Open Affect Flow is very simple application inspired by Stack Overflow, created to illustrate architecture principles.
Work log:
- Created a project based on https://openliberty.io/guides/getting-started.html.
- Did a number of small adjustments in the
pom.xmlandDockerfile - Used Jakarta EE 8 Web Profile instead of Micro Profile
- Got rid of the Java code
- Got rid of the static web assets and created basic html and css assets.
With this version, we can:
- We can run the project in dev mode with
mvn liberty:dev - We can build a Docker image with
./build-image.shand run it with./run-image.sh - We can visit a page showing the "Open Affect Flow" title
Work log:
- Created a first CI pipeline with GitHub Actions
- The pipeline is run when commits are pushed on master or when pull requests are opened on master (every commit to the PR then also triggers the pipeline)
- The pipeline builds the .war file with maven, build the Docker image and uploads it to the GitHub Docker Registry (at the time of writing, in beta)
- Created a personal access token and used as a secret in the project repository, so that the GitHub action can log into the GitHub Docker Registry
- Manually configured the Docker image (which is a package at the organization level) to make it publicly available
With this version, we can
- Build and publish a Docker image packaging our application
- Any interested user can now
docker run -p 9000:9080 ghcr.io/openaffect/flow
Work log:
- Implemented a "side" use case to illustrate the overall code structure, based on ideas from the hexagonal architecture, the onion architecture, the clean architecture and domain driven design. The user can visit the
/jokespage, which runs a query to get a list of jokes. On this page, he can also send a command to propose a new joke. - Code has been added in all layers of the application:
- In the domain layer, there is now a Joke entity and a JokeId identifier for it.
- Also in the domain, there is an IJokeRepository interface that defines the contract to be fulfilled by Joke Repositories.
- In the infrastructure / persistence layer, there is an in-memory implementation of the IJokeRepository interface.
- In the application layer, there is a facade that handles commands and queries built in the user interface layer. There is one command to propose a new Joke. There is a query to retrieve a list of Jokes.
- In the ui / web layer, there are endpoints that react to incoming HTTP request.
- The first endpoint reacts to GET requests when users visit the jokes page. The endpoint prepares a query object, invokes the facade, receives a DTO from the facade and finally delegates the rendering of the page to a JSP.
- The second endpoint reacts to POST requests when users have filled the form and clicked on the submit button. The endpoint prepares a command object, invokes the facade and goes back to the Jokes page.
- We also have unit tests, one integration test (which tests the collaboration between the facade and in-memory repository) and one e2e test (which tests the use case from the end user point of view).
With this version, we can:
- Visit /jokes, write jokes and see them in a list.
- Jokes are stored in an in-memory "data store" (i.e. a map)
- Run an end-to-end test with CodeceptJS
Work log:
- Introduced a Person entity in the domain, with a PersonId and a PersonRepository
- Refactored the code to introduce an abstract InMemoryRepository (with Java generics), extended by the revised InMemoryJokeRepository and InMemoryPersonRepository
- Deal with unique constraint on the username, with a unit test designed to validate concurrent behavior
- Introduced an Identity Management facade, with a Register command and a Login command. Introduced exceptions.
- Updated the e2e tests
WIth this version, we can:
- Register and login
- Access /jokes only if we are logged in
- Logout
- Run end-to-end tests to validate this behavior