ietf-mta-filters
[Top] [All Lists]

Re: spamtestbis

2005-02-04 12:20:06

Hi ned(_dot_)freed(_at_)mrochek(_dot_)com,

--On February 3, 2005 2:08:56 PM -0800 ned(_dot_)freed(_at_)mrochek(_dot_)com 
wrote:

>> FYI the draft for the new version of the spamtest extension is now
>> available. The only change between this and the current RFC is the
>> addition of a :percent argument to the spamtest test to allow for a
>> numeric range of 0 - 100. The value -1 is used to indicate a message
>> that was not categorised in anyway.
>
> First, I don't think the -1 is a good idea because of how it interacts
> with numeric comparators. I'd much rather retain 0 as the value
> indicating no test was done.

So we would have:

    spamtest    interpretation
    value

                 0          message was not tested for spam
                 1          message was tested and is clear of spam
        2 - 99     message was tested and has a varying likelihood of
                            containing spam in increasing order
                 100        message was tested and definitely contains spam

Yes, I think this is what we want.

> The only other issue I see with the draft is the business of making
> require "spamtest" and require "spamtestpercent" mutually exclusive. I
> certainly agree that we want require "spamtestpercent" to subsume require
> "spamtest" - - require lists in sieve can be really long - but I wonder
> if making them mutually exclusive goes too far.

I can change the 'MUST NOT' to a 'SHOULD NOT' for that requirement.

> Additionally, one thing that isn't entirely clear with the current text
> is that the following is legal:
>
>    require "spamtest"; if spamtest :contains "foo" ...
>
> That is, spamtest without a :percent is allowed when require
> "spamtestpercent" is used. I think this should be clarified.
>
>

Here is some alternative text for this:

Old:

   SIEVE implementations that implement the "spamtest" test have an
   identifier of "spamtest" for use with the capability mechanism.  If
   the ":percent" argument is used with any spamtest test, then the
   capability idenitifier "spamtestpercent" MUST be present, and the
   "spamtest" capability MUST NOT be present.

New:

   SIEVE implementations that implement the "spamtest" test use an
   identifier of either "spamtest" or "spamtestpercent" for use with
   the capability mechanism.

   If the ":percent" argument is not used with any spamtest test, then
   one of either the "spamtest" or "spamtestpercent" capability
   identifiers MUST be present, though both SHOULD NOT be present together.

   If the ":percent" argument is used with any spamtest test, then
   the "spamtestpercent" capability identifier MUST be present, and
   the "spamtest" capability identitifer SHOULD NOT be present.

Much better. I'm happy with this.

                                Ned


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