dappl and pineappl Artifact for OOPSLA 2025

Introduction

This artifact provides the implementation of dappl and surrounding experiments.

Hardware Dependencies

These experiments require a working Docker installation. This artifact was tested by the authors on x86 Ubuntu 22.04 and ARM macOS. Instructions will assume a Linux-like shell environment for running command to set up the container.

To run full experiments, expect a server with >=512GB of RAM to take about 12 hours.

Getting started

1. Get a working docker installation.
2. Run the following commands in a shell to pull the container from DockerHub:

  docker pull minsungc/dappl:oopsla_2025
  docker run -it --rm --platform linux/amd64 minsungc/dappl:oopsla_2025

3. Once in the container, cd in to the dappl/ folder and run the command dappl. You should get the dappl help page. If not, run alias dappl='./dappl/_build/install/default/bin/dappl' and try again.

WARNING: the image is rather large in size and requires the linux/amd64 architecture. The build may fail if there is not enough disk space (~10gb) or if the architecture is unspecified.

Running your own dappl programs

Several example dappl programs are given in the examples/ folder. To run your own dappl program, or one from the examples, type in the container

  dappl run $FILE

where $FILE is the path to your .dappl file. There are also optional debug and caching options available for toggle; type dappl run -help for details.

Recreating experiments

Replicating the experiments is expensive, requiring about 12 wall clock hours on ~512GB of RAM and an AMD EPYC CPU. A server or workstation is recommended.

To replicate experiments, type into the console

  python3 experiment.py

If you would like to see a specific experiment ran, comment out the other experiments in experiment.py.

The resulting numbers are stored in a .csv file in the numbers/ folder.

"Kicking the tires"

The "kick the tires" scripts replicates a small fraction of the Bayesian network experiments in Section 6.1 of the paper. To do this, run

  ./kick_the_tires.sh

On an ARM Mac with no other processes, this took about 7 minutes to run.

The resulting numbers are stored in a .csv file in the numbers/ folder.

Navigating the source repository

• bin/: command line tooling.
• derkinderen/: scripts from Derkinderen et al, ECAI 2020.
• examples/: examples of dappl programs.
• experiments/: Python files for automated experimentation and result output
• lib/: the lexer, parser, and interpreter
• testgen/: the main test generation tool. within:
  • bn/ houses different bayesian networks that can be generated
  • mdp/ houses the DR benchmark
  • ladder houses the network ladder diagnosis benchmarks
  • grid/ houses the gridworld benchmarks