Coding for Failure
summary
Advocates three practical rules: fail fast by adding sanity checks (ping external APIs before a multi-step purchase), fail loud by logging and using HopToad to aggregate/count exceptions instead of flooding inboxes, and zero trust components by validating every input (qmail-style process isolation) and checking exit codes for system calls and cron jobs. Shows concrete fixes: use save! or transactions for batch updates, rescue specific Net::HTTP/SMTP error groups, avoid inline rescue, and use a primary file to run one-off tasks on a single server.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:48Fail Whale
- 3:31Failure Points
- 5:35When to Relax
- 7:20Fail Fast
- 12:45Email Exceptions
- 14:58Zero Trust
- 16:15Exceptions
- 20:12Cron Jobs
- 22:58Inline Rescue
