I agree, but you have to take extensibility into account. You either have an extensible language and extend it[0], you make it an array language to the core, or

I agree, but you have to take extensibility into account. You either have an extensible language and extend it[0], you make it an array language to the core, or you just have a few slightly convenient functions (not a language). (I don't have enough experience to judge which category Pandas / numpy are in) [0]: https://github.com/phantomics/april https://github.com/phantomics/april