Lyo/LyoRIO

From Eclipsepedia

< Lyo
Revision as of 10:58, 17 November 2011 by Jconallen.us.ibm.com (Talk | contribs)

Jump to: navigation, search

Contents

Lyo Reference Implementation for OSLC

This document gives you a quick overview of the Reference Implementation for OSLC (RIO), explain how RIO is organized and how to build and run the code.

What is Lyo RIO?

RIO is a simple, bare-bones reference implementation of the OSLC specifications. It is written in Java as a standard Java EE web applications with minimal dependencies. It is intended to help those who are adopting OSLC by providing a functioning system that can be explored via a simple UI and REST services, or by taking a look at the source code.

Goals of RIO

The goals of RIO are:

  • Provide minimal reference implementation of the OSLC specifications
  • Provide a tool for provider and consumer implementations to reference and experiment with
  • Provide a framework to prototype proposed additions to the OSLC specifications

RIO is *not* intended to be:

  • A full implementation of OSLC
  • A full featured ALM tool
  • A performance benchmark
  • A framework or SDK

RIO architecture

RIO is a standard Java EE web applications with minimal dependencies and it organized into four modules.

Modules and Dependencies

RIO is organized into the following components:

  • RIO Core JAR - base classes for services, RDF triple store and query syntax parser
  • RIO Core Webapp - common JSP pages and static resources used by RIO web applications
  • RIO CM Webapp WAR - the RIO Change Management web application
  • RIO AM Webapp WAR - the RIO Architecture Management web application
  • RIO RM Webapp WAR - the RIO Requirements Management web application

The major dependencies of RIO are:

  • Java Servlet API
  • Java Server Pages (JSP)
  • Open RDF / Sesame RDF parser and triple-store
  • ANTLR parser generator
  • Maven build system

We choose to use a very minimal set of dependencies for RIO because we want it to be really simple. The web parts of RIO are implemented with only the Servlet API and JSP pages. For simplicity's sake, there is no webapp framework, no Dojo and no OSGI.

  • For RDF, we choose to use !OpenRDF / Sesame over Jena because it seemed easier to work with.
  • For the build, we chose Maven for these reasons:
    • Allows developers to very easily *build and run RIO with any IDE or no IDE at all*, i.e. via command-line
    • Allows us to pull in dependencies at build-time and we *do not have to store or distribute any 3rd party jars*
    • Maven is very well known and supported, it is the de facto standard build system for Java based open source software


Areas for improvement

Areas we are looking at improving:

  • Support for all OSLC domains
  • Complete support of OSLC Query Syntax
  • Make RIO a consumer as well as an OSLC provider
  • User management and Oauth base authentication.
  • Better UI via CSS, Javascript, etc.
  • Hosting RIO on a public site, data wiped clean nightly
  • Binary releases, with bundled Jetty so its easy to unzip and run
  • Experiment with alternate backend technologies (e.g. Jena, ARQ, JAX-RS, etc).

To track enhancements and bugs, search on product Lyo in Bugzilla.

RIO architecture design discussions

How to build and run RIO

See Building and Running reference implementations