About appeals

[Frank Ellermann]

Doug Ewell wrote:

 [registry file format] 
> Note that these are two different arguments; UTF-8 need not 
> imply XML.

Not necessarily, right.  It is a combination justifying an UTF-8
registry as far as I was concerned.  IANA is anyway interested in
XML formats for its registries, and XML offers a way to indicate
this charset *within* a document (unlike text/plain BOM kludges).

In comparison with XML "record-jar" is an obscure format, it has
no MIME type, it will be handled as "some kind of text/plain" in
practice.  And text/plain means ASCII or any local charset, as it
happens this is not UTF-8 on my platform (was OS/2, now Windows).

BCP 137 (RFC 5137) offers hex. NCRs for this situation.  And BTW,
adding hex. NCRs to BCP 37 was inspired by RFC 4646 among others.

> The group spent ages discussing this and I don't think it can
> be said that all views were not heard.

I don't go as far as fearing *DOOM* for this issue, I trust that
the reviewer and IANA can manage this.  But registry users will
get text/plain files when they download a registry, and this can
cause havoc.  "Hostile to users", I passionately hate such ideas.

  [review period for maintenance updates]
>> It could be also bad if the actual update is forgotten due to
>> other discussions here.
 
> When that happens, it is almost always my fault as the Reviewer's 
> clerical assistant, and in that case RFC 4646 permits -- nay,
> encourages -- this list to tug on Michael's and my sleeves and
> remind us of pending business.

Yes, but when I wrote "forgotten" I meant all reading this list,
not only you.  I also forgot the Moldova record.  I only found
it, because I feed my XML converter manually, and the registry
file-date was still the same when I expected an updated version.

 [scope vs. extlang]
> Changes in 4646bis happen on the LTRU list.

Yep, that was meant as example that I'm not against all changes,
and updated this point in my script for the purpose of checking
4645bis months after I left the WG.  It has of course no effect
for RFC 4646 registries, there can't be an <extlang> or <scope>.

 [EU, UK, EA, IC, AC, DG, TA, CP]
> I do feel a formal appeal is better than making veiled 
> allegations about the legitimacy of WG decisions and integrity
> of the I-D editor on mailing lists.

Nothing about it was veiled.  I considered it as not legitimated,
and considered an appeal proposing to replace the 4646bis editors.
But I arrived at an "unsubscribe LTRU" decision, waiting for the
IETF Last Call.
  
You: <http://article.gmane.org/gmane.ietf.ltru/9168/match=eu+cldr>
Me : <http://article.gmane.org/gmane.ietf.ltru/8836/match=eu+cldr>

The other codes are of course far more ridiculous, and excluding
the only code that could remotely make sense, because users will
always confuse UK and GB, is "hostile to users".
 
> Language tags -- specifically, region subtags -- may look as if
> they are related to ccTLDs, but they are not.

Yes, you don't need to convince me again, you and others did that
already in the pre-4646 discussions.  In practice region subtags
and ccTLDs are related, as they are both based on the same source,
ISO 3166.  With some historical exceptions (including AC, EU, UK)
on the side of the ccTLDs.

There are also historical exceptions (including BU, CS, DD, YD)
on the side of region subtags, that is how RFC 4646 is designed.

> The continued existence and maintenance of the .su domain [1]
> sort of kills the argument that ccTLD usage of ISO 3166-1 is
> exemplary

Actually EU kills that argument.  Maybe SU might be phased out.
 
> [1] http://en.wikipedia.org/wiki/.su

<http://en.wikipedia.org/wiki/ccTLD#ISO_3166-1_and_ccTLDs>
You can guess who added the RFC 1591 quote in this chapter. 

 Frank