Claire Foster

@michaellindon the `destructure()` function may already do what you want (apologies for the lack of documentation visible on the web. There is some docs available in the repl.). Demo: ```...

Your problem is that the list comprehension has failed type inference, so `v` is an array of `Any`. Eventually I'd like to make `destructure()` into a more general tool (eventually...

By the way, you can definitely change the values inside your `m=Array{Vec{2,Float64},1}()`, you just need to replace whole elements at once rather than individual components: ``` julia m = Array{Vec{2,Float64},1}()...

That's rather a lot of symbols to digest, and it's a bit hard to diagnose without any type information for the inputs to FFBS. To focus our attention, could we...

Ah, now how did I not know about `I`, that's very nice! However... `UniformScaling` doesn't seem to interoperate with `Mat` yet. Ah you added it just the other day... so...

@mschauer thanks for making that PR with randn, I've been away on holiday for the last couple of weeks or so. @SimonDanisch I agree with Tim regarding size checks. IMO...

Alas, as expected v0.5 has changed Expr enough that this is broken there. Needs more work.

I just ran into this as well. We should definitely return a fixed size array for slices, at the moment a tuple comes out: ``` julia julia> Vec(1,2,3)[1:2] (1,2) ```

Good point about the type problems. The tuple approach seems like the best thing we can do at the moment IMHO, though for anything larger than 3 or 4-vectors, the...

Yes, to me it looks like you're implicitly creating a kdtree during your splitting process. Whether that's actually useful is something I don't know - ray tracing traversal is a...