Q1 is mapping on lookup permanent or transitional?

[John C Klensin]

--On Tuesday, April 07, 2009 17:15 +0200 Alexander Mayrhofer
<alexander.mayrhofer at nic.at> wrote:

> Thanks for the feedback. Comments inline:
> 
>> Of course, user experience changes regularly, both between
>> implementations and between versions of the same
>> implementation. Users get confused and disturbed.  Then they
>> complain.  And then they adapt.
> 
> Or they get frustrated, and blame the industry. Maybe i
> overestimate it, true.

No, you probably don't.   I might have said "get frustrated and
blame the industry" somewhere before  "...then they adapt".  But
it wouldn't change the conclusion.

> I wish i could see statistics about this - but unfortunately
> The current lowercasing / folding mechanics / stats would only
> be visible 
> to the User agents (hmm... Firefox URL filter list, maybe...?)

No, it is visible in Google's statistics about what shows up in
<a> elements, too.  The problem there is that most of what they
are seeing falls into the category of "not conforming to the
standards".  The question as to whether those cases are more
important, less important, or just different from the conforming
cases makes the interpretation of those statistics different,
but they are there.

> This would give a good insight - i'm fine if it turns out that
> nobody really enters / clicks
> On links that would fail under IDNA2008. (and i know i'm
> web-centric here, sorry)
  
>> And, if any of 
>>     Mac System 9.x -> Mac System X
>>     Windows 3.1 -> Windows 95
>>     Windows XP -> Windows Vista
>>     MS Office 2003 and earlier -> MS Office 2007
>>     Internet Explorer 6 -> Internet Explorer 7
> 
> True, but i think there's a significant difference whether a
> change is  the user's choice or it's forced down his throat.
> (and yes, i still run Office 2003 and XP ... See below for
> "significant advantage")

Sure.  And I've got no machines running Vista, several with XP
and two (one in very active use) running Win2000 (as well as
some FreeBSD boxes).  I do have Office 2007 running on one
machine, but only one.  But I've also gone to the trouble to set
up arrangements on Unix boxes to replicate Multics command names
I learned in the 60s and early 70s.  On the other hand, neither
of us are even vague approximations to typical end users.

>> can be used as examples (and I have many more, but they would
>> probably be less familiar), they adapt a whole lot faster than
>> "forever".   What we need is a plausible reason for making the
>> change other than "we did this to irritate you".  The users
>> don't necessarily have to believe that reason, but it has to
>> be plausible.
> 
> Basically i agree. But there needs to be a *significant*
> advantage For a user to happily adopt. And - being a bit
> egoistic here - i don't  see that advantage for the scripts
> that are relevant for us. We put  quite some effort into
> explaining to our registrars that the lack of  the "sharp s"
> was the best solution. They won't take the IETF too serious if
> we come back and tell them it's now the  other way round ...

Tell them that the IETF made a stupid mistake, that you went
along because you had no choice, and that things are being
corrected so that, in the future, if they construct domain names
that use words or word fragments contain Sharp S, they can spell
them correctly.   Some (like Georg) will consider that a very
significant improvement.  Others will consider it, and the
implied need to go register an additional name if their users
think of their names as containing Sharp S, an annoyance at
best.   Such is the nature of almost any change, even a
significant improvement.

Getting people to understand that less mapping and ambiguity
about domain names and their particular formats is A Good Thing
is going to be a much harder sell to the typical user, but that
doesn't make it incorrect.

> (and yes, i know it's not that easy)

There we agree.

     john