Where can read about SA's FARAWAY tags

2007-11-26 Thread reader
Where can I read about how SA arrives at the FARAWAY tags.

Its not always from charset headers or two letter country indicators
in header address's.  At least I think I'm seeing some mail that
contains neither but SA has still correctly identified them.



Re: How to make sa use just specific tests

2007-11-18 Thread reader
Per Jessen <[EMAIL PROTECTED]> writes:

> [EMAIL PROTECTED] wrote:
>
>> Per Jessen <[EMAIL PROTECTED]> writes:
>> 
>>> [EMAIL PROTECTED] wrote:
>>>
 Summary:
 How can I set things up so that sa runs only certain specific tests
 on incoming mail?
>>>
>>> Just load a single config file with those tests defined.
>> 
>> The `test' that creates FARAWAY tags is the language choice config
>> setting:  (for me ok_locales en)
>>   
>> You mean to leave /etc/mail/spamassassin blank and put
>>ok_locales en
>> 
>> ~/.spamassassin ?
>
> I don't know how you call SA, but I use spamd with a siteconfigpath.
> Using that you can point SA to any config you desire.  I'm sure can
> come up with something similar if you're using procmail.
>
>> But I want to call in full default SA testing later on.  So would I
>> need  a second local conf or what?
>
> Yes.  If you want two different sets of tests, you'll need two different
> configs.

OS=Gentoo - kernel 2.6.21
spamassassin-3.2.3
procmail-3.22

Ok, I'm sorry to be such a dimwit here and about my windy response
below, but delving into all the manpages for spamc spamd spamassassin
procmail (has 5 man pages)

I'm getting really confused.  I've been calling spamd thru spamc in
procmailrc for years and have experimented somewhat with my own SA
rules (years ago now) but recent versions are so good I just use it in
default mode with a few config setting in
/etc/mail/spamassassin/local.cf.

But spam has gotten so intense it is taking more and more processing
time to run SA.  So I wanted to cut down on SAs' work load. 

/etc/mail/spamassassin/local.cf

 report_safe 0
 ok_locales en
 use_pyzor  0
 skip_rbl_checks 1
 dns_available no 

The spamd man page doesn't seem to really explain where spamd gets
user settings from or it is so complicated I'm missing it.
For example:  --siteconfigpath=path Path for site configs

Appears to mean a directory where one puts a collection of files but no
indication of how a specific file is read.

No one uses path to mean path and file in my experience.  That is
usually called an `absolute file name'.  If it meant file I think it
would or should say so.

On a gentoo system one has an /etc/conf.d/spamd file to specifiy how
it is called system wide.  I have root on this system so can set it
there I guess.  But not sure what settings I would want there.  Or how
to use one setting and then another from a site wide location.

Does --siteconfigpath=path  really mean siteconfigpath=/some/path/file?

Further, examples I used originally to setup SA and procmail long
ago, suggested this method in promailrc:

  :0fw
  | /usr/bin/spamc

  :0:
  * ^X-Spam-Status: Yes
  spam.sa_spam1.in

And that is how I've done it for years.

Do I need to start calling spamd directly?

man spamc says spamc spools the message to spamd but there appears no
way to send spamd a config choice thru spamc. 

I guess what I want to do boils down to lightening the load on
spamassin by calling SA thru spamc/spamd but only to use whatever it
does when `ok_locales en' is set, 

Then pass any mail that gets past that point thru several of my own
procmail recipes and finally call SA again when a fair bit of the mail
has already been dispatched. Something like the scheme below 
(necessary steps to specify config are excuded since I'm not sure how):

~/.procmailrc

## Use what ever technique is required so that SA only reads and
## filters with `ok_locales en' and nothing else.  Possibly the
## shortcircuit plugin but not at all clear how one would shortcircuit
## after using `ok_locales en'

   :0fw
  | /usr/bin/spamc

  :0:
  * ^X-Spam-.*FARAWAY
  spam.ForeignBunkem.in

[several more procmail recipes of my own design.  Maybe 8-10 ]

## The amount of mail is now down to half or less
## This time use full spamassassin in default config

  :0fw
  | /usr/bin/spamc

  :0:
  * ^X-Spam-Status: Yes 
  spam.sa_spam1.in

  :0
  WhateverIsLeft.in

So to get that to happen it is not at all clear how to call SA in each case.



Re: How to make sa use just specific tests

2007-11-17 Thread reader
Per Jessen <[EMAIL PROTECTED]> writes:

> [EMAIL PROTECTED] wrote:
>
>> Summary:
>> How can I set things up so that sa runs only certain specific tests on
>> incoming mail?
>
> Just load a single config file with those tests defined. 

The `test' that creates FARAWAY tags is the language choice config
setting:  (for me ok_locales en)
  

You mean to leave /etc/mail/spamassassin blank and put 
   ok_locales en

~/.spamassassin ?

But I want to call in full default SA testing later on.  So would I
need  a second local conf or what?



Re: How to make sa use just specific tests

2007-11-17 Thread reader
Theo Van Dinter <[EMAIL PROTECTED]> writes:

> On Sat, Nov 17, 2007 at 10:16:12AM -0600, [EMAIL PROTECTED] wrote:
>> Summary:
>> How can I set things up so that sa runs only certain specific tests on
>> incoming mail?
>
> What you really want to do is run all the tests, but run some first and stop
> if they hit.  aka: Short Circuiting.  There's a plugin in 3.2 which lets you
> do this.

Yes it is and thanks... one further question to save a little time
spent testing...

The test that causes the FARAWAY headers that I want to look for, near
as I can tell is really just this: 

   ok_locales en

To me, this doesn't appear to be like other tests.  It is more a
configuration.

The documentation for the short circuit plugin appears to be based
around actual tests.

So can you tell me would I simply put something like:

 cat local.cf

 shortcircuit ok_locales en on

I doubt that will do...




How to make sa use just specific tests

2007-11-17 Thread reader
Summary:
How can I set things up so that sa runs only certain specific tests on
incoming mail?

Details:
I'd like to run only tests that produces the TAG with FARAWAY in it
(there are several) at one point near the top of  .procmailrc.
Dispose of that mail with a test next in line to /dev/null and only
then run the full set of SA tests somewhere futher down after several
local procmail tests.

I'm thinking that would lighten the load on SA and in general decrease
mail processing time by a significant factor.

In my experience, I see many SA tags on foreign mail along with the
FARAWAY tags, but I have never seen mail with FARAWAY tag that I want.

However that mail makes up something like 25/30 percent of incoming
mail.  So the other tests are running for no good reason on 25/30
percent of incoming.

But It is possible, even likely, that I do not really understand the
process of SA and procmail working together.



CHARSET_FARAWAY_(HEADER|MIME)

2007-11-12 Thread reader
I often see these:
   CHARSET_FARAWAY_HEADER
   CHARSET_FARAWAY_MIME

On foreign mail (In my case anything not in english).  But sometimes
the Charset is not mentioned in Content-Type or in the Mime headers.

How does spamassain determine the Charset used?