Consensus Call on Latin Sharp S and Greek Final Sigma

[Patrik Fältström]

On 29 nov 2009, at 18.35, Mark Davis ☕ wrote:

> The people on this list need to carefully consider their positions, since
> they will not have to live with the compatibility and security problems;
> German and Greek speakers will.

I was editor of IDNA2003, made a choice then, have seen since then how (as Harald said) bad it was to have mappings so deeply integrated into the standard so that not even domain name holders know whether they hold a domain name that is mapped to something, or not mapped to something else (whether the actual domain name include the 'ß' or 'ss'.

I now understand that choice was bad, and although (as you say) transition will be difficult, IDN is in practice not deployed yet. We have had ability to have IDN domain names deployed in .SE for a number of years, but deployment is still very very slim.

We better get rid of such mappings before we get wide deployment.

IF 'ß' is PVALID, there is an ability for the domain holder to have both the domain name with 'ß', and with 'ss' registered. Just like people in Sweden now (after several years) have started to have domain names with 'a' and 'å' registered (for example). Also, the registry (of .SE) had the ability to make decisions on bundles.

If 'ß' is DISALLOWED, those choices can not be made, and once again, my view and experience is that we definitely must get rid of mappings -- at least to the point that domain name holders know what domain name they have registered.

Mappings as a UI "helper" is of course definitely helpful! But that does not involve the registrant. It is a help for the one looking up the domain name with the help of various programs, protocols etc.

So yes, I have very carefully considered my position, and yes, I know it is a change from IDNA2003, but I think the IDNA2003 architecture is not possible to use long term.

For long term, we must move potential mappings to the UI, as a help for the one looking up domain names. Further away from the actual use of domain names in protocols. Away from for example the existing confusion regarding comparisons when using certificates, DNs when being used in various protocol parameters etc.

Because of that, my choice (as you all saw earlier) is (1).

   Patrik