Re: Next step (Was: I-D ACTION:draft-klensin-unicode-escapes-00.txt

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

--On Wednesday, 24 January, 2007 11:03 -0500 der Mouse
<[e-mail]> wrote:

>> The mail seems to me to fall into three categories:
> 
>> (1) Evidence that I still cannot do ABNF properly, plus a few
>> other editorial issues.  The most important of the ABNF
>> issues is that "u" and "U" cannot be specified as different
>> without escaping both into their character code positions
>> (since character-literals in ABNF are case-insensitive).
 
> This says to me that ABNF is the wrong tool for the job here.
> It seems to me that we really want to specify "backslash"
> "lowercase u" and "backslash" "uppercase U" rather than octet
> sequences, so that the result is not tied to ASCII (while it's
> unlikely that anyone wants to inject Unicode into EBCDIC text,
> I just don't like seeing a problem "fixed" by introducing
> other problems in a workaround instead of addressing the real
> problem).

It is clear to me that, if we can decide what we are trying to
do, we can invent ABNF to do it -- without introducing the
constraint about which you are concerned.   It may be a bit
clumsy, but it is feasible.   

The main problem here is a personal one: for historical reasons
going back long before there was an IETF, I've never been a big
ABNF fan.  I'm even less of a fan of some of the extensions in
RFC 2234/ 4234 which I consider fussy, too extensive, and
unnecessary. While I have accepted the IETF consensus on ABNF
(with the extensions), that distaste, aggravated by the onset of
old age, has impeded my learning it well and preventing my
remembering all of the details that I have learned from one time
to the next.  The IETF should certainly not make decisions on
the basis of my problems with ABNF; even I won't make decisions
on that basis.  I very much appreciate the efforts of those who
have learned that they have to check my work and their patience
in correcting it.  But let's make the right decisions and not
get hung up on my idiosyncrasies. 
 
> If ABNF can't do what we want, I'd say, use something
> better-suited to the need.  Perhaps it would be enough to say
> that the spec uses ABNF except that the character after the
> backslash is case-sensitive?

As I said, I'm confident we can figure something out.  That
would certainly be among the alternatives although my instinct
would be to hunt up Dave and Paul and ask them for advice, at
least as a first step.

> That said, I don't consider this a showstopper - not that my
> opinion of what is or isn't a showstopper necessarily
> matters....

Nor mine.

     john