Alexis Montoison

I suggest you to use macOS Intel binary for the moment, many artifacts are not available for M1 architecture too. It requires to recompile and potentially adapt the build scripts...

@dpo I rebased #244 and we have the same error. `ARM64 - linux` is Tier 2 so I don't think that the errors are related to Julia. The operations have...

@giordano I will compile the last release of SCOTCH 6.1.x before this PR, can I do multiple folders (SCOTCH@6, SCOTCH@7) like MUMPS or OpenBLAS to keep both `build_tarball.jl` ?

@sshin23 I updated HSL.jl to drop `BinaryProvider` if it can help you: https://github.com/JuliaSmoothOptimizers/HSL.jl/pull/112

I can work on the support OF quadratic constraints in `NLPModelsJuMP.jl` if we agree on the storage for them.

@ViralBShah It's relevant to have a random `v0`. It should be related to the initial vector of the Arnoldi process. We use it to generate the Krylov subspace and the...

@SobhanMP I will recompile a new NOMAD_jll as soon as I can.

> Would it be a simple pull request for https://github.com/JuliaPackaging/Yggdrasil/blob/b63bb8b3553d614d20a95137528de1ae506fbef8/N/NOMAD/build_tarballs.jl to the latest git hash? No, because the headers are generated dynamically and I need to generate a new patch...

@jpsamaroo off topic: Do you know if it's possible to add a generic fallback for `reflect!` in `GPUArrays.jl`? I don't know how to implement it without a kernel.

> While it doesn't necessarily hurt (and I won't block merging on switching them back), we don't need the `T` type parameter exposed from most of these functions. Sorry, I...