Re: SORT document, draft -20

[Alexey Melnikov <alexey.melnikov@xxxxxxxxx>]

Mark Crispin wrote:

> On Fri, 7 Mar 2008, Alexey Melnikov wrote:
>
> > >    A server which implements the SORT and/or THREAD extensions MUST
> > >    collate strings in accordance with the requirements of I18NLEVEL=1,
> > >    as described in [IMAP-I18N], and SHOULD implement and advertise the
> > >    I18NLEVEL=1 extension.
> >
> > Typo: the last I18NLEVEL=1 should be I18NLEVEL=2.
>
> I accidentally deleted your previous message where you asked about this.
>
> I18NLEVEL=1 is correct.  Currently, there are servers that implement 
> SORT/THREAD in accordance with the requirements of I18NLEVEL=1, but do 
> not advertise I18NLEVEL=1.  We want those servers to upgrade sooner, 
> rather than later, when imap-i18n (finally) comes out; because there 
> is an ambiguity that a server might advertise SORT/THREAD without 
> complying with I18NLEVEL=1.
>
> Note the next paragraph:
>       Discussion: the SORT and THREAD extensions predate [IMAP-I18N] by
>       several years.  At the time of this writing, all known server
>       implementations of SORT and THREAD comply with the rules of
>       I18NLEVEL=1, but do not necessarily advertise it.  As discussed
>       in [IMAP-I18N] section 4.5, all server implementations should
>       eventually be updated to comply with the I18NLEVEL=2 extension.
>
> I don't think that SORT/THREAD can mandate I18NLEVEL=2 right now, even 
> with a SHOULD.

But this contradict the text you already have in section 7:

  As stated in the introduction, the server SHOULD support the
  [IMAP-I18N] I18NLEVEL=2 extension and follow its rules to perform
  collations in the SORT and THREAD extensions.

  If the server does not support I18NLEVEL=2, the rules of I18NLEVEL=1
  MUST be followed; that is, strings MUST be collated according to the
  i;unicode-casemap collation described in [UNICASEMAP].