Exception table (was: Re: 06FD and 06FE should be PVALID for Sindhi)

[John C Klensin]

--On Wednesday, April 02, 2008 10:13 AM -0400 Andrew Sullivan 
<ajs at commandprompt.com> wrote:

> Dear colleagues,
>
> On Wed, Apr 02, 2008 at 03:53:54AM -0400, John C Klensin wrote:
>
>> It is perhaps also worth reminding ourselves that standing
>> WGs to review  things have rarely worked well and that the
>> odds of spinning up a full WG  to review a small handful of
>> characters (or fewer) are very small (I offer  the length of
>> time we spent of the charter for this effort as an example of
>> why).
>
> I think I'm probably confused.  While I have a great deal of
> sympathy for the IANA registry option, the above consideration
> seems to me to be a problem no matter whether we prefer
> issuing a new short document for each revision or updating the
> relevant registry each time.  For in either case, the problem
> is surely that a considerable amount of consultation with a
> wide group of people will be needed to alter the list of
> exceptions.

In principle, you are certainly correct.  In practice, please 
don't confuse "should" with "will".

> I am familiar with one case where the RFCs that were an output
> from a working group were altered and moved along the
> standards track without creating a new WG.  All the review
> happened on the WG's old mailing list.  That seemed to work
> ok, but it was a pretty narrow community of users (I'm
> thinking of EPP), so I don't know that it's a good analogy.
> Still, won't we need something like that whatever approach we
> use for updating the exceptions list?

Watch out, you could turn into an optimist :-(

The problem is that it is often very hard to get real, in-depth, 
review of anything that is complex and very specialized.  We get 
a level of energy in active WGs that promotes review, 
discussion, and mutual education.   Even then, we've had 
documents go through the WG process that are reviewed carefully 
and with deep knowledge of the subject matter by only the author 
and one other person (and sometimes not that many).   "Review 
teams" don't help with this if the reviewers don't have the 
specialized knowledge needed to do the evaluation.  Even the 
IESG review may involve elements of "the eyes glaze over and I 
certainly hope that someone else understands this".  So we have 
approved standards-track documents that were completely 
understood by less that three or four people, with or without 
WGs (in recent practice, the review quality standard for 
independent submissions is sometimes higher than that for WGs 
because the question of "who else understands this and have they 
reviewed it" is more likely to be explicitly asked, while, with 
WGs, it is assumed that someone exists and has).

For character lists, assignments, classes and properties, our 
colleagues on the UTC have devoted very large fractions of their 
professional lives to understanding both the general and the 
specific issues.  While we (and I'm certainly no exception) 
spend a lot of time pointing out what they got wrong and what 
was questionable, we all need to remember that, because of that 
level of devotion, they have gotten a huge fraction of circa 
100K assigned code points either right or "close enough".

One of the advantages of an extraordinary review process is that 
it stays extraordinary.  One could write rules like "one must 
get input from UTC" and "one must find relevant registries and 
get input from them" and expect that they will be followed.  In 
theory, one would do that with an IETF WG as well, but, in 
addition to having to form the WG (and a half-dozen experts, if 
we can get even that many, won't be able to do that), we've got 
the problem that, despite all of the time that is spent on them, 
provisions of WG Charters are frequently ignored with complete 
impunity.

best,
   john