Bruce Momjian <pgman@candle.pha.pa.us> writes:
> I assume that for a vacuum that only hit pages indicated in the bitmap,
> it would still be necessary to do an index scan to remove the heap
> pointers in the index, right?
Given the current vacuum technology, yes. However, bearing in mind that
indexes should generally be much smaller than their tables, cutting down
the table traversal is certainly the first-order problem. (See also
discussion with Simon from today.)
regards, tom lane