Generate a set of pre-configured scripts

Inizia dal flusso di bootstrap basato su template e dall’entry point harness-generate esistente, quindi esamina come vengono rappresentati i tipi di applicazione e le sovrascritture dell’utente. Esamina le aspettative relative al README.md del progetto generato. Il completamento deve essere definito per la generazione degli script, preservando gli script esistenti e documentando come vengono eseguiti i comandi frontend o backend generati.

Currently, cloud-harness is using a template-based file/folder hierarchie to bootstrap an application in a project. The problem is that, we still need to remember or have complex command lines that needs to be written by hand if we want, for example, to run the backend in local. For a flask + fastapi application, we need to remember to run uvicorn, what are the env var we need to pass on the command line. Usually, this is manually written in a README file somewhere on the project repo, but it can be an information sometimes complicated to find back, if written in the first place.

I wonder if it wouldn't be possible to generate some scripts, perhaps 1 by application that could just wrap the command line that needs to be passed on the cli to run the application:

• for something identified as a frontend application, a run-{name-of-the-frontend-app}.bash file that would just run a yarn install && yarn dev or more elaborated strategy
• for something identified as a django/python backend application, a run-{name-of-the-backend-app}.bash file that would just run CH_CURRENT_APP_NAME={name-of-the-app} CH_VALUES_PATH=deployment/helm/values.yaml {other-app-depepden-env-var} uvicorn --host 0.0.0.0 --port 8000 main:app
• for complex setup where port-forwarding is also required to make the local backend communicate with the deployed db, perhaps some scripts. This could be integrated to the bash file running the backend by also providing a set of defaults options/flags to pass to the command line to target the remote db (if applicable).

The generation strategy would consider that if the script doesn't exists, it generates it, but if it does exist, it is not. The idea is to let the user being able to override those files if required (e.g: refine the command use to run the frontend, or modify some env vars)

If we push a little bit this idea, we could also generate files that generates the backend and frontend binding from an openapi specification. This is currently done by harness-generate, and even if it works well for cloud harness project that uses directly openapi specification, it doesn't work well if we use django-ninja for example, or any other kind of not-yet-supported stack. Generating a bash script for the application which is based on its type (this generated script could be part of the application template) would allow us to have a flexible way to accomodate for future evolutions without having to drastically change cloud-harness, only the templates while being flexible as we would be able to modify those scripts for each applications if we really need it.

Having those scripts generated would also allows us to generate a part of the README.md with instructions about how to run the project in a very agnostic way