pypy

You should report this scenario to PyPy - they're waiting for and love examples where Python is faster than PyPy.

Can PyPy work out that a list is fixed sized?

Tracing JITs like LuaJIT and PyPY as fast to warm up compared to method JITs and PEs.

But stock Python 2.7 code can be run under PyPy and it's very fast if speed is a concern.

PyPy might have helped as well by making ordinary Python -- the kind you would otherwise write in C -- run fast.

Nevertheless SPJ has suggested to the PyPy/RPython people to write a meta-tracer for Haskell, just to be able to measure what's possible.

Could you explain why you think contributing to PyPy would better achieve the goal of replacing the Python implementation in Dropbox's codebase?

RPython and PyPy are (IMO) the coolest Python projects out there.

You could read it on PyPy blog and Python mailing lists.

In part because I have done an implementation with the Stackless module of PyPy.

Whereas with PyPy you want to do the opposite.

Dropbox using 2.7 is no different from Dropbox avoiding PyPy, they are evidently ultra-conservative (and probably have a lot of old code they don't want to port).

If "business needs" was a real consideration, they'd just contribute to PyPy instead of writing a new implementation.

For instance, one restriction I had when I was working with PyPy was "str.strip can only take one character", but this has been removed in subsequence PyPy versions.

I always had the impression that the more "RPython-like" your Python code is, the better PyPy can optimize it.

Pypy, for example, claims to have a simple libffi interface.

In Pypy, atLeast2 is 1.9x slower than atLeast4.