Artifact for "Making Formulog Fast"

This repository contains the material for the artifact for the OOPSLA'24 paper Making Formulog Fast: An Argument for Unconventional Datalog Evaluation by Aaron Bembenek, Michael Greenberg, and Stephen Chong. The official artifact on Zenodo was reviewed as being "Reusable" and "Reproducible," and was recognized as a Distinguished Artifact.

The two primary reasons for open sourcing this repository are:

1. the repository includes the experimental infrastructure, results, and analysis for the scaling experiments reported in Section 5.3.4, which are not present in the reviewed artifact (check out the scaling branch of this repository); and
2. the GitHub UI makes it easy to skim the artifact's material without downloading the Docker images (for example, you can easily examine the case study implementations in the benchmarks/ directory).

Instructions for reviewers (and folks interested in running the artifact) are in the file 434/README.md; you'll need to download the appropriate Docker image from the "Releases" section of the GitHub repository.

Build Instructions

These are internal instructions for building the artifact.

First, you need to download lib_facts.zip from the GitHub repo "Releases" to the base directory.

We build two images, one for AMD64/x86_64 and one for ARM64/AArch64:

docker build --platform linux/amd64 -t mff-amd64 .
docker build --platform linux/arm64 -t mff-arm64 .

To distribute the images, save them as archives:

docker save mff-amd64 | gzip > 434/vms/mff-amd64.tar.gz
docker save mff-arm64 | gzip > 434/vms/mff-arm64.tar.gz

Saved images will be put in the "Releases" section of the GitHub repo.

Ubuntu 22.04 Setup

• Put script setup_ubuntu_x86.sh in scripts/ directory
• Run script ./scripts/setup_ubuntu_x86.sh
• Put archive lib_facts.zip in benchmarks/scuba/ directory
• Add /usr/local/lib/ to LD_LIBRARY_PATH variable

echo "export LD_LIBRARY_PATH=/usr/local/lib/" >> ~/.bashrc