Skip to content

115 個拒絕站點中 82 處(78%)沒有任何測試斷言其訊息——#202 後的測試債盤點 #216

Description

@kiki830621

問題

#202 把 115 個字串拒絕改成 throw,全套件 383 tests 0 failures,但 verify 的 regression lens 對照 Tests/ 逐站點比對:104 個單行字面站點中 82 處(78%)沒有任何測試斷言其訊息——是既有債務,#202 沒有製造也沒有惡化它,但它讓「文字逐字元不變」這類契約只能靠 diff 比對而不是測試守。

DA 的 mutation probe 同時證明:對沒有 protocol 案例的站點,把 throw 改回 return 全套件仍綠——所以這份債的代價是靜默回歸,不是「測試少一點」。

前十個高風險站點(regression lens 的排序)

| # | 位置 | 工具 | 拒絕內容 |
| 1 | Server.swift:7084 | recover_from_autosave | E_NO_AUTOSAVE — document '…' has no known source path. |
| 2 | Server.swift:7088 | recover_from_autosave | E_NO_AUTOSAVE — no autosave file at '…'. |
| 3 | Server.swift:7183 | revert_to_disk | document '…' has no known source path to revert from |
| 4 | Server.swift:7219 | reload_from_disk | document '…' has no known source path to reload from |
| 5 | Server.swift:7519 | replace_text | invalid regex pattern '…' |
| 6 | Server.swift:7391/7393 | insert_paragraph | table index / table[..] cell (row,col) out of range |
| 7 | Server.swift:8698/8700 | insert_image_from_path | table[..] cell out of range |
| 8 | Server.swift:9963 | splice_omath_from_source | formatSpliceError(…) 全部六種 |
| 9 | Server.swift:13752 | protect_document | Invalid protection type. … |
| 10 | ReadbackTools.swift:293 / 315 | update_equation / delete_equation | expected paragraph at … |

要做的事

  • 以 tool 為單位補 protocol 案例(直接打 handleToolCall,斷言 isError == true + 訊息 needle),優先 wrap_caption_seq(10 站點)、insert_equation(19)、insert_image_from_path / insert_paragraph / insert_caption(各 8)。
  • RefusalIsErrorSweepTests 的 source sweep 互補:sweep 擋「形狀」,protocol 案例擋「站點」。

Source: surfaced during /idd-verify #202 (Step 5b) — regression §2, DA rulings; report: #213 (comment)

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions