labels, M* (U-label) is an M-label where M* is  
M-inverse. M and M-inverse are not unique in the event that more than  
one M-label maps into the same U-label. If we allow M to include the  
null mapping, as Alex proposes, then U-labels become a subset of M- 
labels.

I don't necessarily propose that language above belongs in the  
specifications, but I am trying for precision in understanding the  
meaning of M-label.

Vint Cerf
Google
1818 Library Street, Suite 400
Reston, VA 20190
202-370-5637
vint at google.com




On Apr 8, 2009, at 5:21 AM, Alexander Mayrhofer wrote:

>> won't making M-label a subset of U-label destroy an important
>> property
>> of U-label as defined in IDNA2008?
>>
>> Under IDNA2008 there is a one-to-one mapping of every U-label to A-
>> label and vice-versa, without any mapping.
>>
>> M-label needs mapping to become a U-label and is thus a superset, I
>> believe.
>
> I fully agree. I think that "M-Labels" would definitely be a  
> superset of
> "U-Labels", and the relation might be as follows:
>
> - Any valid U-Label can also be considered an M-Label (because of that
> superset property, "null" mapping..)
> - many M-Labels may map to a single U-Label (because the mapping might
> loose information)
> - therefore, the mapping M-Label -> U-Label is not reversible
>
> Also:
>
> - A string that contains codepoints for which no mapping is defined  
> (eg.
> Unassigned) is just a string, not an "M-Label". (or, "any valid M- 
> Label
> must be successfully mapped to a U-Label"?)
>
> Does that sound roughly right?
>
> Alex
>
>
> _______________________________________________
> Idna-update mailing list
> Idna-update at alvestrand.no
> http://www.alvestrand.no/mailman/listinfo/idna-update