Vincent-lau / erlog

Erlog: A distributed datalog engine

Build

To build this engine, install Erlang/OTP (>24.0) and rebar3 and then run

 rebar3 compile

Run

For the syntax of a datalog program, you can check out a few examples in apps/erlog/doc/examples/. Once you have a datalog program, change the field in erlog.app.src which says {mod, {erlog_app, ["path/to/file"]}} to point to your datalog program. You can also modify the num_tasks env variable to control how many tasks you want to split the original input into.

Then open up a few shells, and use one of them as the coordinator, and the rest of workers. For example, you can do

# shell 1
rebar3 shell --name 'coor@127.0.0.1'

# shell 2
rebar3 shell --name 'worker1@127.0.0.1'

# shell 3
rebar3 shell --name 'worker2@127.0.0.1'

This is setup the cluster and once these Erlang nodes are up, run erlog_srv:start_working() in the coordinator node and the engine should now start to computing your program!

Tests

To test the engine, use Erlang's Eunit and Common test framework

rebar3 do eunit, ct

Note

An alternative way of structuring this code can be found in the two-app branch, where I separated one application into two to avoid the need for distinguishing two different types roles in the engine: coordinator and worker.