Nightmare Before Best Practices
summary
Four real failures: a payment flow that authorizes but never captures causes a €180,000 loss until a hastily run fix uses an undocumented refund endpoint and doubles the damage; a registration/voting system collapses under 1,000 attendees exposing unauthorized check‑ins and DOS from simultaneous votes; a photo API switch from location IDs to GPS ruins estimates and kills the project before coding; operational singletons like an SSH tunnel dependency and screen sessions break payments and newsletters. Fixes: specs up front, peer + automated review (Codacy/Cotesy), stress tests, deployment process and documentation.
AI-written summary · about
chapters (27)
- 0:00Introduction
- 1:07Who am I
- 1:45Todays topic
- 2:07The story
- 2:55The bank explains
- 4:48The script
- 6:04The bank
- 6:57The conference
- 7:39The visitor
- 9:03The anonymous tip
- 9:50Voting
- 10:42Security
- 11:04Travel
- 13:00Google Street View
- 14:22Wrong Time Estimation
- 14:53The Laptop
- 15:42Newsletter Platform
- 16:37Temporary Solution
- 18:22Best Practices
- 21:35Comments
- 21:50Asking Comments
- 22:08JavaScript
- 23:07Java Comments
- 23:50Regular Expressions
- 26:30Links
- 26:52Technical Standards
- 27:41Technical Specification
related talks
28mLearning from High Profile Failures
41m12 Factors of Pain and Suffering
39mPEBCAK
47mThreat Modelling Stories From The Trenches
46mApplication security from start to finish
06mWhat I Learned From A Dress, an On-Call Nightmare
58mSix Little Lines of Fail
01h 02mFive things every developer should know about building mission-critical systems