That just slices rows but something like df[cols] for example would be nice (and is more intuitive), the __slice__ api in python is quiet powerful and I wish polars would take more advantage of it