kimtth Goal: failures that are visible, contextual, and handled deliberately.
Use for:
- deciding how functions signal and propagate failure
- reviewing try/catch, error types, and recovery
- fixing swallowed errors and vague messages
Workflow:
- Distinguish expected failures from programmer bugs.
- Model expected errors explicitly (results or typed exceptions).
- Handle errors where you can act; otherwise propagate with context.
- Add context as errors cross boundaries, without losing the cause.
- Fail fast on invariant violations; do not limp on.
- Surface actionable messages; log details once, near the source.
Patterns:
- typed errors or result types over generic throws
- wrap-and-rethrow to add context, preserving the cause
- retry with backoff for transient failures only
- a single place that maps errors to user-facing responses
Rules:
- never swallow an error silently
- do not catch broadly just to continue; handle or rethrow
- include context, never the bare message
- distinguish retryable failures from permanent ones