Got an idea for a new feature? Great! The good news is due to Element43's structure you can use your favorite tools as long as they 'speak' gRPC and can be deployed in a Docker container. Extra points for low resource usage and fast response times. Note that the setup described in this guide is not intended for production use.
First, check with the rest of the project's participants if someone is already working on a similar function, maybe you can collaborate! Then you need to think about what you want to add to the application and which components will be affected. Can your feature be implemented purely in the frontend, can you use existing services or do you want to develop a new service? Once you've done that, file an issue in this repo detailing your basic design so everyone can see what you're working on and there is no effort wasted by working on the same functionality in parallel.
There are three basic scenarios for developing services for Element43. In many cases you don't even need to run many components locally.
This is the simplest of the three. You can just work with the production API to test your modifications locally. First, install NodeJS and yarn. Next, fork and clone vue43 and add the following line to store/index.ts just before the apiClient is defined (do no check this change in):
httpSettings.baseURL = 'https://element-43.com/api'That way you're directly working with the production API. Run yarn install to install vue43's dependencies and then run yarn dev to launch the dev server. Wait for the compilation to finish (this may take a couple of minutes), then launch a browser and access the URL displayed in your console. Most likely your browser will block access to Element43's public API due to CORS restrictions. Temporarily disable CORS checks in your browser while developing - do not forget to re-enable them once you're done. You're ready to go now! Consider using the excellent vue-devtools for debugging. Once you're done, review your changes and open a pull request. From there it should not take long for your change to land in production!
If you want to modify an existing service, you may be able to get away with a minimal setup depending on the changes. Again, please file a new issue or comment on an existing one beforehand, so we know you are working on it.
Change the code and open a PR in the service's repo, comment on the issue in the main repo. Done.
If your change requires making RPCs to another service, download and run the other services locally. This should suffice in most cases.
Alternatively, you can spin up the necessary instances using the Kubernetes manifests from this directory:
- Get minikube and create a local cluster or bring your own K8S cluster
- Get and set up telepresence for integrating your local service's instances into the cluster's network
- Launch the required services and their dependencies (check their readme) by running
kubectl apply -f SERVICE_NAME.yml. You can find the manifests indev/manifests. Some services such as esi-markets need an ESI token which you need put into the manifest. You can generate your token using the script inutil/tokenfetcher.pyafter registering an app for testing. - Launch your local services using telepresence. You now have access to your cluster's network.
If your change affects the service's gRPC interface, another PR containing the changes to the gRPC interface is required, too.
This one requires special consideration. Maybe start with defining your service's gRPC description. That way you have a clearly defined specification you can work towards. Once you have picked your language/tools/persistence, ensure to discuss your plans with the rest of the project's participants beforehand. From then on the process is:
- Implement backend service
- Merge PR for new service's gRPC spec into the main repo
- Add CI to new service, publish Docker image
- Deploy service to production
- Deploy changes to jumpgate if needed
- Implement/merge/deploy frontend changes if needed