Weekly Sync - 23 April 2024

[github-actions[bot]]

Please add agenda items below and give your weekly update in a comment. Thank you!

Agenda

• @JoblersTune discussion about localenv setup when balancing dev experience and hot module reloading vs keeping Docker images closer to a production use case

Discussion

• How do we get the best possible developer experience, which means hot module reloading (HMR), while ensuring we never neglect interacting with our production state Dockerfiles
• Packages like the frontend Admin UI run using Remix. To run Remix in dev mode for HMR you need to have a container that includes the ability to build and run code. This means we need dev dependencies inside the container. Whereas for a production Dockerfile we have everything layered so that there is a build stage, and only the build assets are copied over, there is a production dependency layer so that only production required packages are installed and then there is a runner mode so that all the code is statically copied into the Docker image and the container is lightweight and only has what it needs to run.
• Now we need to have production ready Dockerfiles, so if we include HMR it means having separate dev Dockerfiles in addition to prod Dockerfiles. The worry is that if we take an approach like this we run the risk of not interacting regularly with these production state Dockerfiles and may miss bugs or issues that we would have otherwise caught.
• The hope is that the integration tests should catch most of these problems if they come up.
• There are no integration tests for frontend but we are planning on adding end-to-end tests for frontend
• Idea to add tests to the CI that check production version of containers do at least spin up without issues.
• The concern remains that some things slip through the cracks and go unseen.
• Discussion about whether to have Helm charts default to dev values or prod values. Currently there is an issue running with the Helm charts in prod mode since frontend fails because it looks for a dev server that doesn't exist
• There was discussion around the idea of spinning up a Kubernetes cluster locally to resolve these conflicts between the different Dockerfile needs but it was decided it was not worth the additional development complexity.
• The summary is that there is no good solution. So we are moving forward with the approach of having two Dockerfiles, one for dev and one for prod, we can have two docker compose set ups so that we can still easily spin up our production Dockerfiles as well Implement Separate Docker Compose Configurations for Development and Production Dockerfiles #2672
• Perhaps we can touch base on this as the other packages begin implementing HMR and it becomes clearer how that looks in other contexts as well. Follow this issue: [DevEx] allow for hot reload of backend and auth #2650
• In order to make HMR possible there are some instances of application memory that need to be addressed still so that the services are stateless and rely fully on the databases to maintain state.
• Deadline: would like this to be completed so that the hackathon participants can use this before the 2024 summit in October.

[njlie]

• Added Admin API security via HMAC signatures
• Working on adding logs to auth server graphQL resolvers

[sabineschaller]

working on backend graphql refactor an adding logging

[BlairCurrey]

• merged in open payments changes for making quote optional
• opened PR and incorporated feedback on making quote optional

[JoblersTune]

Merged PR in to use Kratos to secure the Admin UI.
Busy researching end-to-end testing for Admin UI - leaning toward Playwright.