https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6582

--- Comment #7 from Henrik Krohns <[email protected]> 2011-05-08 12:56:16 UTC ---
First tryout with 1000h/1000s corpus, making rendered() return LIMIT bytes.

LIMIT  SCANTIME
orig   197s
10000  127s (-35%)
20000  134s (-32%)
30000  137s (-30%)
50000  144s (-27%)
100000 159s (-19%)

Only thing it affected was ~30 (few less for 10k, few more for 100k) identical
chinese (GB2312) trap spams, which were all 200-450k in size. Pretty much only
FREEMAIL_REPLY and FUZZY_XPILL were the rules that changed.. obviously we don't
have many chinese rules.. the scores dropped from ~4.7 score to 0-2.

Now I'm going to run my full old corpus (63834h/46929s) with 30000 and 100000
limits to see how they do in detail.

-- 
Configure bugmail: 
https://issues.apache.org/SpamAssassin/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

Reply via email to