Distributed Experiment Workflow Representation
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
BNF
Constraints
Generators
UI Adding code for web surveys Oct 2, 2018
WebSurvey
translators/bash
LICENSE
README.md

README.md

DEW (Doer!): Distributed Eperiment Workflows Representation

DEW unifies behavior and topology of an experiment, but captures only relevant topological details.

Design overview

DEW has several design goals:

High-level representation

It was important to us that DEW contains a human-readable, short description of what the experiment is supposed to do. Such a description facilitates reuse, because researchers could, at a glance, understand the experiment's actions and judge if this experiment is useful to them. Moreover, expressing what an experiment will do in broad strokes captures the process, which so far has existed only in the researcher's mind, and eases experiment design.

Generic language

DEW should support many diverse experiments, and thus its language must be expressive enough to be broadly applicable, regardless of the experiment's goal or testbed infrastructure where it will be realized.

Self-contained representation

We wanted DEW to contain sufficient details to facilitate automated generation of experimental topologies and scripts, which would run on these topologies. If we could achieve this, then researchers could work with experiments at a high level, delegating tedious, detail-oriented and error-prone tasks to machines. This would further facilitate generation of topologies and scripts in many different languages, and for many different testbed infrastructures, enabling portability.

Decouple behavior from topology

We wanted to decouple the intended behavior from the topology where it will be realized, thus enabling the same experiment to be scaled up and down easily, by changing a few lines of DEW. Our goal was to capture only the necessary topology details in the form of constraints the behavior must place on resources.

Structured representation

We wanted to impose some natural structure on DEW, enabling researchers to easily locate and focus on the important pieces for their goal(s). This structure should facilitate experiment design and running, which are aligned with human cognitive process.

Installing

You'll need: - Python (either 2.x or 3.x) - pyparsing - networkx 2+ - Enum - numpy

Optionally you'll also want: - spaCy (for Natural Language Processing) - spaCy english model (en) - CEFTB Xir description language (for constraint processing and saving experiment descriptions)

See the readme in the Constraints directory for setting up and using the constraint server.

For spaCY install and model install instructions see: https://spacy.io/usage/ https://spacy.io/usage/models

For installing Xir see: https://github.com/ceftb/xir https://github.com/ceftb/xir/tree/master/lang/python

Running

To run the GUI: % python dew_gui.py

Organization

You can find the BNF notation of DEW in BNF folder.

Folder UI has our UI implementation

Folder translators will house translators from other representations into DEW