serverless VS relay-starter-kit

Here is the actual package used [1]. I thought it would have been similar to serverless-pg[2] with initiation outside the handler as with the node implementation. Any insights into how they achieve connection caching (similar to pooling?) on the edge?

[1]: https://github.com/neondatabase/serverless

https://github.com/kriasoft/react-starter-kit (or, relay-starter-kit)

GitHub

It also helps to have a REPL shell for Knex.js — https://i.imgur.com/P5bVdoC.gif

I generate TypeScript enums and types from the actual database schema using knex-types as part of the yarn db:reset script that used during local development (after making changes to the db schema migration files). See example project.

I tried to take advantage of TypeORM in a couple of projects, but then reverted back to knex + knex-types — just a PostgreSQL client with query builder works great, IMO, especially for GraphQL API development.

GraphQL API and Relay Starter Kit— an example of implementing GraphQL API optimized for the serverless environment (Google Cloud Functions) + a reverse proxy using Cloudflare Workers

It helps to have a monorepo setup with packages/projects such as "api", "worker" (cron job) etc. all consuming the same environment variables (db connection settings etc.), when the cron job "serverless" function is deployed, it just picks the required environment variables from the shared "env" folder/package.