xtensor-stack / xtensor-julia Goto Github PK

Or inherit Julia info from the JlCxx target.

Hi,
I was wondering if we could talk about packaging for julia inside the conda-forge framework.
@mkitti and I from the julia-feedstock are trying to figure out what people really want and how we can help.
We recently did some housekeeping and fixes and we would like your input.

In your readme here: https://github.com/xtensor-stack/xtensor-julia#installation-of-the-julia-package, there are some typos. It should be conda install xtensor-julia -c conda-forge

The assignment operation, i.e.,

xt::jltensor<double, 1> test(xt::jltensor<double, 1> u) {
  xt::jltensor<double, 1> f = u;
  return f;
}

cannot be performed in-place through

void test(xt::jltensor<double, 1> f, xt::jltensor<double, 1> u) {
  f = u;
}

This is understandable since the argument is value instead of the reference.
Is there a workaround for the in-place mutation?

edit: an explicit for-loop can do this

void test1([T](xt::jltensor<double, 1> f, xt::jltensor<double, 1> u) { 
  int nu = u.shape(0);
  for (int i = 0; i < nu; i++) {
    f[i] = u[i];
  }
}

But is there an easier way?