Skip to content

Latest commit

 

History

History
36 lines (25 loc) · 4.09 KB

README.md

File metadata and controls

36 lines (25 loc) · 4.09 KB
Raw

benchmarx

Infrastructure for implementing benchmarx: benchmarks for bidirectional transformation (bx) tools. Also contains a collection of example benchmarx and test runners for various and diverse bx tools.

Running the benchmarx without getting your fingers dirty

We have a plug and play (via remote desktop) Share virtual machine available from: http://is.ieis.tue.nl/staff/pvgorp/share/?page=ConfigureNewSession&vdi=Ubuntu12LTS_BenchmarX.vdi

How to setup and execute the benchmarx

  1. Clone this repo: git clone https://github.com/eMoflon/benchmarx.git benchmarx
  2. Download the latest version of the Eclipse Modeling Tools for your platform. Currently tested for: http://www.eclipse.org/downloads/packages/eclipse-modeling-tools/neon3
  3. Start Eclipse in a workspace of your choice and import the Eclipse project BenchmarxFamiliesToPersons (benchmarx/examples/familiestopersons/BenchmarxFamiliesToPersons) from the working tree of the benchmarx git repository you just cloned.  
  4. Choose the tools you want to execute by appropriately manipulating /BenchmarxFamiliesToPersons/src/org/benchmarx/examples/familiestopersons/testsuite/FamiliesToPersonsTestCase.java/tools().
  5. (Optional) If you want to setup the bx tool BiGUL, then you'll have to work through /BenchmarxFamiliesToPersons/src/org/benchmarx/examples/familiestopersons/implementations/bigul/README-SETUP to do this.
  6. Choose the project BenchmarxFamiliesToPersons and select "Run As/JUnit Test" to execute the benchmarx "Families to Persons" for all tools chosen in Step 5.
  7. You can compare your results with /BenchmarxFamiliesToPersons/results/TestResults.xlsx.

How to extend the benchmarx by adding a new implementation (and/or new test cases!)

To add an implementation of an existing benchmarx example, e.g., FamiliesToPersons with your favourite bx tool, do the following (taking FamiliesToPersons as an example, substitute as appropriate for another example):

  1. Add a folder for your tool to /examples/familiestopersons/BenchmarxFamiliesToPersons/src/org/benchmarx/examples/familiestopersons/implementations/<your_bx_tool>.
  2. Add a Java test runner that is responsible for invoking and interfacing with your bx tool. If your bx tool is EMF-based, take a look at emoflon/EMoflonFamiliesToPersons.java and medini/MediniQVTFamiliesToPersons.java. If your bx tool does not use EMF, then take a look at bigul/BiGULFamiliesToPersons.java.
  3. If your implementation needs extra artefacts (such as Eclipse projects, etc) then add these to /examples/familiestopersons/implementationsArtefacts.
  4. Extend /examples/familiestopersons/BenchmarxFamiliesToPersons/src/org/benchmarx/examples/familiestopersons/testsuite/FamiliesToPersonsTestCase.java to include your new test runner.
  5. Feel free to add new JUnit tests that demonstrate the strengths of your tool (and reveal weaknesses of others). Please try to classify your new tests using the existing structure in the benchmarx.
  6. If your bx tool requires additional setup steps in addition to compiling a standard Eclipse project, add a README-SETUP file to this effect in your bx tool's implementation folder (cf. bigul/README-SETUP).
  7. Please remember to create and send us a pull request from your fork so we can update the benchmarx!

How to extend the benchmarx by adding a new example

  1. Copy the existing structure for /examples/familiestopersons. This consists of a folder metamodels for the source and target metamodels in an EMF representation, a folder BenchmarxFamiliesToPersons for the actual testsuite, and a folder implementationArtefacts for any further implementation artefacts.
  2. The folder BenchmarxFamiliesToPersons is an Eclipse project that can be run as a JUnit test suite and contains all test cases as JUnit tests, all resources (input and expected models), any implementations of the benchmarx example, and auxiliary classes for both metamodels used to simplify the comparison of output and expected models (or model representations).
  3. Please remember to create and send us a pull request from your fork so we can update the benchmarx!