channel subscription rules (was: Re: [ticker-dev] Latest draft)

[Ian Lister]

On Mon, 19 Apr 2004, David Arnold wrote:
>so, if there are people who feel strongly that requiring ticker
>clients to subscribe using fold-case() is the right approach, i'd be
>willing to entertain proposals for such an amendment to the spec (in
>the next few days :-)

I suggest the following amendment:

In section 6.1, replace the following text:

   The Elvin subscription language provides operations to
   transform strings to canonical representations to ensure
   that strings using different representations of the same
   characters are correctly matched.  Implementors of Tick-
   ertape protocol clients SHOULD use these features to
   ensure that user expectations are fulfilled.

with:

   The Elvin subscription language provides operations to
   transform strings to canonical representations to ensure
   that strings using different representations of the same
   characters are correctly matched.  Implementations of the
   Tickertape protocol MUST use the decompose-compat and
   fold-case subscription langage functions to normalise
   references to the 'Group', 'From' and 'Message' attributes
   unless explicitly instructed to the contrary by the user.

   For example, a subscription to a group named 'Chat' would
   be made with the expression:

      fold-case(decompose-compat(Group)) ==
      fold-case(decompose-compat("Chat"))

   When sending a message as a reply to a previously received
   message, implementations MUST set the value of the 'Group'
   attribute in the reply to be exactly the same as the value
   of the 'Group' attribute in the message to which it is a
   reply unless explicitly instructed to reply in a different
   group.

Ian