[Top] [All Lists]

Re: document status: 3028bis, body, editheader

2006-03-22 09:06:20

On Wed, Mar 22, 2006 at 07:13:31AM -0800, Alexey Melnikov wrote:
I've already sent a comment to Arnt asking him to allow en;ascii-casemap 
to work on Unicode characters. The advantage of this is that Unicode 
characters outside of ASCII are legal and not going to cause matching 
failure anymore.
I though you agreed to this in one of your messages to the mailing list.

So, are there any objections to change en;ascii-casemap to work on 
Unicode characters?

I would appreciate it, because right now the base spec does not allow
to match one arbitrary character.  To begin with, I interpreted 3028 to
talk of unicode characters, not octets, because the text was confusing
and nothing else made sense to me.  I just changed my implementation to
match octets for RFC conformance, after we discussed the topic to death.
It sounded like all other implementations already worked that way before.

Changing the comparator breaks all existing implementations.

If it was just for me, I'd say: So what? It is a change for the better,
having en;ascii-casemap working on octets stinks, using variables
"?" matches a single octet of a multibyte UTF-8 sequence (which is bad
and unexpected to most users) and progress sometimes has to break things.

But I am afraid somebody is going to say something about "compatibility"
and "violating existing standards" and stuff.  I can almost hear it.
Sieve is now widespread and since we could always invent a new comparator,
that works like "en;ascii-casemap" but on characters, perhaps you really
shouldn't listen to me. ;-)