I've update the patch to fit the current master branch. I wasn't however able to modify the diff submission. I therefore created a new one: D5078.
Otherwise I've additionally attached the new patch to this message.

Great. I think then 0.1 should be used as mentioned in the paper.
I'd be quite happy with the final solution as mentioned by Daniel Santana =).

This should probably be evaluated in the following sense and for the following reason:

In D5062#115123, @Daniel Santana (dgsantana) wrote:

This should give the same result as the F1442978, since the template function will expand at compile time to the Matrix3f version, which is the what is proposed in the diff, there should be no difference in speed between the 2.