https://github.com/meooow25/recopt

https://github.com/meooow25/recopt

Last synced: 26 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/meooow25/recopt
• Owner: meooow25
• License: mit
• Created: 2022-05-10T22:33:36.000Z (over 2 years ago)
• Default Branch: master
• Last Pushed: 2022-05-15T06:39:50.000Z (over 2 years ago)
• Last Synced: 2023-04-03T21:42:20.270Z (almost 2 years ago)
• Language: Python
• Size: 5.86 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# recopt

Optimize recursion in [PyPy](https://www.pypy.org/download.html)

---

```py

@recopt.optrec('fact') # <--- this

def run_fact(n, mod):

    def fact(n):

        if n == 0:

            return 1

        r = fact(n-1)

        return n * r % mod

    return fact(n)

```

**What is this?**  

Recursion in PyPy is very slow (why? I don't know). This function transforms the [AST](https://docs.python.org/3/library/ast.html) of a recursive function to remove the recursive calls while preserving the behavior.

**How does it work?**  

A function call is about executing a function and resuming where we left off once the called function is done. This is achieved here by splitting the body of a recursive function into multiple functions, such that resuming means executing another function. The functions to resume are maintained on a stack.

There are some restrictions on what is supported.

* The recursive function must have positional arguments only.

* Recursive call statements must be simple calls `f()` or assignments `x = f()`.

* The only control flow statements supported are `if`, `while`, `for`.

**How fast is it?**  

See [`bench_fact.py`](bench_fact.py) and [`bench_dfs.py`](bench_dfs.py).

**Tell me more about how to use it**  

Don't.