Erlang

Erlang Python

Apache 2.0

Bidirectional integration between Erlang and Python. Call Python from Erlang. Call Erlang from Python. True parallelism.

Bidirectional Integration

Erlang

Erlang / Elixir

  • py:call()
  • py:eval()
  • py:async_call()
  • py:stream()
Erlang calls Python
Python calls Erlang

Shared State (ETS)

Accessible from both sides

Python

  • from erlang import func
  • erlang.call()
  • state_get/set()
  • state_incr()

Features

True Parallelism

Sub-interpreters (Python 3.12+) or free-threaded Python (3.13+). No GIL contention.

Async/Await

Call Python async functions, gather results, stream from async generators.

Dirty NIF Execution

Python runs on dirty schedulers. Never blocks the BEAM.

Bidirectional Calls

Python calls Erlang. Erlang calls Python. Reentrant callbacks supported.

Streaming

Iterate over Python generators chunk-by-chunk. Memory efficient.

AI/ML Ready

Examples for embeddings, semantic search, RAG, and LLMs.

Code Examples

Erlang Erlang calling Python 
%% Call Python functions from Erlang
{ok, 4.0} = py:call(math, sqrt, [16]).

%% Evaluate expressions
{ok, 45} = py:eval(<<"sum(range(10))">>).

%% Async calls
Ref = py:call_async(math, factorial, [100]),
{ok, Result} = py:await(Ref).

%% Stream from generators
{ok, [0,1,4,9,16]} = py:stream_eval(<<"(x**2 for x in range(5))">>).
Python calling Erlang 
# Python calling Erlang functions
from erlang import my_func, state_get, state_set

# Call registered Erlang functions
result = my_func(10, 20)

# Use shared state (ETS-backed)
state_set('user:123', {'name': 'Alice', 'score': 100})
user = state_get('user:123')

# Atomic counters
from erlang import state_incr
count = state_incr('page_views')
Register Erlang functions for Python 
%% Register Erlang functions callable from Python
py:register_function(my_func, fun([X, Y]) -> X + Y end).

%% Python can now import and call it
py:exec(<<"
from erlang import my_func
result = my_func(10, 20)  # Returns 30
">>).

Parallelism Modes

Mode Python Version Parallelism
Free-threaded 3.13+ (nogil) True parallel, no GIL
Sub-interpreter 3.12+ Per-interpreter GIL
Multi-executor Any GIL contention (fallback)

Use Cases

ML/AI Integration

Use PyTorch, TensorFlow, sentence-transformers, or any ML library. Run inference, generate embeddings, build RAG pipelines.

  • + Semantic search
  • + RAG with LLMs
  • + Embeddings at scale

BEAM Parallelism

Fan out Python work across lightweight Erlang processes. 10x speedup for parallel workloads.

  • + Process-per-request
  • + Parallel map
  • + Linear scaling

Async I/O

Call Python async functions with async_call/async_await. Stream from async generators.

  • + aiohttp requests
  • + Async database queries
  • + Concurrent API calls

Data Processing

Leverage pandas, numpy, scipy for computation. Erlang handles orchestration.

  • + Pandas DataFrames
  • + NumPy arrays
  • + Scientific computing

Installation

rebar.config 
{deps, [
    {erlang_python, {git, "https://github.com/benoitc/erlang_python.git", {branch, "main"}}}
]}.

Requires Erlang/OTP 27+ and Python 3.12+

Read the Docs View Source