RE: [exim] Testing what ports are open on the sending host.

--On March 7, 2005 14:59:20 -0500 Eli <eli-list@???> wrote:

> Ian wrote:
>> There's a clue in the grammar here. Eli said "my filter
>> *is*", and you said "your filter *will be* nothing of the sort".
>> So, he's talking about something that he's implemented, and you're
>> talking about something that he's proposing.
>
> I am Eli, and I said the latter quote so I'm assuming you meant
> Marc said "my filter is".

Ah, sorry. Yes, that's right.

> Assuming that he'd either implemented it already, or was
> merely thinking about it (his initial post was about a buddy who was
> mentioning the idea to him, which is where I took the interpretation of
> him either testing or thinking about putting it in to production) makes no
> difference to me.

The difference is that he was claiming that his current filter *is* very
good, but he wants to improve it - not that his proposed host profiling
*would be* very good on its own. Either because he's not satisfied with
whatever errors it makes, or just to keep ahead of the game.

> I'm just trying to point out that I think the test
> would be rather worthless overall - mixed with all the other spam/ham
> checks he's mentioned before I think this wouldn't attribute much
> to the
> final outcome to make it worth his while.

Well, the test on its own isn't very predictive - we're all aware of that.
The question is, can it be used in conjunction with other tests to improve
their results. The technical contributions to the discussion DO suggest
that the contribution will be small, but they don't prove that it will be
negligible. That's an empirical question - we'll only ever know if someone
tries it. And each one of us will only know how it works for us if we try
it ourselves: it seems like most of us won't, unless good results are
reported.

> Just because I say he shouldn't do it though doesn't mean he shouldn't -
> he can do what he likes. Marc throws out many ideas to the list, and
> although he says he doesn't want to get flamed, he should notice that
> it's not so easy to post radical ideas and not brew something up.

Yes, I guess that's true. I'm going to change my tune here a bit, and say
that the discussion has thrown up a lot of useful cautions [cache results,
be careful about interpretation], as well as some spurious objections - and
overall the signal to noise ratio has been good.

> Heck,my grammar just got put in to question and look what it started!
>
> Eli.
>
>



--
Ian Eiloart
Servers Team
Sussex University ITS