Wolf Vollprecht

> Arrays are fully C++ compliant, can be held in standard C++ collections. Can be included in larger structs without having to write custom move constructors / assignment for those...

To give some more motivation in terms of interoperability: So far xtensor supports natively 4 languages – C++, Python, Julia and R in their respective packages: - xtensor-python: https://github.com/QuantStack/xtensor-python -...

Hi, just to reply to your first point: Writing generic functions in xtensor is "easy" -- you'd write something like ``` template auto func(xexpression& e) { auto& de = e.derived_cast();...

> Then I would say xtensor does not yet meet the minimum requirements provided by Fortran 90. It should be entirely natural to do something like: > void f(Array &a);...

@slayoo I don't have a lot of experience with libdynd. we have talked with one of the authors a couple of weeks ago. At the moment there is not much...

Hi @emmenlau we do have a benchmark suite that compares against several other libraries here: https://github.com/wolfv/xtensor-benchmark It might need a little bit of love to get it to compile with...

btw seeing your HighFive contributions, we actually use HighFive as part of `xtensor-io` (github.com/QuantStack/xtensor-io) that one of our core contributors (@tdegeus) has added (and also backported some cool changes to...

Hi @asottile, while I understand your reasoning, I would like to point out that this behavior of conda is a "historical artifact" and something that the Anaconda team is actively...

Unfortunately `conda create` does _not_ work with `environment.yml` files (it does, however, work with plain text files or explicit lock files). On the other hand, `conda env create` _only_ works...

I would love to understand better what bothers you about such a small change? It would be a great help! As I explained, I would like to move away from...