I'm using it as my main language but there are several reason why I don't push
it in my data science communities:
* I'd like a jupyter kernel/REPL for quick experimentation (arguably this is
possible now with hot-code reloading).
* Vtable (or working generic methods) would be very useful
* ecosystem is too small, especially for data visualizating and
transformation:
* data and serialization formats like protobuf, yaml or even jpeg would
benefit a lot from an unified API
* image loading and saving
* sound files loading and saving
* data visualization
