Simon Riggs <[EMAIL PROTECTED]> writes: > Looks good. I think it explains a few other wierd perf reports also.
Could be. I went back to look at Sam Mason's report about three weeks ago, and it definitely seems to explain his issue. The "fuzzy cost comparison" logic is new in 8.0 so it hasn't had all that much testing... regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq