A Containerized Server To Execute Code Remotely

Runs a dockerized server that allows you to execute Python code via HTTP requests. All the Python code executions will run concurrently; the server will not block any requests. This means that the server will not wait for the execution of the Python code to finish before accepting new requests.

Make sure to init the submodules before running the server: git submodule update --init --recursive

Running the Server

To run the server, you need to have Docker installed on your machine (preferably with the gVisor runtime for security reasons). Then, you can just use the ./build_and_run.sh script to build the container and run the server on your machine. Alternatively, you can use ./pull_and_run.sh to pull the container from Docker Hub instead of building it.

If you are feeling dangerous, you can also just run the server directly with ./run.sh. You'll need rust installed on your machine to compile the server along with whatever runtime the language of the code you want to execute requires.

Increase Open File Limit

If you plan to run crazy amounts of code in parallel, you might want to increase the open file limit of your machine:

sudo sysctl -w fs.file-max=1000000
ulimit -n 1000000

Calling the Server

We provide a simple Python library to interact with the server, which you can find in the ./code_exec_reqs.py file. The server runs on port 8000, so you can call it from http://127.0.0.1:8000.

You can also use whatever http client to interact with the server. There are three endpoints:

• /py_exec: executes Python code. Expects a json with field code containing the Python code to be executed.
• /any_exec: executes code in any language. Expects a json with fields code containing the code to be executed and lang containing the language of the program. See list of supported languages below. It's preferred to run python with the /py_exec endpoint, as it doesn't have the overhead of running MultiPL-E evaluators.
• /py_coverage: executes the Python code and returns the coverage of the code. Expects a json with field code containing the Python code to be executed.

Both /py_exec and /any_exec will return the exit code of the program, followed by the stdout and stderr of the program after the newline character.

Other Languages

Yes! The server supports about 29 languages:

• clj
• cpp
• cs
• dfy
• dlang
• elixir
• fs
• go
• hs
• java
• javascript
• julia
• lean
• lua
• luau
• matlab
• ocaml
• php
• pl
• python
• racket
• r
• ruby
• rust
• scala
• sh
• swift
• ts
• v

This is thanks to the MultiPL-E evaluators, check them out here.