Nathan Daly

So I just want to leave a status update here. So I think this basically works. The benchmarks run (and after the merged-changes in https://github.com/JuliaCI/PkgBenchmark.jl/pull/75, they should correctly and precisely...

Hey @ericforgy, I'm glad you pointed this out! I agree it seems like probably surprising behavior, but I'm not too sure how best to proceed either. It's certainly surprising behavior...

Sorry for the delay @EricForgy. Thanks again for bringing this up. Just adding some more thoughts: > If this is an intentional design choice that has been discussed / vetted...

**Excellent** news! We can now close this PR because Julia+LLVM now do this optimization themselves, automatically! 🎉 Since Julia 1.6, when `*(::FD{Int64}, ::FD{Int64})` took ~8ns, now it takes 0.8 ns,...

Sure, makes sense. Cool! Thanks! On Jun 26, 2017 9:28 AM, "Bart Janssens" wrote: > True, that would be more consistent. I did discuss with @shashi > about issue #43...

One other thing to consider: if tests allow grouping within a file, along the discussion in https://github.com/julia-vscode/TestItemRunner.jl/issues/5, the testsetup could be per-test-group.

CC: @nickrobinson251

+1 for doing the nesting by directory/file structure. this maps to what they do in golang, and that seems to work well enough for them! :)

I'm ambivalent on the sub-file nesting, but it does seem like it could be nice, yeah! Either by having yet another grouping macro, or using nested names as you suggested.

interesting. makes sense! No matter what else we do to investigate this, can we eliminate whatever caching is going on in the video that's causing it to _run out of...