Re: Minor issue in draft-ietf-sieve-spamtestbis-02.txt

[Cyrus Daboo <[e-mail]>]

Hi Ned,

--On June 1, 2006 2:07:26 PM -0700 Ned Freed <[e-mail]> wrote:

>> > I just noticed that although the draft makes heavy use of the
>> > relational :value match type, it says nothing about how :count should
>> > be handled. I suggest adding a note to say that the "count value" is 1
>> > if there's a spamtest or virustest result to check, 0 otherwise.
>
>> I like that... I could swear that had been suggested at some point but
>> can't find it, so probably not.
>
> Interesting. I too have a vague recollection that this came up before,
> but a quick search of the list archives failed to find any references.
> Perhaps at a WG meeting or bar-BOF?

Interesting - I can't ever recall that being discussed before, at least in 
my presence :-). But it does provide an interesting solution to the 
'untested' case which is somewhat awkward with the :percent option.

So here is a proposal: remove the current "untested" return value and 
instead require scripts to use :count to determine that case with the 
behavior of :count being defined as Ned suggested. An example:

           require ["spamtestplus", "fileinto",
                    "relational", "comparator-i;ascii-numeric"];

           if spamtest :count "eq"
                       :comparator "i;ascii-numeric" "0"
           {
               fileinto "INBOX.unclassified";
           }
           elsif spamtest :value "lt"
                       :comparator "i;ascii-numeric" "37"
           {
               fileinto "INBOX.not-spam";
           }
           else
           {
               fileinto "INBOX.spam-trap";
           }


-- 
Cyrus Daboo