The ontology analysis workflow is intended for validating ontologies based on the core IMCE ontologies. It uses OWL ontologies (TODO: OML AS INPUT) as input, and produces test reports in the quasi-standard JUnit format. The following will walk through the environment setup, and give instructions on how to run the workflow (a) locally and (b) on a CI system such as Jenkins.
Running the workflow requires a number of tools to be configured: JRuby, SBT, Java 8 (JDK), Make. When using a CI server (such as Jenkins), a number of plugins are recommended to be installed. This is mentioned at the bottom of this section.
Ensure that Ruby 2.0.0 is installed. Note: it is recommended to stick with version 2.0.0 to avoid any potential unforeseen side-effects.
Ruby 2.0.0 can be fetched from, e.g., here.
Ensure that JRuby 1.7.24 is installed. Note: it is important to install version 1.7.24, since there have been incompatible changes to the syntax and supported libraries in later versions.
JRuby 1.7.24 can be fetched from here. Ensure that an environment variable JRUBY exists, or that the command which jruby directs to the correct installation.
Install the latest version of the Java 8 JDK from Oracle's website. It is recommended not to use OpenJDK.
Java 8 JDK may already be pre-installed on your system. To check if this is the case, run java -version in a terminal window.
Install the latest version of SBT 0.13.x from the Scala SBT website. Note: do NOT use SBT 1.x (currently a pre-release / beta) - there are known differences in syntax that will break part of the workflow.
Ensure that "make" is installed on the system. This is typically already the case for most Unix-based systems. Make is part of the GNU Utils. On MacOSX, this may require installing Xcode.
When running in a CI system (such as Jenkins), the agent may need to be configured appropriately. For Jenkins:
- Add an environment variable JRUBY under "Manage Jenkins" > "Manage Nodes" > (node name) > "Configure". The value should be the absolute path to the jruby executable: e.g., /usr/local/jruby/jruby-1.7.24/bin/jruby . Note: this is particularly important when using Jenkins 2.4.x, since Jenkins may NOT be aware of the system's environment settings.
- Install the SBT plugin (simply called "sbt plugin"), and configure an installation of SBT 0.13.x .
The above assumes a recommended standard installation of Jenkins 2.4.x, with the recommended plugins installed (e.g., GitHub plugin, JUnit plugin)
MISSING: SETTING UP EXPORT RESULTS / RUNNING OML2OWL
Firstly, ensure that there is an environment variable METADATA set that points at the metadata file produced by the SysML 2 OWL exporter.
To run, execute the following commands in a terminal, starting at the project root:
sbt setupTools setupExportResults
cd workflow
. env.sh
make bootstrap
make location-mapping
make validate-roots
make run-audits
cd ..Note carefully: The above assumes that an existing Fuseki instance is running on the port specified in workflow/env.sh, and with the content produced by loadprod (see profile generation workflow) already exists. This is done to avoid having to reason over basic ontologies again. Run the workflow described here before executing this workflow (profile generation step is not necessary)!
For Jenkins, a pipeline script can be found in the root directory. This file is called Jenkinsfile.
The expected setup for Fuseki can be found here.
Note that results are stored in JUnit format in files under target/*.xml. With the JUnit plugin installed, Jenkins can read the analysis results. Note that also Eclipse's JUnit framework (and other JUnit-supporting (test) frameworks) will be able to parse and display the results of the analysis workflow.