What are the advantages for Julia in terms of package development?
So how does this translate to packages? First Data Science and ML
What about Scientific Computing?
a competitive advantage since package developers then only need to write a generic implementation of said algorithm and then it can work on any hardware for which someone has implemented a compatible type. As Moore's law comes to an end and variations on GPUs / FPGAs become required to get a speedup, allowing the user to specify the hardware interaction while still using a standard package will definitely be an advantage!