Re: [idn] Requirements I-D

[John C Klensin <klensin@jck.com>]

--On Thursday, May 18, 2000 07:32 -0400 "Brian W. Spolarich"
<briansp@acm.org> wrote:

>   I think this is an important issue to recognize.  The DNS as
>   perhaps originally envisioned and the DNS as it is used
> today are different in many ways.  I'm not a linguist, but I'm
> reminded of the way that language itself is used, in contrast
>... 
> passions, ideas, families, etc.  This implies a certain amount
> of "pressure" on the DNS to accomodate those desires.  One can
> complain that "the DNS wasn't designed for that purpose", but
> the reality is that people want to use it that way.

Absolutely.  It is also important to stress that, if you are
hearing "wasn't designed for that purpose" from the "old DNS
hands" (or me), it is almost never "so don't do it", merely that
the evolutionary process may be a little painful and that there
are a lot of issues that we need to look at carefully to be sure
that we don't accidentally disrupt the infrastructure. 

>   Perhaps this is a naive question (I haven't done the
>   research to see if I can find the answer), but I'm assuming
> that the IDN will support a wider character set even for the
> English implementation, beyond [A-Za-z0-9-].

I think most of us have been assuming that, if we open things up
to permit essentially any character,  there is little residual
motivation to preserve the restrictions for "English" (plain
ASCII)

>   What were the
> original assumptions behind the restrictions against
> 'HelloWorld!.com' and the like?  Shell escapes?

No, long before that.  There were a lot of issues, and I can't
reconstruct all of them.  And some of the definitions/ rules
were carried forward from pre-DNS ARPANET Hostnames.   But, for
example:

 * It was desirable to avoid code positions that were associated
with "national use" characters in IS646, for exactly the same
reasons we are having difficulty with locality issues here.

 * There were some human factors issues.  E.g., if someone
writes a domain name down with a pencil, and is not very
careful, it is hard to distinguish a few days later between
foo-bar and foo_bar.  Banning the latter, especially since it
was associated with some other ambiguities, just seemed like a
good idea at the time.

 * There might have also been some issues with unambiguous
EBCDIC conversions -- I just don't remember, perhaps someone
else does.

And there were probably others, possibly including "we have
eliminated so many special characters, lets get rid of
everything else besides a name separator ("-") and the label
separator (".") and make an easy-to-understand rule", but you
get the idea.

   john