团队规范
adr-process-error-handling
更新于 2026-10-26
ADR Process · 错误处理规则
Error-handling rules for ADR Process: classification, typed codes and single-point mapping to user outcomes.
- 技术栈
- ADR Process
- 主题
- 错误处理
- 条目
- 9 条
- 版本
- v0.2.0
✓ 应当这么做
- ✓ Classify failures once: expected (validation, not-found) vs unexpected (bugs, dependencies down)
- ✓ Return typed errors carrying a stable code plus human-readable context
- ✓ Map errors to user-visible outcomes at the outermost layer only
✕ 常见反模式
- ✕ String matching on error messages to decide control flow
- ✕ Swallowing an error to keep the happy path green
- ✕ Leaking stack traces or internal identifiers to end users
◎ 交付前验证
- ◎ Typecheck (`tsc --noEmit` or equivalent) passes with zero errors
- ◎ Lint passes and no rule was disabled inline
- ◎ Node/browser runtime compatibility checked against the project engines field
同分类条目
17 篇团队规范
Git Conventions 工程规则
Commit rules: conventional prefixes, scoped messages and revertible history
团队规范
Branching 工程规则
Branching rules: short-lived branches and a single integration source of truth
团队规范
Monorepo 工程规则
Monorepo rules: workspace boundaries, affected builds and lockfile discipline
团队规范
Dependency Upgrades 工程规则
Upgrade rules: pinned versions, changelog review and staged rollout
团队规范
ADR Process 工程规则
ADR rules: one decision per record with explicit consequences
团队规范
Local Dev Environment 工程规则
Dev environment rules: one command bootstrap and reproducible toolchains