Comments on draft-ietf-idnabis-defs-10

[John C Klensin]

--On Monday, August 31, 2009 17:43 +0200 Patrik Fältström
<patrik at frobbit.se> wrote:

> On 31 aug 2009, at 17.37, John C Klensin wrote:
> 
>>> "An 'xn--' prefixed label MUST before one tries to convert it
>>> to a U-label be lowercased. I.e. all ascii characters [A-Z]
>>> in the label MUST be mapped to their lowercase variant [a-z]
>>> (the label must be turned into an A-label) before removing
>>> the 'xn--', and punycode decoding is taking place."
>> 
>> With small editorial adjustments (which I can make), that
>> formulation works for me if others are ok with it.
> 
> It is VERY important it does not say "...an A-label must be
> lowercased..." because according to the definitions we have
> today, an xn-- prefixed label with uppercase ascii can never
> be an A-label.

I agree.  See note just sent to Andrew.  My "editorial
adjustment" comment referred strictly to your hastily-written
English, e.g., "MUST before...lowercased" -> "MUST be lowercased
before one tries to convert it to a U-label".  

It is a little more complicated than this because the A-label
definition depends on conversion from a U-label (again, see note
to Andrew), so the upper-case forms are R-LDH labels that are
XN-labels but not A-labels.  If I recall, the text in Defs says,
approximately, that the set of XN-labels that are not A-labels
are prohibited as trash.  So, if we are going to do it this way,
Defs probably will need to be changed to define three classes of
R-LDH Labels:  A-labels, things that can be turned into A-labels
by lowercasing, and trash.  Not rocket science but, as Andrew
suggests, a little sensitive to be doing at this late stage.

     john