Olivier Gauthé

I have also been annoyed by the changing index dtype. I would favor switching to `np.int64` as default, since having a fix dtype everywhere is much simpler. In my case,...

Hi! I tried using `pre-commit` with @FeryET repo. I did not manage to run the hooks with `rust` as language: `pre-commit` could not locate a `Cargo.toml` in the _commit hook...

I confirm everything is fine with the latest release. Thank you!

Hello, I just ran into this issue with numba `0.53.1+8.gddf817af0.dirty`. As @dizcza, I am using `prange` with `njit(parallel=True, cache=True)`, no `ThreadPoolExecutor` or `ProcessPoolExecutor`. The first run, which call the compiler,...

new feature request: `Base.:*(::BlockSparseArray, x::Number)` and `Base.:/(::BlockSparseArray, x::Number)` I updated the first comment.

new issue: `1im * a` does not modify `a` data type if `a` is empty. It crashes if `a` contains data.

new issue: `similar(a::BlockSparseArray, eltype::type)` do not set new `eltype`. `similar(a::BlockSparseArray, eltype::type, size)` behavior depends on `size` type. - For `size::NTuple{N,Int}`, it returns a dense julia `Array` with correct `eltype`. -...

Feature request: `display(::Adjoint)`. The function `Base.adjoint` is well defined and returns an `Adjoint` type, but it throws an error at display. Closely related, it is not clear to me how...

The solution to accept any `AbstractUnitRange` is to replace the definition of `Axes` with ```julia Axes

I continue in exploring the effect of `dual`. `eachindex` is broken on such an array. This affects `==`. ``` julia> eachindex(m1) ERROR: MethodError: no method matching AbstractUnitRange{…}(::NDTensors.GradedAxes.UnitRangeDual{…}) Closest candidates are:...