http://bugzilla.spamassassin.org/show_bug.cgi?id=4400
------- Additional Comments From [EMAIL PROTECTED] 2005-06-13 13:30 ------- (In reply to comment #5) > Since the Pg code has been heavily benchmarked and tweaked accordingly, > I'm pretty certain I won't make this change without similar benchmarks > to back it up. Please see: > http://wiki.apache.org/spamassassin/BayesBenchmark Your benchmark has at least one issue that I can see---unless I missed something, all entries you put in the test database will all have the same ID. This seems unlikely to be the case in most real installations, and the selectivity of the id column could have implications for PostgreSQL's planner's choice of index for queries of the form "where id = ? and token in ()". OTOH, I don't know that it's likely enough to be worth worrying about. ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
