ietf-mta-filters
[Top] [All Lists]

Re: The last remaining issue in the Sieve variables draft

2005-04-25 03:31:38

On Mon, 2005-04-25 at 10:38 +0100, Nigel Swinson wrote:
Which reminded me: can we change the last sentence to:

   Unless explicitly allowed by a Sieve extension, the user can not 
specify a namespace when setting variables with SET.

Once again, I don't like to be too restrictive for no good reason.

the Syntax line would have to change, too, to refer to variable-name
rather than identifier.  the above change would make it legal to set
numbered variables.

I'd prefer other extension to add their own action if they need it.  the
way I figure is that most uses of namespaces will be read-only, or with
specialised needs for setting values.  you don't really want

   SET "mimestructure.body.2.1" "Hi there";

do you?

I would also prefer extensions to explicitly allow specifying a
namespace.  And as Kjetil points out I think it very unlikely that
we'd agree to any extension that permitted setting of "silly"
namespaces, but I think it likely that we'll use namespaces a lot in
the future of the language, and prohibiting the possibility of setting
variables in namespaces sounds to me like something we'd regret.  You
said yourself "most" uses will be read-only.  You didn't say "all"
uses will be read only.

okay, but it seems like a lot of hair to add, since we now have to add
in wording about the validity of a namespace and the use of numbered
variables.  do you have a use case?
-- 
Kjetil T.


<Prev in Thread] Current Thread [Next in Thread>