Re: Versioning (Re: [idn] The layers and character handling in DNS)

[John C Klensin <klensin@jck.com>]

--On Monday, 19 February, 2001 17:08 +0100 Harald Alvestrand
<Harald@Alvestrand.no> wrote:

> actually versioning nameprep does not work, for any case but
> the one where a string illegal in NameprepM is now legal in
> NameprepN (M < N).

That has been my impression.  But, if it is the case, it implies
that the first, and every subsequent version, of nameprep must be
perfectly correct for any unbanned characters.  Sorry, but being
certain about that...   Well, the WG just doesn't appear have the
level of divine guidance needed to be that confident; we need to
design things on the assumption that we will need versions
because we got something wrong and it turned out to be important.
If we can't, then the system fails.

> - in the human-readable string. Will be pelted with rotten
> bananas. - in the DNS entry for the name. But since we can't
> get to the DNS-stored entry before we have done nameprep, the
> chicken-and-egg conundrum strikes again. - in the global
> /hosts.txt file?????
> 
> For any existing name, Nameprep can't change.

Again, requires perfect wisdom.  Since we don't have that, I
think you are adding to my argument that the current approach
just cannot work.

    john