ewoks 0.1

Many workflow management systems exist to deal with data processing problems that can be expressed as a graph of tasks, also referred to as a computational graph or workflow. The main purpose of a workflow management system is to provide a framework for implementing tasks, creating graphs of tasks and executing these graphs.

The purpose of ewoks (ESRF Workflow System) is to provide an abstraction layer between graph representation and execution. This allows using the same tasks and graphs in different workflow management systems.

ewoks has been developed by the Software group of the European Synchrotron.

Getting started

Install ewoks

pip install ewoks

Create a test workflow in JSON format (“acyclic1” is an test workflow that ships with ewoks)

ewoks convert acyclic1 --test test.json

Execute the test workflow and print the output of all tasks

ewoks execute test.json --outputs all

For a desktop GUI, install ewoks with the orange binding

pip install ewoks[orange]

Open the test workflow in the GUI

ewoks execute test.json --engine orange

For a web GUI, install ewoksserver with the frontend

pip install ewoksserver[frontend]

Start the ewoks web server, open the link in a web browser and load the test.json file

ewoks-server

Ewoks has different interfaces to create and/or execute a workflow

• Python
• Command line
• Job scheduling
• REST API
• Desktop Graphical Interface
• Web Graphical Interface

Documentation

• Hello world
• Ewoks events
• Use different engines
• Benchmark workflow execution
• API documentation

Binding documentation:

• ewokscore : create workflows and implement tasks

• ewoksorange : create and execute workflows with a desktop GUI

• ewoksppf : execute cyclic workflows

• ewoksdask : parallelize workflow execution

• ewoksjob: distribute workflow execution

• ewoksserver: REST server to manage and execute workflows

• ewoksweb: web frontend to create, visualize and execute workflows

• ewoksutils : developer utilities