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

Re: mailfilter



 ******  Edmund GRIMLEY EVANS = E. 
 
>E.: ĸ <lfpor@lujz.org>:
>E.: 
>E.: > laŭ mi legis en:
>E.: > http://sourceforge.net/mailarchive/message.php?msg_id=6161460
>E.: 
>E.: Tiu TTT-adreso ne plu funkcias.

 Strange, por mi ĝi funkcias... Mi sendas anekse ĝin.
 
>E.: Tio dependas de la aliro. Kun IMAP eblas demandi pri la MIME-strukturo
>E.: de mesaĝoj. Kun POP3 eblas nur peti la ĉapaĵon kaj la unuajn X liniojn
>E.: de la korpo. Eblus peti 50 liniojn kaj kontroli, ĉu tia aldonaĵo

 Nu, mia estas per pop3. Do mi provos vian programon, kaj eble petos ion
 de servilo, eble kiel if.ufrgs.br, ke hostsul.com.br instalu spamassassin.

dankon al vi kaj MJ Ray, 

Luiz
Email Archive: mailfilter-users (read-only)

From: Ron Johnson <ron@co...>
Re: 4 suggestions for simple improvements to MailFilter  
2003-09-29 09:58

 On Mon, 2003-09-29 at 10:23, Sundial Services wrote:
 > MailFilter is, of course, priceless for us right now.  I'd like to make 3 very 
 > simple suggestions for improvements that I think would make it vastly more 
 > useful to us:
 > 
 > [1]  Allow the 'n' of 'TOP messagenumber n' to be user-specifiable.  Currently 
 > it appears to use '0' but many more messages could be filtered if this were a 
 > user-specifiable option.  '25' or '30' would allow many more junk messages to 
 > be detected.
 > 
 > [2]  One obvious characteristic of spam-mail, at least for us, is that 
 > spammers always send great numbers of nearly-identically-sized messages and 
 > do so all-at-once.  Therefore, if two messages are identically-sized (or I 
 > would prefer, "identical +/- N bytes") and they don't come from a friend, 
 > they're spam.  Let me delete all of them or choose to keep Z copies.  (Z is 
 > user-specifiable; default 1.)
 > 
 > [3]  It might be much more complicated, but it would be nice, if the program 
 > could read the text of messages and compare the number of words they have in 
 > common in the first section.  I envision that this algorithm would parse the 
 > message text for words, promoting words seen more than once and eventually 
 > dropping words seen only once, keeping only so-many words in the pool.  Then 
 > go back and develop a signature based on word occurrence and count, and throw 
 > out dupes.
 
 Then you'd have SpamAssassin inside mailfilter.
 
 > [4]  Attachments!  If an attachment matching a certain pattern like "*.EXE" or 
 > "*.SCR" or "*.PHP" could be seen, then it's junk.
 
 The problem in that the content descriptors for such attachments
 may be *many* (like 200-300) lines into the body of the message,
 and to read so much, while only a fraction of a 150KB swen mail,
 is *all* of most emails, which means that you'd read each email
 twice: once by mailfilter, and once by the process that "really"
 gets the whole email.
 
 And on a disl-up line, that would just about defeat the purpose
 of mailfilter, it seems to me.
 
 -- 
 -----------------------------------------------------------------
 Ron Johnson, Jr. ron.l.johnson@co...
 Jefferson, LA USA
 
 "As the night fall does not come at once, neither does 
 oppression. It is in such twilight that we must all be aware of 
 change in the air - however slight - lest we become unwitting 
 victims of the darkness."
 Justice William O. Douglas
 


 

Thread View
Thread 	Author 	Date
4 suggestions for simple improvements to MailFilter	Sundial Services <miker@su...> 	2003-09-29 08:30
      Re: 4 suggestions for simple improvements to MailFilter	Ron Johnson <ron@co...> 	2003-09-29 09:58
            Re: 4 suggestions for simple improvements to MailFilter	Douglas A. Augusto <daaugusto@ya...> 	2003-09-29 17:20
                  Re: 4 suggestions for simple improvements to MailFilter	Til Schubbe <t@gm...> 	2003-09-29 21:41

Reply to: