Installer souffle avec Homebrew, Nix

Liens du projet

• Page d'accueil
• Dépôt

Commandes et alias

• souffle
• souffle-compile.py
• souffleprof

Historique du projet

The official documentation says Soufflé was initially designed at Oracle Labs for crafting static analysis in logic. Its publication list shows the research lineage: an ASWEC 2015 experience report, CAV and CC papers in 2016, and later work on indexes, data structures, provenance, interpretation, and incrementalization.

The public GitHub repository was created in March 2016 and presents itself as the official repository for the Soufflé language project. The README describes Soufflé as similar to Datalog, with records, and frequently used as a domain-specific language for analysis problems.

Historique d'adoption

Soufflé's adoption has been strongest in program analysis and research tooling. The official docs list applications including Java static analysis with Doop, network analysis, data analytics, cloud-computing security analysis, and smart-contract security analysis with Gigahorse.

Packaging reflects that specialist audience. The official install page documents Ubuntu/Fedora/Oracle Linux packages, source builds, and a Homebrew tap; the Homebrew metadata in this batch also records a packaged `souffle` formula in Homebrew and Nix.

Modes d'utilisation

Users write relations, facts, and rules in Soufflé, then run the `souffle` CLI to evaluate the program or synthesize efficient C++ for larger analyses. The ecosystem also includes profiling and compile helper tools such as `souffleprof` and `souffle-compile.py`.

Common usage is not a general scripting task but analysis construction: points-to analysis, taint analysis, security checks, graph reachability, and other rule-heavy computations where Datalog is concise and generated code needs to be fast.

Pourquoi les passionnés de paquets s'y intéressent

Soufflé is package-nerd significant because it packages a research compiler and Datalog runtime into a usable developer CLI. It is a bridge between papers, static-analysis frameworks, and installed command-line tools.

It also shows why package managers matter for research software: reproducible installation makes it much easier for analyzers, benchmarks, and downstream tools to depend on the same language implementation.

Chronologie

• 2015: Soufflé-related Datalog static-analysis experience report published.
• 2016: Soufflé CAV and CC papers published and public GitHub repository created.
• 2021: PLDI paper published on an efficient interpreter for Datalog by de-specializing relations.
• 2025: Soufflé 2.5 released on GitHub.

Related projects

• Datalog is the language family Soufflé extends for large-scale analysis.
• Doop is named by the official docs as a Java static-analysis application using Soufflé-style Datalog analysis.
• Gigahorse is named by the official docs as a smart-contract security-analysis application.

Sources

• https://api.github.com/repos/souffle-lang/souffle
• https://api.github.com/repos/souffle-lang/souffle/releases
• https://github.com/souffle-lang/souffle
• https://souffle-lang.github.io/docs.html
• https://souffle-lang.github.io/install
• https://souffle-lang.github.io/publications
• input.json source_facts.package-manager