Standardizing on IDNA 2003 in the URL Standard

Gervase Markham gerv at
Fri Aug 23 12:03:49 CEST 2013

On 23/08/13 10:49, Vint Cerf wrote:
> If we go down the IDNA2008 + TR46 path, I think we ought to be very
> explicit about a date certain to drop TR46 treatment so as to eliminate
> mapping and to instantiate the uniqueness properties of IDNA2008. Is
> that possible and what timeframe makes sense? The longer we wait, the
> harder it will be to get there.

What exactly do you mean by "TR46 treatment"?

As Mark says, TR46 is 3 things:

1) Different treatment for the 4 deviations (ZWJ, ZWNJ, eszett, sigma)
2) Allowance of non-letters
3) A mapping, e.g. casefolding

The use of a mapping such as the one defined in 3) is permitted by
IDNA2008; I would expect that mapping to continue being used
indefinitely, as a valid part of the system.

So are you asking about when we should drop 1) and 2)?

We should consider the option of not tying the two together.


More information about the Idna-update mailing list