[gaius]

Julias big draw is performance, in my eyes. Ease of use is good too, Python syntax + types + macros is great, but the speed is absolutely killer once you get it tuned up.

Yes I’ve heard that but the counter-argument is just use Numba and leverage your existing Python/NumPy code onto GPU. Everyone always forgets that the Python community is very comfortable with offloading heavy lifting to other languages.

[qop]

Those people generally aren't of the same caliber as Matlab programmers.

I'm talking about professional-grade numerical work. People use matlab or R for this currently. Mathematica too.

These people, who are very talented mathematical minds but maybe not necessarily the most talented programmers, are going to eventually find Julia to be more convenient and more attentive to their needs doing numerical work than Python+PyJunk libraries.

Python sucks. Plain and simple. No types, no macros, it's slow, it can be a pain to get gpu support up and running, the main guy just up and left (?!?!?), and the primary survival trait of the community at large is figuring out how to write python and make something other than python do the work.

It would be like buying a car and the dealer hands you a complimentary bus pass and says "try and keep her off the road if you can help it."

So yeah. I don't consider it a major threat to the ecosystem.