• STAGE 1: wide but shallow. Build a lot of components that make up a Rails-like framework, but only invest 60% in what they should be.
• STAGE 2: operable demo. Build tooling - CLI, testing, and a demo app to showcase the framework
• STAGE 3: tooling polish. Make day to day development operations easy, such as adding models, controllers, logic and tests (by code generation, macros, or helpers)
• STAGE 4: indepth development. Make the 60% go to 80% of functionality where needed. Things like documentation (website?), missing API, supporting various deployment scenarios (on docker, other platforms), feature flags to cut down functionality
• STAGE 5: go wide and expand. Focus on building various kinds of demo apps

To start a new project, you can use cargo-generate:

$ cargo install loco-cli
$ loco new

cd examples/demo

1. create a database (through your postgres admin app):

loco_app

2. migrate + generate entities:

$ cargo loco db status
$ cargo loco db migrate
$ cargo loco db entities

3. run:

= terminal 1 =

$ cargo loco start

= terminal 2 =

$ cargo loco workers

Using the framework

• Convention over configuration, though with a static language like Rust, this is highly diminished
• Developer happiness (testing-first, CLI, tooling, beautiful code)
• ActiveRecord is a good thing, fat models, slim controllers
• Secure defaults
• Optimize for the solo developer, the "one man framework"
• Take advantage and preserve the Rust x-factor: performance, single-binary deploys, coding tools, community

Building the frameowrk

• Use a library, and force it into shape (build some glue code / shims / wrappers), rather building it
• If no choice, build it
• Stick to Rails concepts, way of working, naming when can't decide (naming is hard)

• axum for serving (controllers, router, requests)
• seaorm for data layer (ORM, ActiveModel, migrations)
• sidekiq-rs for background jobs
• mailers, views, tasks - homebrewed