What registries might do (was: Consensus Call on Latin Sharp S andGreek Final Sigma)

[Alexander Mayrhofer]

> But to be fair, the "registry completely restricts" policy won't work
> if ß is PVALID in IDNA2008.  Suppose .at by policy excludes (and
> doesn't map) ß.  What will happen then is that IDNA2003 clients will
> map ß to ss, and so a label ${example}ß.at will work for them; when
> the human on the other side of that client does its next upgrade,
> ${example}ß.at suddenly stops working.  This is the problem that
> opponents of making ß PVALID are talking about.

Andrew,

Thanks for that - this is exactly what i am worried about - plus, i am really concerned about incompatible mappings. Sure we could do strict bundling to avoid this, but what would be the gain compared to the current situation?

Alex