tree: f115a57d3f004b870600d9a4683b3203ca0bf017 [path history] [tgz]
  1. bin/
  2. dev/
  3. integration_test/
  4. lib/
  5. test/
  6. tool/
  7. .gitignore
  8. analysis_options.yaml
  9. app.yaml
  10. build.yaml
  11. cloudbuild_app_dart.yaml
  12. cloudbuild_app_dart_deploy.yaml
  13. cloudbuild_app_dart_docker.yaml
  14. Dockerfile
  15. index.yaml
  16. pubspec.lock
  17. pubspec.yaml

Dart backend for cocoon

This folder contains a Dart based backend for Cocoon.

Building and running


  • Install the Google Cloud Developer Tools Command Line Interface (gcloud). Then initialize it and authenticate yourself by running:
gcloud auth login
gcloud init
export PATH="$PATH":"path/to/flutter/bin/"
flutter upgrade
flutter pub get
export PATH="$PATH":"path/to/flutter/bin/cache/dart-sdk/bin/"

Running the tests

$ dart test

Running codegen


To update the JSON serialization generated code, run:

$ dart run build_runner build

Any updates should be checked into source control.


To update the Protocol Buffer generated code:

  1. Download and install the protocol buffer compiler (protoc). Once installed, update your PATH to include the path to the protoc binary.

    On Linux, use sudo apt-get install protocol-compiler to install. On macOS, use brew install protobuf

  2. Install the protoc_plugin Dart package. Once installed, update your PATH to include the path to the protoc_plugin/bin directory (or $HOME/.pub-cache/bin if you used pub global activate protoc_plugin).

  3. Run the following command:

    $ protoc --dart_out=. lib/src/model/proto/**/*.proto
  4. Remove the unused generated files:

    $ find . -regex '.*\.\(pbjson\|pbserver\)\.dart' -delete

    (you can remove the *.pbenum.dart files too, except for protobuffers that actually define enums, like build_status_response.proto)

Generating cloud datastore indexes

To update the indexes in the App Engine project, run:

$ gcloud datastore indexes create index.yaml

Local development

Using physical machine

  • Setting up the environment

This environment is needed as you don't have access to the remote redis instance during local development.

  • Starting server
dart bin/server.dart

If you see Serving requests at the dev server is working.

Using Docker

  • Running a local development instance

Once you've installed Docker and have the docker command-line tool in your path, then you can use the following commands to build, run, stop, and kill a local development instance.

# Build the docker image
$ docker build -t local .

# Start the local container, clearing the console buffer and tailing the logs
$ container_id="$(docker run -d -p 8080:8080 local)" && \
    clear && \
    printf '\e[3J' && \
    docker logs $container_id -f

# Stop the local Docker container
$ docker container ls|grep local|tr -s ' '|cut -d' ' -f1|xargs docker container stop

# Remove the local Docker image
$ docker images|grep local|tr -s ' '|cut -d' ' -f3|xargs docker rmi -f
  • ssh into instance
$ docker exec -it <container name> /bin/bash

Deploying a release to App Engine


Cocoon auto deployment has been set up via Google Cloud Build daily on Workdays.


  • Using the cloud build

This is easy to deploy if you simply want a new version based on the latest commit. Open Cloud Build dashboard and click run in the push-master trigger (example)

$ dart dev/deploy.dart --version version-$(git rev-parse --short HEAD) --project flutter-dashboard

The deploy script will build the Flutter project and copy it over for deployment. Then it will use the Google Cloud CLI to deploy the project to AppEngine.

For more options run:

$ dart dev/deploy.dart --help

Branching support for flutter repo

Add targeted branches in dev/branches.txt, based on which cocoon API filters targeted branches and then runs tests on those branches. With tests running against different branches, the frontend then supports listing commits on a specific branch (defaulting to master).

Once the PR has merged, go to to flush the cache and have the latest change take effect. Otherwise, it can take 12 hours for it to propagate. Ensure you are signed into cocoon dashboard otherwise it will throw a 403.