Re: [Evolution] Need some help (I Got IT!!!)



On Wed, 2004-03-31 at 13:10, Tony Earnshaw wrote:
ons, 31.03.2004 kl. 12.07 skrev guenther:

Another reason for being a stick-in-the-mud and keeping to Evo 1.4,
whilst keeping SA ("what version was it again?") on the Postfix snapshot
smtp amavisd-new proxy MTA. *Where it belongs*. And yes, we have
LDAP-based amavisd-new per-user SpamAssassin preferences at this site.

Whose bright idea was this enforced Evo SA innovation? I can see us all
having to go back to Mozilla yet once again :(

Bullshit. Sorry, Tony, but that is plain FUD to me.

Call it "ignorance" if you wish, but FUD? FUD applies to what Microsoft
and SCO spread.

Nope. FUD can apply to other cases, too. And your comments mainly where
spreading fear, uncertainty and doubt.


You can entirely disabling filtering through SpamAssassin in Evolution.
No one is forced to. IMHO it is a better approach in UN*X land to use
applications, that already exist and proved to work very well. SA does
for you. There is no need to implement a Bayesian filter for any mailer
out there.

This I don't understand. One of the strongest assets of SA is a Bayesian
filter.

Yes, I agree with that (smilingly looking at my BAYES_99 ratio ;).

What I meant: There already is a very good spam-fighting product. So why
should every mailer implement their own version of the same thing? The
UN*X way traditionally is re-using already existing good code. And
Evolution follows this path of re-using, instead of implementing yet
another Bayesian classifier.


[...[

Sorry again Tony, I don't want to be offending or confrontational. And I
sure don't want a flame war, but IMHO you where overreacting.

Wouldn't surprise me :) Jeff calls it being "confrontational".

Different wording... ;)


I'm not too happy with SA being used by default, too. Any user should
decide to *enable* it, not disable it. And the additional dependency
isn't fun anyway to me.

Then have in mind that 1.5.x is *development* software. It is not yet
intended for the users out there. Hopefully a proper documentation,
warning and maybe a defensive default (SA disabled) will be part of
Evolution 2.0. (Anna, Aaron?)

SA is highly specialized software, intended for MTAs, not MUAs.

This one is open to discussion.

SA sure should not be running on dedicated MTAs. IMHO automatic
filtering is not a good idea without any user interaction in this case.
I know you are aware of false positives. They can get a PITA for people
dealing with certain circumstances. Having the opportunity to check your
incoming spams is essential. Thus any MTA not at the very end should
never run SA.

Now, how does my fetchmail/procmail/IMAP setup differ from a mail
client? Both fetch mails and process them. In fact, any mailer
implements the basic functionality of fetchmail and procmail, namely
retrieving mails and processing (sorting) them. You don't wanna tell me
(and the lion's share of the SA list subscribers) that we should not
call SA using procmail, do you?


It needs
careful configuring and maintenance on a regular basis, it presumes a
site admin who is capable of understanding and implementing
site-specific Perl regular expressions and is capable of doing 'man' and
running tests on a regular basis.

Maintenance can be an issue, right. But the vast majority should be
happy enough with a default SA install. Heck, not even the SA list
subscribers seem to be able to maintain their own server and keep it up
to date...


SA runs normally as an extension to
MTA server software. It has its own medium-volume mailing list devoted
to the above.

I do know. Yet another mailing list we share... ;)


If Evo people start posting on it, they'll be talking
another language. The Mozilla Bayes filter constitutes a reasonable
subset (Bayes filtering) which actually complement site-wide filtering
and needs no specialist knowledge.

This ain't gonna happen. All SA related questions by Evolution users
will end up here. That's one of my fears.

I have set up SA with default only configuration. In fact, I offered to
customize it. They declined, and they seem to be very happy [1]. The
defaults should suit well enough.

...guenther


[1] The only spoiler that got added way later where some additional net
tests, which improved the rate even more.


-- 
char *t="\10pse\0r\0dtu\0  ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]