jPipe - Justified Pipelines

The jPipe environment supports the definition of justification to support software maintenance activities. The name comes from "justified pipelines", as the key idea is to design an environment supporting the justification of CI/CD pipelines by design.

General Information

Principal Investigator:
- Sébastien Mosser, McSCert, McMaster University.
Active Contributors:
- Kalvin Thuan-Phong Khuu, McSCert, McMaster University. PhD Student.
Contributors:
- Nirmal Chaudhari, McSCert, McMaster University. Undergraduate Research Assistant.
- Aaron Loh, McSCert, McMaster University. Undergraduate Research Assistant.
- Deesha Patel, McSCert, McMaster University. Master Student.
- Corinne Pulgar, École de Technologie Supérieure (ETS). Master Student.

Repository organization

jpipe-lang: ANTLR4 grammar and generated lexer/parser
jpipe-model: domain model (justification elements, symbol table)
jpipe-operators: composition operator extension point and built-in operators
jpipe-compiler: compiler pipeline (parsing, model building, validation, export)
jpipe-cli: command-line interface and fat JAR entry point
docs: technical documentation and architecture decision records
bin: launcher script templates (POSIX sh and PowerShell) used by the packaging channels
scripts: maintainer tooling (release.sh)
debian: Debian source packaging metadata for the Ubuntu PPA

Developer Setup

Required tools

Tool	Version	Purpose
JDK	25	Compilation and runtime
Maven	3.x	Build tool
adr-tools	latest	Browsing and creating architecture decision records

Optional tools

Tool	Version	Purpose
MkDocs Material	latest	Previewing the documentation site locally (`mkdocs serve`)

Building

mvn package

The fat JAR is produced in jpipe-cli/target/.

Branching

Work is integrated on dev; main is release-only, and every commit on it is a published version (ADR-0024). Branch from dev and open a pull request for anything with reviewable content. Low-risk changes — version bumps, changelog close-out, CI configuration, a small fix carried by its tests — may be pushed straight to dev once mvn verify is green locally. Only main requires a pull request.

Releasing a new version

A release is prepared on dev, merged into main through a pull request, and started by pushing a vX.Y.Z tag on main. The tag triggers the pipeline, which publishes the GitHub Release and updates Homebrew, Scoop and the Ubuntu PPA.

scripts/release.sh prepare X.Y.Z       # on dev: set the version, close the changelog
                                       # then push, and open the dev → main PR
scripts/release.sh preflight X.Y.Z     # on main, after the merge: re-run the
                                       # pipeline's checks before the tag exists
git tag vX.Y.Z && git push origin vX.Y.Z

git switch dev && git merge origin/main            # after the release
scripts/release.sh post-release <next>-SNAPSHOT    # e.g. 2.4.1-SNAPSHOT

Read docs/releasing.md before cutting a release — it is the full checklist, including the dependency review, the Ubuntu series review and the Windows smoke test that the ADRs require. The pipeline's mechanics and the repository secrets it needs are documented in ADR-0020.

Code style

Formatting is enforced by Spotless (Google Java Format) and Checkstyle. To auto-format before committing:

mvn spotless:apply

Build output / shade warnings

The fat JAR build (jpipe-cli) merges many dependencies and would normally emit overlap warnings from the Maven Shade plugin. These are suppressed by default via .mvn/jvm.config. To re-enable them for a single run:

mvn package -Dorg.slf4j.simpleLogger.log.org.apache.maven.plugins.shade.DefaultShader=warn

Usage

Installation

Released versions are published to one package manager per platform (ADR-0025):

# macOS
brew tap jpipe-mcscert/mcscert && brew install jpipe

# Ubuntu
sudo add-apt-repository ppa:mcscert/ppa
sudo apt update && sudo apt install jpipe

# Windows
scoop bucket add mcscert https://github.com/jpipe-mcscert/scoop-mcscert
scoop install mcscert/jpipe

Each channel installs a jpipe launcher on your PATH; run jpipe doctor to check the installed version and its runtime dependencies. The fat JAR is also attached to every GitHub Release.

Running the compiler

After building from source, the fat JAR is at jpipe-cli/target/jpipe-cli-*.jar. The process subcommand is the default, so these two invocations are equivalent:

java -jar jpipe-cli/target/jpipe-cli-*.jar -i my.jd -d MyModel -f dot
java -jar jpipe-cli/target/jpipe-cli-*.jar process -i my.jd -d MyModel -f dot

Common flags

Flag	Short	Description
`--input FILE`	`-i`	Input `.jd` source file (default: stdin)
`--output FILE`	`-o`	Output file (default: stdout)
`--diagram NAME`	`-d`	Name of the model to export (required)
`--format FORMAT`	`-f`	Output format — see table below (default: `JPIPE`)
`--headless`		Suppress the logo banner
`--log-level LEVEL`		Log verbosity: `OFF` `ERROR` `WARN` `INFO` `DEBUG` `TRACE`

Output formats

Format	Description	Requires Graphviz
`JPIPE`	Canonical jPipe source (round-trip)	No
`DOT`	Graphviz DOT source	No
`PNG`	Rendered PNG image	Yes
`JPEG`	Rendered JPEG image	Yes
`SVG`	Rendered SVG image	Yes
`JSON`	JSON model dump	No
`PYTHON`	Python object model	No

Install Graphviz and run jpipe doctor to verify that the dot binary is on your PATH before using image formats.

Other subcommands

# Parse and validate without exporting — prints diagnostics, statistics, and the symbol table
java -jar jpipe-cli/target/jpipe-cli-*.jar diagnostic -i my.jd

# Same report as JSON, for IDEs and other tooling
java -jar jpipe-cli/target/jpipe-cli-*.jar diagnostic -i my.jd -f json

# Check runtime dependencies (Graphviz)
java -jar jpipe-cli/target/jpipe-cli-*.jar doctor

Minimal `.jd` example

justification MyModel {
    conclusion  c  is "Our claim"
    strategy    s  is "Argument"
    evidence    e1 is "Evidence A"
    evidence    e2 is "Evidence B"

    s  supports c
    e1 supports s
    e2 supports s
}

Compile it to a DOT diagram:

java -jar jpipe-cli/target/jpipe-cli-*.jar -i my.jd -d MyModel -f dot -o my.dot
dot -Tpng my.dot -o my.png

More examples are in the examples/ directory.

How to cite?

@software{mcscert:jpipe,
  author = {Mosser, Sébastien and Khuu, Kalvin Thuan-Phong and Chaudhari, Nirmal and Loh, Aaron and Patel, Deesha and Pulgar, Corinne},
  license = {MIT},
  title = {{jPipe}},
  url = {https://github.com/jpipe-mcscert/jpipe-compiler}
}

How to contribute?

Found a bug, or want to add a cool feature? Feel free to fork this repository and send a pull request.

If you're interested in contributing to the research effort related to jPipe, feel free to contact the PI: Dr. Sébastien Mosser.

We do have undergrad summer internships available to contribute to the compiler, as well as MASc and PhD positions in Software Engineering at Mac.

AI assistance policy

Parts of this codebase were developed with the assistance of Claude (Anthropic), an AI coding assistant. We are transparent about this use and welcome AI-assisted contributions, subject to the following conditions:

Pull requests must not be 100% AI-generated. Every contribution must reflect the understanding and judgement of a human author.
Human authors are fully responsible for the correctness, quality, and appropriateness of their contributions, regardless of whether AI tools were used in their preparation.
Reviewers may ask contributors to explain any part of their submission.