Re: [idn] We are quibbling about WHAT?

[Dave Crocker <dcrocker@brandenburg.com>]

At 08:13 PM 7/29/2001, Eric Brunner-Williams in Portland Maine wrote:
>I think the encapsulate vs extend debate is about architecture. It
>just happens that 63 bytes does focus some attention on effective
>characters per mechanism, and encapsulation schemes using only 5 bits
>plus a nickle per octet are at a disadvantage over extension schemes
>that use at least 7 bits per octet.

1.  My impression is that, in fact, expressions of concern have more often 
had, really, to do with encoding efficiency than anything else.  Or, at 
least, the arguments with the most substance.  And to continue the 
analytical quibble, my impression is that the character sets that are more 
restricted with UTF-8 *AND* with ACE have more information density that 
such un-dense stuff as ASCII.

2.  The name of "architecture" is a very double-edged sword.  Architecture 
that worries only about computer science purity, for example, often winds 
up being highly impractical.  It does not attend to a sufficient range of 
real-world requirements/constraints.

That was why I went through the transition discussion, separately.  The 
difference in transition ease, risk, etc., between ACE and UTF-8 is exactly 
the reason that the horrible technology, called MIME, was used for 
introducing binary, segmented, labeled data to email content.

>The transition impact discussion can't be refuted. It was the subject
>of RFC 801, and others. I'm particularly fond of lines 51 - 64 of
>that transition plan by the way.
>Time to choose an ACE and move on? I'll pass thanks. A half-dozen
>browser-hack marketeers, a trio of monopolies, and the DNSO food
>fight, is inadequate motivation to solve a complex problem. A good
>solution is, and differences in that are why we are where we are.

Having not yet seen any response to the analysis I offered, I beginning 
whether folks really are not interested in attending to transition pragmatics.

Where is the detailed response, rather than just rejection or 
acceptance?  Please skip the personalization of this.

Just as the tendency in the IETF is to say that the primary -- and maybe 
the only -- design concern is scaling, I've come to believe that the only 
concern when upgrading an installed base is the transition ease.

d/

----------
Dave Crocker  <mailto:dcrocker@brandenburg.com>
Brandenburg InternetWorking  <http://www.brandenburg.com>
tel +1.408.246.8253;  fax +1.408.273.6464