5 ms·

Have you considered Lua? You can develop web apps with Lapis [1] and do your data analysis with Torch [2]. The tradeoffs compared to Python: LuaJIT is much fas

Have you considered Lua? You can develop web apps with Lapis [1] and do your data analysis with Torch [2]. The tradeoffs compared to Python: LuaJIT is much faster, which should ease your scaling worries, but the ecosystem is not as developed. [1] http://leafo.net/lapis http://leafo.net/lapis [2] http://torch.ch/ http://torch.ch/

aldanor 9y ago
It’s really not fair to compare CPython with LuaJIT. If you compare LuaJIT with PyPy or Numba, the “much faster” argument will simply not be true. What’s different is that both of the above only cover a part of the language, but for specialised (e.g. numeric) application that’s often not a problem.
- whyever 9y ago
  I think LuaJIT is generally faster, not only because Python is harder to optimize.
- shakna 9y ago
  LUAJIT is often on par with C. How would that make it slower than Pypy?
- etiene 9y ago
  > If you compare LuaJIT with PyPy or Numba, the “much faster” argument will simply not be true Source? AFAIK it does hold true. Last time I benchmarked those, LuaJIT was significantly faster, with PyPy almost being on pair with vanilla Lua, due to Python being a much more complex language and harder to optimise.
agibsonccc 9y ago
Just want to note a ton of the torch people have largely moved to pytorch. Lua with LuaJIT probably works for some people, but I don't see it really growing in marketshare over python..
- etiene 9y ago
  This is true, but the reason is just people already knowing python / python being more popular. Lua is a much simpler language than python or julia though, I'd absolutely still recommend it for data scientists who are new to programming.
  - agibsonccc 9y ago
    Right but with a good enough FFI and wanting to actually deploy models in some way, having the infrastructure to deploy general purpose applications in the same language (even if it's slower than say: go or the JVM) is really appealing. Lua itself while simpler owns a very different part of the application space than python does. There's a reason more peopple know it. I would actually recommend against lua because of the lack of libraries for doing every day data science tasks. I mean look at what facebook had to do to justify using lua, it had to invent the notebook for it.

aldanor 9y ago

It’s really not fair to compare CPython with LuaJIT. If you compare LuaJIT with PyPy or Numba, the “much faster” argument will simply not be true. What’s different is that both of the above only cover a part of the language, but for specialised (e.g. numeric) application that’s often not a problem.

whyever 9y ago
I think LuaJIT is generally faster, not only because Python is harder to optimize.
shakna 9y ago
LUAJIT is often on par with C. How would that make it slower than Pypy?
etiene 9y ago
> If you compare LuaJIT with PyPy or Numba, the “much faster” argument will simply not be true Source? AFAIK it does hold true. Last time I benchmarked those, LuaJIT was significantly faster, with PyPy almost being on pair with vanilla Lua, due to Python being a much more complex language and harder to optimise.

agibsonccc 9y ago

Just want to note a ton of the torch people have largely moved to pytorch. Lua with LuaJIT probably works for some people, but I don't see it really growing in marketshare over python..

etiene 9y ago
This is true, but the reason is just people already knowing python / python being more popular. Lua is a much simpler language than python or julia though, I'd absolutely still recommend it for data scientists who are new to programming.
- agibsonccc 9y ago
  Right but with a good enough FFI and wanting to actually deploy models in some way, having the infrastructure to deploy general purpose applications in the same language (even if it's slower than say: go or the JVM) is really appealing. Lua itself while simpler owns a very different part of the application space than python does. There's a reason more peopple know it. I would actually recommend against lua because of the lack of libraries for doing every day data science tasks. I mean look at what facebook had to do to justify using lua, it had to invent the notebook for it.