Oleg Bartunov <oleg@sai.msu.su> writes:
> I'm also wonder if we're on the right way, since the only serious
> issue with indexscan was possible problem with slaves, but read-only slaves
> delayed to 8.5, so this is not an issue now. In 8.5 development cycle we'll
> certainly return to this issue, so why do we disable index scan for 8.4 ?
So that we have a trouble-free feature in 8.4. I have no confidence
in the solution that was proposed, and am more than willing to accept
the loss of plain-indexscan support to avoid the risk of worse bugs.
regards, tom lane