Removing symbols (was: RE: Proposed Charter for the IDNAbis Working Group)

[John C Klensin]

--On Thursday, 27 March, 2008 12:43 -0700 Shawn Steele
<Shawn.Steele at microsoft.com> wrote:

>> Personal opinion only, but this is more or less the boundary
>> point between "discuss charter and what this is about" and
>> "need to read the drafts before discussing them."
> 
> Reasonable point.  I'd've probably expected the charter to
> guide the drafts rather than the other way around though.
> Since I haven't been involved in this process before, perhaps
> I misunderstand what the charter's for.

If this WG were following the sequence "charter first, then
discussion, then drafts", that would be the case.  And that
order is certainly common around the IETF.   In this case, there
have been extensive discussions and corresponding drafts, for
well over a year and the draft charter makes reference to
reviewing those documents, so, "if you want to start digging
into details, look at the documents" becomes appropriate.

> Dropping a few code points and breaking back-compat is a lot
> different than dropping all of the symbols.  My understanding
> of the draft is that we intend to drop all of the symbols, so
> I'd expect that to be clearly allowed by the charter.  My
> interpretation of the charter was that "breaking" changes
> would be small in scope and limited to reduce their impact.
> 
> I'm not trying to argue about dropping symbols, I don't think
> they cause much harm, but also don't see how they're very
> useful, so I don't mind if they're in or out.  I do think that
> removing that many code points does break IDNA2003 (whether or
> not they were practical) and so the charter should allow this
> "big" of a change.  It wasn't clear to me if the charter was
> intended to reduce the scope of the drafted symbols behavior,
> or if it was allowed.

I certainly have no problem with the charter being changed if
necessary to be explicit about that.  I agree that it is a
significant element of the new proposals and not just a detail.  

However, FWIW, the combination of guidelines and good sense has
resulted in a situation in which removal of symbols (and
punctuation, and other non-letter characters) almost certainly
"breaks" fewer domain names (renders legitimate registrations
that identify real hosts and resources inaccessible) than, e.g.,
changing the behavior for ZWJ and ZWNJ from "discarded without
comment" to "allowed to appear in the ACE" (with some
restrictions) would.

> I still don't know how to interpret the "(a charter change is
> required) to remove any character mapping".  I understand now
> that doesn't mean the set of allowed characters, but if
> mappings between are being moved to "outside the protocol",
> then it seems like all of the mappings are being removed from
> the core proposals.

I'll have to go back and reread it (or hope someone else does
before I get back from an errand I'm about to go run).  But my
recollection was that the expectation was that we would remove
character mappings from the protocol itself (a basic part of the
new model) and that a charter change would be needed iff the WG
decided that a large selection of character mappings in the
protocol were necessary.  If that text has somehow gotten
unclear in the various iterations, we need to restore its
clarity.

   john