On Wed, 22 May 2002 the voices made Craig R Hughes write:

> Tony L. Svanstrom wrote:
>
> TLS>  I know it's GA-evolved, but isn't the score (2.741) for SUBJ_HAS_SPACES
> TLS> uncalled for when we're just talking about "linebreakingspaces", indenting the
> TLS> continued subject-line...?
>
> TLS> X-Spam-Status: No, hits=3.3 required=4.5 tests=SUBJ_HAS_SPACES,
>                     ^^^^^^^^^^^^^^^^^^^^^^^^^
>
> SpamAssassin notches up another true-negative.  I don't see how your example
> here is an argument for changing anything -- SA is working as intended.

 I seem to just love looking like a fool on this ML; and after reading RFC
(2)822 I see that strictly speaking SA is doing the right thing ("Unfolding is
accomplished by simply removing any CRLF that is immediately followed by
WSP.").

 IRL, however, I see unfolding done in a s/(\n\s*(?=\s\S)/\ /g-ish way (take a
look at A.5. in 2822); and I'm getting e-mails where the subject's been folded
with added whitespace (basically indenting the second line the length of the
name of the header + 2).
 AFAIK these added whitespaces are not shown in most mailreaders.

 Based on a blurry "I'm pretty sure"-ish feeling I can't check due to me
messing with all the headers before they're saved, I think that the
SUBJ_HAS_SPACES rule would benefit from ignoring "indenting" WSP
("/(?<=\S)(?:\s{4,}|\t)/"?).


 Would it be possible to set up an environment where you can test changed/new
rules against a set number of spam and non-spam, preferably the same ones used
for the current GA-scores?


        /Tony
-- 
# Per scientiam ad libertatem! // Through knowledge towards freedom! #
# Genom kunskap mot frihet! =*= (c) 1999-2002 [EMAIL PROTECTED] =*= #
-- Random URL (6/8):
<URL: http://www.macslash.org/ > When http://apple.slashdot.org/ isn't enough.





_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to