Luca Palmieri

This is not something I have time to follow right now.

That could definitely work! I hadn't realised there was an `Advice` severity level 😅 I'll experiment and report back.

It depends on how far-reaching are our plans for `LinalgScalar`. It can be quite convenient to be able to divide and subtract when implementing linear algebra routines that go beyond...

Looking at this purely from an ecosystem integration perspective, it would be amazing! It would make it seamless to interoperate Rust code using `ndarray` with systems developed in other language...

It looks cleaner, which is always good API-wise, but I have to admit that the whole dimension-shape situation is a little big foggy to me (especially since reading #367, I...

I was rereading https://github.com/rust-ndarray/ndarray/issues/367 and I think, even though there might some technical difficulties that will have to be ironed out, that what @termoshtt was proposing makes intuitive sense from...

I like @jturner314's design - wouldn't it be better though to use two separate traits for Shapes and Strides? (even if they might offer exactly the same methods initially) Currently...

> In the community there is some shift in preferences and recently some people favor [eyre](https://lib.rs/crates/eyre) over anyhow. This claim needs a source. `eyre`'s download count is extremely stable: https://crates.io/crates/eyre...

> As a "library-level" crate, we should use more specific error enums if possible. That error variant must be opaque. The only alternative, here, is to define a custom opaque...

> > This claim needs a source > > https://blessed.rs/crates (scroll down to “Error Handling”) That's someone opinion—they're entitled to it, but blessed.rs has no official sanctioning whatsoever. Even the...