http://bugzilla.spamassassin.org/show_bug.cgi?id=3538
[EMAIL PROTECTED] changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|major |normal
------- Additional Comments From [EMAIL PROTECTED] 2004-07-06 16:50 -------
lowering severity to normal
I tested 500 spam and 500 ham using the current tree as a control and the
one line workaround included above. Each was tested with all tests (no
Bayes or AWL, but with DCC, Pyzor, and Razor plus DNSBL tests) as a baseline
and with only URIBL tests as a worst case for the timing bug. DNS cache
was cleared before each run and -j 8 was used. Three runs of each:
count of messages with one or more URIBL hits:
ham-old-ALL 4 4 4
ham-old-URIBL 2 3 4
ham-new-ALL 4 4 4
ham-new-URIBL 4 4 4
spam-old-ALL- 433 434 432
spam-old-URIBL 375 402 407
spam-new-ALL 433 433 430
spam-new-URIBL 402 402 405
The lack of a patch was only significant in one of three runs (the 375).
Total URIBL hits across all three runs:
ham-old-ALL 30
ham-old-URIBL 16
ham-new-ALL 30
ham-new-URIBL 25
spam-old-ALL 3527
spam-old-URIBL 2971
spam-new-ALL 3523
spam-new-URIBL 2834
So, yes, it helps, but it's only an increase of about 5% in the hit rate.
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
