Recipe for an (Almost) Perfect SDK
summary
This talk describes Parrot’s four-year effort to build a robust drone SDK: use a tiny, stable public API with semantic versioning, require tests and API reviews, and keep implementation private until needed. The team rebuilt the SDK over two years using a private/public git branch workflow with git filter-branch, provided a simulator and Android Studio plugin that run and test apps, enforced annotations and non‑crashing behavior, and shipped example apps plus a clear license.
AI-written summary · about
chapters (24)
- 0:00SDK creation intro
- 2:54SDK goal and design
- 4:23Know target and simplify API
- 7:14Accuracy vs simplicity
- 9:02API consistency
- 10:52Platform consistency
- 14:01Avoid breaking changes
- 15:27Library code writing
- 17:17Logs and debugging
- 19:49Crash prevention rules
- 20:32Code documentation
- 22:19Annotations and tests
- 23:02Testing importance
- 24:29Usage examples
- 25:34Dependency management
- 27:49Performance optimization
- 28:53SDK release prep
- 31:06Branching and releases
- 34:42Beta distribution
- 36:10Distribution format
- 37:11Versioning and release notes
- 38:13User feedback
- 40:03SDK adoption tips
- 42:35Conclusion
related talks
53mHow not to fuckup with building SDK and make clients happy
34mSuccessfully publishing a tested, upgradeable and documented Open Source Library
09mDeveloping and supporting 12 SDKs with a team of three
32mBest Practices in SDK Development
32mHow to successfully shift from app development to SDK development
22mBest Practices When Building Mobile SDKs
28mLessons From Becoming an SDK Developer
27mSDK or Not