Alvaro Herrera <alvherre@xxxxxxxxxxxxxxxxx> writes:
> Maybe we could load it in a more compact form after the heap cleanup
> pass, instead of a plain TID list.
Hmm ... a non-lossy bitmap would be an interesting alternative.
I think I went with bsearch mainly because I knew I could generate
the TID list already sorted. Somebody should try to measure the
probe time of the tidbitmap code against bsearch ...
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faq
