Re: [I18ndir] Will there be an I18N directorate meeting at IETF 108?

[John C Klensin <[e-mail]>]

--On Tuesday, June 23, 2020 14:15 -0500 Nico Williams
<[e-mail]> wrote:

> On Tue, Jun 23, 2020 at 02:41:40PM -0400, John C Klensin wrote:
>> Obviously the more fundamental question is "is there such a
>> thing as an I18N Directorate, or is there, at best, a review
>> team"?
> 
> I didn't think there was any doubt about the existence of an
> I18N Directorate.  I don't understand the distinction between
> "directorate" and "review team".  The security directorate is,
> really, just a "review team", yet they meet.

Let me try to explain the distinction because it is important to
the history of the directorate and the purposes for which it was
requested and, presumably, organized.  Traditionally, a
"directorate" (or sometimes an "Area advisory group") in the
IETF was an advisory body, normally for an area.  Its membership
was expected to show a high level of subject matter expertise
(both individually and collectively and expertise, not just
opinions) and to work together.  Strategic issues, including
what work areas or documents should be developed and scoped,
were the primary focus, with document reviews being secondary
_and_ generally being considered a collaborative effort.  And
they started to develop precisely because, in a number of Areas,
the expectation that the ADs would be experts on all topics
affecting that Area was unrealistic.  When Patrik and I
submitted the BOF that evolved into this group, what we had in
mind was something with that orientation and at least that
strong, possibly with a role of advising the whole IETF and not
just the Apps/ART ADs.  

Since you used the security area directorate as an example, my
recollection is that it came into being precisely to provide
perspective and advice on the ADs on topics in which they were
not fully expert.  One implication of that expectation about
expertise is that, if a topic were brought to that group that
was clearly related to network security, but no one in the group
had a clue, it would imply either that the AD(s) had done a bad
job of selecting and approving members or that the Security Area
was in deep trouble.  I18ndir is different in another way: while
I suppose it includes the chairs of every WG in the ART area
that is focused in i18n issues -- in whole-area directorates,
that facilitates detection and avoidance of conflicts about the
work of different WGs--  the number of such WGs in ART has been,
I believe, zero since PRECIS wrapped up.   If secdir has really
evolved into a review team, I know several prior security ADs
who would be disappointed, and maybe quite concerned.

By contrast, a review team is typically just that: a group of
individuals that do reviews on a volunteer and/or rotating
basis.  One would hope the reviews would be appropriately
expert, but, if you have read a significant number of
review-team reviews in the last few years, you will have seen
many in which the reviewer does not understand the subject
matter of the document being reviewed.  The better (IMO) of
those reviews are clear about what they don't know and proceed
to do what they understand, which is to pick general nits.
Others do not acknowledge (or even understand) what they don't
know and therefore declare a document ready to go when what they
actually mean is that, if there were problems, they aren't
qualified to spot them. 

>> I think it is fairly clear that the latter does not need to
>> meet.
> 
> The security directorate normally meets at lunchtime on
> Tuesdays.

It does not seem to me that "they do it, therefore we should" is
a particularly good reason, especially because they are
area-wide and presumably have WGs to coordinate and we don't.

And, fwiw, the reason for my "is this a directorate, an extended
design team, or just a review team" question is that I have been
posing questions about how to get the supposed directorate to
engage on a number of strategic issues since around the time we
put the "IANA IDNA Tables" issues to rest and especially since
we got what is now RFC 8753 through the IESG in, IIR, early
November.  Of particular interest to Asmus (since he responded
to your suggestion) might be a discussion of the IESG comments
on 5891bis, how to respond to them, and whether the directorate
(not just me or Asmus and me) are willing to defend the (still
evolving and as yet unposted or even sent to Asmus to check
draft-klensin-idna-5891bis-06).  I've written Pete about it
several times and Barry and Pete slightly fewer times, and have
not gotten an actionable response in months.  Partial questions
sent to the list have not gotten much more response.

So, if there is a plan to get things moving again (and moving in
the directorate direction), perhaps with a new co-chair or two,
and a meeting was part of that plan, I'd be completely in favor
of a meeting.  However.. if nothing else, meeting time, even for
all-remote meetings, is not free, especially with time zone
issues, and unless there is an agenda and some indications that
a meeting would be productive...


> IMO it'd be useful to be able to have a regular meeting of this
> directorate.

And that brings me to your follow-up note:

> I should add that I would like to discuss the filesystem I18N
> matter, and others might like to discuss updates to IDNA or
> whatever.  There's plenty of fodder for an actual meeting of
> this group.

I've been away from operating system and filesystem design work
since the middle of the 1970s, but was close enough to it before
that to have been involved in and understood the analysis and
tradeoffs that went into FTP as well as the difference between
the TENEX, TOPS-20, ITS, CP/CMS, and Multics and Unix file
system designs and others that contributed both to the FTP
design and the "if you want to work with our systems you need to
simulate them" attitude that influenced the early design of NFS
(and which led directly to some of the issues that you are now
having with later versions).  But I still remember enough to be
quite confident that your comments so far have vastly
oversimplified the issues especially when one remembers that,
when faced with a widely-deployed and established base, the
question of how one could get people to adopt a new model and
replace what they have already is at least as important as how
wonderful that model would have been if people had thought of it
and been convinced in 1965 (a date not chosen by accident).
While a new design today could much more easily accommodate
Unicode than designs that assume single-octet characters, from
an actual i18n standpoint, those are the easy problems and there
are harder ones as soon as one starts to compare strings [1].  

But you have been around the IETF long enough that a discussion
of a new idea or proposal based on a few comments in an email
message, or a presentation at a meeting, rarely if ever goes
anywhere useful.  If you want to promote your file system ideas,
do what has been explained in almost every Newcomer's
Orientation for decades: write an I-D that explains them (and,
ideally, the advantages and tradeoffs involved) in sufficient
detail that they can be evaluated by others.  Then try to get
someone interested.  The i18n directorate is clearly not the
right group.  If you are convinced it is really an ART matter,
see if DISPATCH wants to play.  It seems to me more likely that
you should be having a discussion with the Internet area.

In the meantime, if you would like something useful to discuss
on this list, or maybe in a meeting, it would be how to get the
NFS group the review they asked for --an analysis of how and why
the document does and does not work from an i18n perspective and
at least hints about how the issues might be addressed -- rather
than (or even in addition to) an explanation about how the would
would be a better place if their entire file system model,
multiple implementations, and a widely-deployed base were
replaced with new ideas.

As to "others might like to discuss updates to IDNA or
whatever", the evidence --to my huge frustration-- is to the
contrary.

best,
   john


[1] Am I the only one who has noticed that, as far as the RFC
index and the IETF datatracker are concerned, there appear to be
two rather separate people with similar-sounding names.  One of
them is named Patrik Faltstrom and the other is named Patrik
Fältström and a search for one doesn't find the other.  Is
that what is wanted?  Did anyone think to ask either the i18n
directorate or, for that matter, Patrik?  I assume other pairs
of people, like Martin Dürst and Martin Duerst, get to look
forward to similar futures.  Now that, IMO, would be an
interesting discussion for the directorate but iff someone with
the ability and willingness to change things were listening.