Re: [discuss] 2.6.19-rc6: known regressions (v4)

> A slightly smarter, but not quite as obviously correct, 

I think it's better to go for the "obviously correct" approach right now
And sorting multiple times should be fine

-Andi