Repository navigation
fix(weixin): show allowed image directories - #5296
Conversation
|
@qwen-code /triage |
6fbe8cb to
5ea3c91
Compare
|
@qwen-code /triage |
✅ Local runtime verification (real tmux, no mocks) — PR #5296Verdict: works as intended, no regressions — safe to merge. I reproduced the issue #4441 rejection path against the real What the PR doesWhen an image is rejected as outside the sandbox allowlist, the error message now also lists the allowed directories. Evidence1. Unit suite + build (PR head) 2. Real runtime — PR head (identical result against compiled Full message verbatim:
On this host 3. A/B against the pre-PR base commit (same harness, same filesystem): Direct before/after: base emits the bare message (exactly the #4441 complaint); head appends the directory list. 4. The new test has teeth — running the PR's regression test against the base code fails as expected: Edge cases checked (against the exact helper logic)
Minor / non-blocking nit: a workspace that is literally a drive root is built internally as Notes
🇨🇳 中文版(点击展开)✅ 本地真实运行验证(真实 tmux,无 mock)— PR #5296结论:行为符合预期,无回归,可以合并。 我在隔离的 tmux 会话中,针对真实的 这个 PR 做了什么当图片因不在沙箱白名单内而被拒绝时,错误信息现在会额外列出允许的目录。 证据1. 单元测试 + 构建(PR head) 2. 真实运行 — PR head(编译产物 完整信息原文:
本机 3. 与 PR 之前的 base 提交做 A/B 对比(同一套脚本、同一文件系统): 直接的前后对比:base 只给出干巴巴的信息(正是 #4441 的抱怨点);head 追加了目录列表。 4. 新增测试是有效的(有“牙齿”)——把 PR 的回归测试跑在 base 代码上会按预期失败: 边界情况核对(针对两个 helper 的精确逻辑)
非阻塞的小提示:如果工作区本身就是盘符根,内部会拼成 说明
|
Summary
Refs #4441
Test Plan
AI Assistance Disclosure
I used Codex to review the changes, sanity-check the implementation against existing patterns, and help spot potential edge cases.