Re: [art] Fwd: New Version Notification for draft-nottingham-how-did-that-get-into-the-repo-00.txt

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

--On Friday, August 17, 2018 11:04 +1000 Mark Nottingham
<[e-mail]> wrote:

>> The scheme part of URIs is case-insensitive so I'm not sure
>> you can say the URI string is matched
>> "character-for-character, case-sensitive". An example
>> that looks like hex, but supposed to be case-sensitive is not
>> ideal.
> 
> The token should be emitted, stored and submitted in the same
> form -- do you really think people will be trying to normalise
> it?

First of all, if a string looks like Hex, yes, I can imagine
people trying to case fold it (which is different from
normalizing it).  More generally, while there has been a general
rule all along that running text should be normalized only as
part of matching or comparison operations, the rule is less
clear for identifiers both because special circumstances (as in
IDNA) may suggest getting things into normalized form as early
as possible and keeping them that way and the consensus about
other identifiers is, IMO, shifting.  So, yes, depending on how
those token are processed and through what, I can imagine them
being normalized without those who are processing them paying
much attention to whether or not that is happening.

   john