Always Learning wrote:
> Peter Bowyer wrote on Fri, 23 Apr 2010 18:23:57 +0100.
>
>> Discarding the countless man-years of effort already expended on this
>> topic would seem to be foolish. See Dynastop, Spamhaus PBL, and many
>> other projects. They've taken the time to get it right and think of
>> all the instances that your simplistic pattern matching falls over on.
>
> 1. One will never ever learn much simply by adopting other people's
> solutions.
Perhaps not. But it works fairly well at avoiding pain or fatalities.
In about 26 countries now, I've always tried to at least drive on the same side
of the road as the 'locals' do.
>
> 2. Yes the implementation of what I perceived to be Regular Expressions
> failed in my Exim 4.63 (no later versions available for my Linux
> distribution). Conducting a test in Perl, the same as Martin did,
> produces the same non-match so the inevitable question is why did the
> same criteria produce a radically different answer in Exim if both are
> using the same Regular Expressions ?
>
> 3. I prefer to get the filtering working satisfactorily in Exim's ACL
> rather than to venture into very heavily RAM dependent (as alleged by
> others) Spam Assassin.
Indeed, SA should be your *last* line of defense. But Exim has all the tools
built-in for the several barreirs ahead of it. no need to reinvent all that.
>
> 4. Yes my pattern matching is indeed 'simplistic'. What can one
> realistically expect from a brand new Exim user of just 6 months
> experience of Exim who implemented his 'simplistic' solution with just 3
> months part-time knowledge ?
>
Blood in my mouth from biting my tongue...?
> 5. If anyone can help explain why this gives the wrong result I shall be
> grateful.
>
>>> deny message = [C06.5] Msg6 Msg2
>>> hosts = ^.*[a].?[d].?[s].?[l]*
>
> It matched 'olga.hinterlands.org'
>
> Regards,
>
> Paul.
>
>
No .. as you've just had (graciously) explained by someone kind enough to
[invest|waste] the time, it matched the 'ands':
'olga.hinterlands.org'
^^^^
WHEREAS..
EITHER a built-in Exim rDNS test OR any number or about a hundred and fifteen
available RBL calls, dynamic or otherwise, would have found Martin's server
clean and pristine. MANY of those RBL's would also have blocked the zombots on
dynamically-assigned ISP pools that *seems* to be the actual object of your
original exercise.
It is all well and good to reinvent the wheel if you are learning something from
that.
But you are no longer so doing.
You've now become fixated on perl-as-a-method instead of the nature of the
original problem.
Go Ogle S.I Hayakawa's citation about left-door syndrome or the famous pike and
minnows experiment. That fish just 'knows what he knows', and will starve to
death in the midst of plenty. For 'plenty' read 'this list archives' if not the FM.
That could save you a lot more than 3 months...
Bill
