On 2/24/2013 9:05 AM, Timon Gehr wrote: > For the moment I'm just sticking with 2.060, because I have failed to reduce all > the forward reference regressions introduced in 2.061. I've found dustmite to be pretty helpful reducing things down.