Jodi Kanter <jkanter(at)virginia(dot)edu> writes:
> I apologize if this is the wrong list. I have posted explain analyzes below.
Given the small size of the tables, I think the planner is doing the
right thing not to use indexes. Seqscans are probably less I/O until
the tables get a lot bigger.
The bitwise-and tests are not indexable anyway... you might want to
reconsider that data representation, if you were hoping that the bit
conditions would be useful for indexing.
regards, tom lane