Kristoffer Carlsson

The reason for the strange error is due to faulty `@inbounds` usage. ``` julia> push!(v, 23) ERROR: BoundsError: attempt to access 0-element Vector{Any} at index [1] Stacktrace: [1] setindex! @...

I did not see this conversation before I put my W.I.P PR version of the VTK exporter for MeshIO. I am fine if mine gets replaced if you decide to...

Okay. Btw, I saw your hack for putting raw binary into the file. Nice :). It is quite annoying that you can't just write_binary or something but then again XML...

Works fine for me: ```jl julia> for i in 1:3 print(i, ": ") @btime $(Symbol(:func, i))($(Ref(12345))[]) end 1: 1.214 ns (0 allocations: 0 bytes) 2: 1.017 ns (0 allocations: 0...

`@btime` also does GC scrubbing between samples which you probably don't want for this?

Seconds is too big of a unit to default to. Just print nanoseconds by default and have an option to print in another unit.

I think nanoseconds are better because an operation takes on the order of nanoseconds. I think we can print `ns` instead of the full thing, especially when it is the...

Google benchmark prints ns https://github.com/google/benchmark#counter-reporting. ``` ------------------------------------------------------------------------------ Benchmark Time CPU Iterations UserCounters... ------------------------------------------------------------------------------ BM_UserCounter/threads:8 2248 ns 10277 ns 68808 Bar=16 Bat=40 Baz=24 Foo=8 BM_UserCounter/threads:1 9797 ns 9788 ns 71523 Bar=2...

> I would rather print leading zeros up through the hundred seconds place, e.g.: Plz no. I would say 90-95% of the benchmarks I do are under 1000 ns. Having...

Just a note that #28 fixes a problem we had where some functions were being completely elided. Was just going to file an issue and then it is already fixed....