UNASSIGNED, DISALLOWED, PVALID, and MAYBE (was: Re: Unicode 5.2 -> 6.0)
John C Klensin
klensin at jck.com
Fri Oct 15 00:51:22 CEST 2010
--On Thursday, October 14, 2010 16:47 -0400 Eric
Brunner-Williams <ebw at abenaki.wabanaki.net> wrote:
> I think gTLD operators (and their respective technical
> services providers, where distinct) should be advised* that
> allowing any of the version delta characters before
> consultations** is premature.
--On Thursday, October 14, 2010 16:51 -0400 Andrew Sullivan
<ajs at shinkuro.com> wrote:
> if this is going to be a regular problem in future, it seems
> like one would be better to have some new class like
> PROBABLY-PVALID where characters we're not sure about live for
> a couple releases of the Unicode tables.
--On Thursday, October 14, 2010 14:34 -0700 Mark Davis ☕
<mark at macchiato.com> wrote:
> As far as an implementation is concerned, UNASSIGNED ==
Folks, this note is more about procedures than about substance,
-- The original proposals for what evolved into IDNA2008
contained a pair of categories, then called MAYBE YES
and MAYBE NO, that were intended as a kind of purgatory
for scripts and languages about which we weren't
confident that we had sufficient information, weren't
confident that anyone actually understood how to use
correctly for domain names, for which we thought more
consultation was needed, and just as a place to put
newly-added characters while they matured enough in use
elsewhere to make rational and informed decisions about
IDN use. Those categories were dropped during the
period just before the WG was formed but were documented
in I-Ds during that period. If my recollection is
correct no one made a serious proposal during the WG's
lifetime to put them back in.
Even though I have some regrets every time an issue like
this comes up (either in practice or in theory) I
personally think we should treat that decision as closed
unless someone is willing to make a serious proposal (by
definition, in the form of an I-D) to open IDNA and put
the categories back in... and then see if they can get
any traction for the idea given that the community
clearly wants to move forward with IDNs and not have to
go through more uncertainty while we work on new
category definitions and the associated tables.
-- The WG discussed, at great length, the questions of
whether UNASSIGNED and DISALLOWED were different in
practice and whether DISALLOWED -> PVALID transitions
should be treated differently from PVALID -> DISALLOWED
transitions. The first conclusion was pretty clearly
that UNASSIGNED was different from DISALLOWED because
implementations, registries, and would-be registrants
needed to assume that UNASSIGNED code points might or
would ultimately be assigned while DISALLOWED ones were
expected to be stable in that state. And the second
conclusion was neither change was acceptable except in
truly exceptional situations.
Again, my personal preference is that we not reopen
either decision because I don't think the advantages of
doing so come even close to outweighing the
disadvantages of destabilizing the IDNA2008 definition.
If someone has a different opinion for which they are
prepared to argue, please generate an I-D with a
proposal. But, at least IMO, we aren't going to make
progress by wishing that the WG's conclusions were other
than they actually were.
It is not clear to me that the above leads to any particular
conclusion about what we should do about these three characters.
But perhaps it could be helpful in focusing the discussion.
More information about the Idna-update