Talento90 / StructuredMinimalApi

An example of a structured minimal api project

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Structured Minimal API

An example on how to structure a minmal API using Vertical Slice Architecture

What is Vertical Slice Architecture?

Vertical slice architecture is an approach for organising your code into features/vertical slices rather than organising by technical conerns (e.g. Controllers, Models, Services etc). Each slice will contain all the code which fullfills a feature or use case inside the application. One of the main benefits of VSA is the ability to structure each feature/slice independently, so each feature/slice can be as simple or complicated as it needs to be.

What does this API do?

This is an API for Twitter/X like social media, where users can make text posts, like and comment on posts and follow other users.

Some Important Design Decisions In This Project

  1. Each endpoint will define their own request/response contract
    • I have found trying to resuse DTOs can be a pain as soon as an edge case requires the DTO to diverge from the common structure. Rather than dealing with that pain later, bite the bullet now and create a seperate request/response DTO.
  2. Let DATA BE DATA
    • I'm not using Domain Driven Design, after using DDD on real world projects, I personally, don't like the approach (if it works for you, keep doing it, don't let me change that), I found I was always trying to search for the code that was actually doing the thing, or trying to figure out which Domain object the code belongs too rather than writing the code.
    • After experimenting with Golang I really like the approach of letting data just be simple structs and using methods/functions to operate over them.
    • So, in this project our Data Types will just be simple data buckets, no logic inside them, they are just there to represent our data. Some people call this an Anemic Domain Model
  3. No IRepository abstraction over EF Core
    • Controversial, I know. My take is EF Core is already a pretty solid abstraction over a database and covers 99.9% of use cases. Some people say "What about unit testing?", I think you shouldn't be mocking YOUR OWN database, if you need to test something which is reading/writing to your database, you should be writing an integration test
    • If EF Core isn't suitable for a specific feature/slice, we can use anything we want (e.g. Dapper) as each feature/slice is independent.

About

An example of a structured minimal api project

License:Apache License 2.0


Languages

Language:C# 98.5%Language:Dockerfile 1.5%