https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6880

--- Comment #7 from Adam Katz <[email protected]> ---
(In reply to comment #2)
> That code is broken somehow. The following rules from my sandbox _do_ have
> descriptions in the base language: [...]

Sorry, I hadn't run   make test   on that machine in a while, so some
descriptions hadn't propagated.  Nothing was wrong with the aforementioned code
block, just the file tree it evaluated.  While at this, I've refined the test
with a lower threshold and to note high scores ("scale" is spam% + 2*ham% +
abs(score)/3, printing all twenty rules exceeding 1.8):

$ svn update trunk
At revision 1429166.
$ make test
[...]
$ awk '$1 == "describe" { print $2 }' trunk/rules/*.cf |grep -Fwvf-
results/freqs |awk '{ bad = sprintf("%.3f", $2 + $3*2 + sqrt($6^2)/3 ) } NR ==
1 { bad = "scale" } bad > 1.8 && $7 !~ /^[T_]_|\(/ { printf "%6s %8s %8s %6s
%5s %5s  %s\n", bad,$2,$3,$4,$5,$6,$7 }' |sort -r

 scale    SPAM%     HAM%    S/O  RANK SCORE  NAME
 7.739   6.0396   0.8499  0.877  0.62  0.00  TVD_SPACE_RATIO
 6.887   2.8959   1.9120  0.602  0.48  0.50  SUBJECT_NEEDS_ENCODING
 6.801   5.6883   0.0396  0.993  0.88  3.10  STOX_REPLY_TYPE_WITHOUT_QUOTES
 6.483   5.6896   0.0801  0.986  0.83  1.90  STOX_REPLY_TYPE
 6.454   5.5854   0.0010  1.000  0.95  2.60  AXB_XMAILER_MIMEOLE_OL_024C2
 6.298   5.8823   0.0660  0.989  0.85  0.85  FSL_NEW_HELO_USER
 5.840   5.5085   0.0007  1.000  0.95  0.99  MSOE_MID_WRONG_CASE
 4.192   2.8809   0.0055  0.998  0.88  3.90  AXB_XMAILER_MIMEOLE_OL_7EB15
 4.007   1.4442   1.2812  0.530  0.43  0.00  TVD_RCVD_SPACE_BRACKET
 3.709   2.4308   0.0058  0.998  0.86  3.80  KB_DATE_CONTAINS_TAB
 3.242   2.4281   0.0021  0.999  0.86  2.43  KB_FAKED_THE_BAT
 2.706   1.4428   0.0316  0.979  0.75  3.60  AXB_XMAILER_MIMEOLE_OL_B054A
 2.473   1.6253   0.0237  0.986  0.78  2.40  REPLYTO_WITHOUT_TO_CC
 2.179   1.9515   0.0206  0.990  0.80  0.56  HTTPS_HTTP_MISMATCH
 1.954   1.2563   0.0024  0.998  0.78  2.08  HK_SCAM_N2
 1.936   0.5018   0.3240  0.608  0.43  2.36  FSL_HELO_NON_FQDN_1
 1.884   1.5858   0.0310  0.981  0.76  0.71  HK_LOTTO
 1.879   0.5114   0.0003  0.999  0.69  4.10  KB_RATWARE_MSGID
 1.821   1.5123   0.0045  0.997  0.80  0.90  AXB_XMAILER_MIMEOLE_OL_4379D
 1.809   1.2603   0.0244  0.981  0.75  1.50  HK_NAME_FM_MR_MRS


Callouts to authors: felicity (3), kb (3), axb (4), maddoc (2), jm (5), hege
(3), as located by this command, which spits out the rule definitions and
sandbox paths (output not presented here):

$ awk '$1 == "describe" { print $2 }' trunk/rules/*.cf |grep -Fwvf-
results/freqs |awk '$2 + $3*2 + sqrt($6^2)/3 > 1.8 && $7 !~ /^[T_]_|\(/ { print
"^ *[^# ps]\w+\s+" $7 "\s" }' |grep -Erf- trunk/r* |grep -wv 72_active.cf

(and   svn annotate trunk/rules/20_body_tests.cf |grep -wC3 TVD_SPACE_RATIO   )

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to