I Would Never Use an ORM
summary
Argument is: ORMs hurt complex projects and replaceable code is better replaced by SQL plus tooling. Shows a concrete alternative: Platformatic DB built on Fastify that generates schema, migrations, REST/OpenAPI and live GraphQL subscription endpoints and lets custom JS resolvers run alongside. Cites a real scale pain: projects with 2000 models and 200 engineers that take about a minute to boot. Demo mostly works locally, cloud deploy attempts hit flaky conference Wi‑Fi but generation, queries and subscriptions function.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:00ORM and MVC problems
- 7:17Pareto Principle
- 10:51ORM failures in projects
- 13:22Better database requirements
- 14:06Platformatic DB
- 15:12Demo and setup
- 18:02Customizing APIs
- 19:28Relations and migrations
- 20:10Deploying to cloud
- 21:34Architecture and testing
- 21:56Q&A
related talks
25mI Would Never Use an ORM
45mORM vs GraphQL
46mBeyond Good and ORMs
30mSimplifying Data Access: ORMs for Web Development with JS and TS
20mYou don't need an ORM
35mNoSQL, No Server, No problem: Using ORM's in Javascript Applications
27mMaking Relational Cool Again (or: JavaScript on ACID)
01h 05mAdvanced Weapons and Tactics