Skip to content

Latest commit

 

History

History
106 lines (60 loc) · 5.08 KB

CONTRIBUTE.md

File metadata and controls

106 lines (60 loc) · 5.08 KB
Raw

Contribute to the MUBench Platform

While we invested a lot into the MUBench platform, to increase its usability and support different experiment scenarios, there is always one more thing to do. Therefore, we always welcome contributions to the platform itself, be it in code or documentation. We recommend the following procedure

  1. Choose an issue or create a new issue to discuss the idea you want to work on.
  2. Fork the project.
  3. Make you changes and add respective tests.
  4. Create a pull request.

The MUBench platform consists of several components. The subsequent documentation presents the purpose and covers specifics regarding the development of each of these components, as well as the project's Continuous Integration/Deployment.

The MUBench Dataset

Is a dataset of known API misuses in various real-world software projects.

The dataset resides in the /data directory. We encourage anyone to contribute to the dataset.

The MUBench Pipeline

Is a Python application that manages the benchmarking dataset, checkout of subject project code, compilation of subject projects, execution of detectors, and capturing of detector output.

Its source code resides in the /mubench.pipeline directory.

Release

See Docker Image.

MUBench Pipeline Java Utils

Is a Java library with some utilities used by the MUBench Pipeline.

Its source code resides in the /mubench.utils directory.

Release

To release a new version of the Java utils, follow these steps:

  1. Set the version in the pom.xml file to the next release version, e.g., 0.0.5.
  2. Run mvn deploy.A
  3. Set the version in the pom.xml file to the next development version, e.g., 0.0.6-SNAPSHOT.
  4. Obtain the md5 hash of the target/mubench.utils-0.0.5-jar-with-dependencies.jar file.
  5. Update the constants __UTILS_VERSION and __UTILS_JAR_MD5 at the top of /mubench.pipeline/utils/java_utils.py accordingly.
  6. Commit and publish the change.

MUBench Pipeline CLI

Is a Java library with the framework and some utilities for MUBench Runners.

Its source code resides in the /mubench.cli directory and its API documentation on the STG artifact page.

Release

To release a new version of the CLI, follow these steps:

  1. Set the version in the pom.xml file to the next release version, e.g., 0.0.14.
  2. Run mvn deploy.A
  3. Run mvn site deploy.A
  4. Set the version in the pom.xml file to the next development version, e.g., 0.0.15-SNAPSHOT.
  5. If the new release changes the way that the MUBench Pipeline should communicate with MUBench Runners, create a respective subclass of RunnerInterface in /mubench.pipeline/data/runner_interface.py.
  6. Commit and publish the change.

MUBench Review Site

Is a PHP application that manages the display of detector findings in MUBench experiments, manual reviews of these findings, as well as computation of statistics on the experiment results.

Its source code resides in the /mubench.reviewsite directory.

Release

See Docker Image.

Docker Image

MUBench comes with a Docker image to enable cross-platform portability of the benchmark. We use this image also as a medium to deliver the MUBench Pipeline and the MUBench Review Site to users.

Release

A development version of the docker image is continuously deployed to Dockerhub with every succeeding CI build of the master branch of this repository.

To release a stable version of the docker image, run ./deploy_release in the repository root. This will publish the version currently checked out to Dockerhub, using the stable tag.

Continuous Integration and Deployment

The MUBench project uses Shippable for Continuous Integration of the platform components described above. This involves a custom Docker image to support building and testing all modules in one build, despite the multitude of different programming languages in use.

The CI configuration resides in shippable.yml.

After a successful build on the master branch, a new version of the MUBench Docker image is built and deployed to Dockerhub, using the latest tag.

ADeployment requires authentication credentials for stg-mubench in your ~/.m2/settings.xml to deploy the new library version to ftp://www.st.informatik.tu-darmstadt.de/mubench/mvn.