http://bugzilla.spamassassin.org/show_bug.cgi?id=4260
------- Additional Comments From [EMAIL PROTECTED] 2005-05-20 12:57 ------- Created an attachment (id=2884) --> (http://bugzilla.spamassassin.org/attachment.cgi?id=2884&action=view) Patch to current trunk svn that uses question data as hash for uniqueness Here is a patch thatshould take care of everything other than the current failures in t/dnsbl.t tests. I haven't had a chance to look at that more than to determine that this patch doesn't make it worse or better. That should be dealt with in a separate bug. This patch uses the data from the question section of the packet and the ID as the hash key for ensuring that a packet reply matches the query. It doesn't need to use SHA or anything like that because perl's own hashing takes care of it at a memory cost of storing the query string for each DNS query made. That makes the code simple too. The patch selects a port at random, which should avoid what I think is the cause of most of the collisions we have been seeing, then uses the hash to make it impossible to get a false result even if there is a collision. The purpose of trying to reduce the collisions in the first place is that there is a cost to receiving and discarding the packets. The purpose of hte hash is that we can't completely eliminate collisions, especially since we don't have control over SPF::Query's use of the DNS -- We can get collisions from replies to packets that it sends out. I would like to check this is and then deal with the t/dnsbl.t test failures. My laptop died again and I don't have as good a development and testing environment as usual. So please review and comment on this patch before I check it in. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
