Comment 9 for bug 590708

We will likely want to review a lot of our schemas as we improve performance, but I think the sort of tradeoff you are suggesting is reasonable in the short term.

The count seems totally waste to me in the case that the returned content is < one page, so I think we're going to want to fix that.

However, dropping down to 250ms + 2000ms would be awesome.

Care to work up a patch to make it query in this fashion?