How to Develop

Dayu keeps the runtime lifecycle in stable service shells and exposes application, scheduling, monitoring, datasource, and visualization behavior through templates and hooks. Start by choosing the extension surface that matches the change instead of modifying the complete request path.

Development environment

The repository declares Python 3.8 and Node.js 20 for local and CI-compatible checks. Docker with Buildx is required only when building images.

git clone https://github.com/dayu-autostreamer/dayu.git
cd dayu

make install-python-dev
make frontend-install
make help

Repository map

Path	Purpose
backend/	Control-plane APIs, template composition, RuntimeService lifecycle, telemetry, and visualization.
frontend/	Vue operator interface and lifecycle state management.
datasource/	HTTP/RTSP sources and manifest-based playback.
dependency/core/	Runtime services, algorithms, common contracts, and application implementations.
template/	Policy, service, component, and visualization configuration.
docker-bake.hcl and build/	Image matrix and Dockerfiles.
docs/	Implementation-facing API, architecture, configuration, hook, operations, and testing references.
tests/	Unit, integration, component, end-to-end, frontend, and optional ML tests.

The repository-local quick start and concepts pages are the best code-oriented entry points.

Choose an extension path

• Add an AI service: follow Add Applications.
• Add a scheduling policy or hook: follow Customize Scheduling.
• Add another runtime hook family: use Hook Functions.
• Add or rename an image: update docker-bake.hcl, the Dockerfile, and run make validate-build.
• Change a public API or lifecycle: update the implementation-facing repository docs and the corresponding website guide in the same change.

Validate the change

Run the narrowest relevant tests while developing, then use the aggregate checks before review:

make validate-build
make test-unit-integration
make frontend-check
make check

ML-backed scheduler and application tests are optional in the default environment. Run make test-python-ml in an environment with the declared ML dependencies when the change affects those paths. See the testing guide for test-layer boundaries.