Slog v1: Data-Parallel Deductive Programming

Slog is a language, compiler, and runtime system for implementing data-parallel deductive programming.

The best way to understand Slog is to read our preprint.

Components

• compiler/ -- Compiler from Slog source code to a parallel RA plan
• backend/ -- MPI-based parallel relational algebra plan (RA) runtime
• runslog -- Script to build and run slog program with given input data

A simple example

Here's how you do Transitive Closure in Slog

[(path x y) <-- (edge x y)]
[(path x z) <-- (edge x y) (path y z)]

The Following guide can help you quickly get started with Slog:

Getting Started

The following video (click the image below) demonstrates the first 2-3 articles below: setting up slog, running transitive closure, and showing the first example from the tutorial article.

• Set up Slog
• A simple example: transitive closure
• Tutorial
• REPL and runslog
• Comparison with Souffle
• Components and Internals (Todo)

If you have issues or find any mistakes, please raise an issue.