Good API design
summary
Argues that the worst mistakes are no API at all and an interface that was never designed, then lists common anti‑patterns: multiple incompatible APIs, fear of commitment, inconsistency, and extremes. Recommends concrete remedies: design around concepts not implementation, provide a layered API, use REST and JSON or platform conventions, ship runnable examples and testable docs, phase transitions (don’t big‑bang), keep a data dictionary, and avoid exposing raw DB schemas or pickle. Cites gmpy/gmpy2 and the Python 2→3 transition as cautionary examples.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:55API design overlooked
- 4:01API design anti-patterns
- 9:03No API consequences
- 17:25Good API design
- 21:41Internal protocols as APIs
- 26:17Fear of API commitment
- 31:12Too many APIs drawbacks
- 35:52API transitions and deprecation
- 40:12API design inconsistency
- 44:21Balanced API design
- 49:01Error messages and documentation in APIs
- 50:30Performance considerations for APIs
- 52:57Q&A
related talks
18mBuild a terrible API for people you hate
47mAPI Design Principles
01h 36mThe Most Important API Design Guideline - No, It's Not That One
01h 01mAPI Design – How to Do It Better (and How Not To)
27mAvoid These API Design Mistakes for Better Business Integration
51mTop 10 Rest API Design Pitfalls
41mAPI = Some REST and HTTP, Right? RIGHT?!
33mDesigning Sustainable APIs