Eric Hanson

I built trivy from this PR and tried running it locally. It seemed to work well on filesystems, but panic'd when I ran it on a docker image: ```julia >trivy...

Sure, I can reproduce by creating a Dockerfile with just: ```dockerfile FROM julia:latest WORKDIR /env RUN julia --project=. -e 'using Pkg; Pkg.add("Example")' ``` Then running `mwe.sh` which contains: ```sh #!/usr/bin/env...

I do think there is some issue here, because if I install release trivy (`brew install trivy`) and change my shell script invoke that one, then there's no panic and...

just to say, this also happens on 1.8.3: ``` Precompiling project... Progress [==> ] 3/70 ◒ Parsers ◑ Colors ◐ Interpolations ◒ HTTP ✓ Glib_jll ◑ GeometryBasics ^Cfatal: error thrown...

## Case 1: explicit import of non-public names like `using Foo: private` yep, that should be possible! IIRC Base has an api for querying if a given name is public...

I think this was fixed, works for me now: ```julia julia> @memoize LRU{Tuple{Any,Any},Any}(maxsize=2) function x(a, b) println("Running") a + b end x (generic function with 1 method) julia> x(1,2) Running...

Following that trail, I get to https://github.com/JuliaGPU/CUDA.jl/blob/0c5bd736f91877c3dfac1d08af3448bd08733d00/lib/cudnn/activation.jl#L32 (upstream docs: https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnNanPropagation_t) where it looks like the default is to not propagate NaNs. Do we want to use another option for Flux?...

On second thought, I think CompatHelper.jl will propose compat bounds for all Project's in a repo though, so this isn't actually super important if you also use CompatHelper, since you'll...

> Regarding the name, maybe `@require`? I thought that could be confusing since `Base.require` is very different, but that's not a public API anyway so maybe it's fine? `@require` does...

I started working on tests and realized the outlined function in my code only works if the msg you pass is a string, not an exception or even arbitrary code...