宜博 / lemongraph

Commit 4cd87c41d1f7e47b66723891d884613dceeb81d8

Authored by bboonnee
1 parent 75162ac6

feat: task.execution_priority column + v_task_board display/edit column 

Requested by user after today's incident retrospective: a manually-assignable
execution-order hint across a project's tasks, shown as a new "执行顺序" column
on v_task_board. Serial steps are whole numbers (1, 2, 3); parallel siblings of
one step share the same integer part as decimals (2.1, 2.2). Purely advisory --
nothing in the scheduler/dispatch loop reads it, it's a human/Boss-Agent-facing
planning aid, not an enforcement mechanism.

- schema.py: task.execution_priority REAL (nullable), migrate_add_task_execution_priority
  (plain ADD COLUMN, no CHECK constraint). Two earlier CHECK-rebuild migrations that
  also touch task (migrate_relax_org_node_constraints, migrate_add_reviewing_planner_
  output_status) now call it as a prerequisite and include the column in their own
  rebuilt task_new, same "depends on a later additive column" pattern
  migrate_relax_org_node_constraints already used for migrate_add_work_graph_fork_columns.
- task_query.py: list_tasks/get_task expose it; new update_task_execution_priority
  (unlike update_task_goal_text, NOT gated to status='queued' -- it's a display hint,
  not a planning input, so it stays settable on in-flight/blocked tasks too).
- backend_server.py: PUT /api/tasks/:id now accepts execution_priority independently
  of goal_text on the same endpoint.
- frontend/app.js: new editable column on v_task_board, PUTs on change.

Also adds docs/todo/append-run-node-id-collision-across-org-nodes-requirements.md,
a third real finding from today's retrospective (task 911861fe: Coder and Validator's
suggested node catalogs both converge on node_type="test_script", so their
self-planned node_ids collide inside a shared merged work_graph_run -- PlanningFailedError,
uncaught by _ROUTABLE_EXCEPTIONS, falls into the same generic execution_error bucket
FR-7 already flagged).

813/813 src tests, 134/134 supervisor tests, frontend smoke passes.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Inline Side-by-side
Showing 8 changed files with 292 additions and 16 deletions
docs/todo/append-run-node-id-collision-across-org-nodes-requirements.md 0 → 100644
  View file @4cd87c4
  1 +# 追加规划的 node_id 跨 org_node 撞名——需求文档
  2 +
  3 +**日期**: 2026-08-13
  4 +**性质**: 需求文档（BRD/PRD 层面），不含详细技术设计。喂给 graphlemon 自己实现的 Goal，本次会话
  5 +不直接改代码。
  6 +**状态**: 草案，待确认后作为 Goal 提交。
  7 +
  8 +## 1. 背景
  9 +
  10 +`work_graph_plan.py::plan_work_graph(..., append_to_run_id=...)`（Phase 2c 的合并 run 架构，
  11 +Coder/Validator 各自往同一个共享 run 里追加自己规划的节点）在检测到新规划的 `node_id` 跟已有节点
  12 +撞名时会直接拒绝：
  13 +
  14 +```python
  15 +collision = existing_ids_set & set(new_node_ids)
  16 +if collision:
  17 + raise PlanningFailedError(
  18 + "append_to_run_id={} node_id collision with existing nodes: {}".format(append_to_run_id, collision)
  19 + )
  20 +```
  21 +
  22 +调用方（`plan_work_graph`）会把已有节点 id 列表写进 prompt，明确要求"不要重复用这些 id",但今天
  23 +（2026-08-13）真实任务 `911861fe` 撞上了这个拒绝：Validator 自己规划节点时选了 `n_test_script`
  24 +这个 id,而这个 id 在同一个 run 里已经被 Coder 用过了。
  25 +
  26 +根因不是"LLM 没读提示词"这么简单——`src/work_graph_plan.py` 里 Coder 和 Validator 的建议节点
  27 +目录（`_CODER_NODE_CATALOG`/`_VALIDATOR_NODE_CATALOG`）**都**给"写一个可机械验证的测试脚本"这类
  28 +节点建议了同一个 `node_type: "test_script"`。两个不同 org_node 在描述"概念上相似的东西"时，天然
  29 +会往同一个规范化命名（`n_test_script`）收敛——即使 prompt 里明确列出了"这些 id 已经被用了"，
  30 +自然语言指令本身也无法保证绝对不再犯。这不是这一次运气不好，是这个架构下的一个系统性碰撞概率
  31 +问题：**只要两个 org_node 的建议目录用相同/相似的 `node_type` 描述概念上相似的节点，撞名概率就
  32 +不是小概率事件**。
  33 +
  34 +`PlanningFailedError` 不在 `scheduler._ROUTABLE_EXCEPTIONS` 白名单里，所以这次撞名直接变成了
  35 +`blocked_pending_human`/`execution_error`，需要人工介入——跟
  36 +`docs/todo/boss-agent-autonomous-blocked-task-arbiter-requirements.md` FR-7 是同一类问题（系统
  37 +自己能分类的原因，不该都堆进笼统的兜底桶），但这条更进一步：**这类碰撞，理想情况下根本不该走到
  38 +需要人工介入这一步**，应该有更根本的预防或自动恢复机制。
  39 +
  40 +## 2. 目标
  41 +
  42 +让追加规划（`append_to_run_id`）在遇到 `node_id` 跨 org_node 撞名时，不需要人工介入就能正确处理
  43 +（预防撞名，或者撞了之后自动恢复），而不是直接硬失败变成 `blocked_pending_human`。
  44 +
  45 +## 3. 非目标
  46 +
  47 +1. **不裁定具体用哪种方案**（见 §7）——命名规范前缀（比如按 org_node 加前缀）、失败后带反馈重
  48 + 规划（跟 planner_reviewer 的重试模式类似）、自动重命名并改写内部引用，三种代价和可靠性都不同，
  49 + 需要设计阶段权衡。
  50 +2. **不改变 `_CODER_NODE_CATALOG`/`_VALIDATOR_NODE_CATALOG` 这两份目录已有的 `node_type` 分类
  51 + 本身**——"Coder 和 Validator 都需要一个 test_script 类型的节点"这个业务事实是对的、不用改，需要
  52 + 改的是"两者都可能选出同一个具体 `node_id` 字符串"这件事。
  53 +
  54 +## 4. 功能需求
  55 +
  56 +| # | 需求描述 | 优先级 |
  57 +|---|---|---|
  58 +| FR-1 | `append_to_run_id` 模式下发生 `node_id` 撞名时，不能直接终结为需要人工介入的 `blocked_pending_human`——需要预防（撞名概率显著降低）或自动恢复（撞名后不需要人工也能继续）之一，具体方案见 §7。 | 必须 |
  59 +| FR-2 | 如果最终方案里撞名之后仍然可能真正无法自动恢复（比如重试次数用尽），这种情况下打的原因也要专属、可区分，不能落入笼统的 `execution_error`（跟 Boss Agent 文档 FR-7 同一方向）。 | 必须 |
  60 +
  61 +## 5. 验收标准
  62 +
  63 +构造一个真实场景：两个不同 org_node（比如 Coder 和 Validator）在同一个共享 run 里，各自的
  64 +建议目录都指向同一个 `node_type`（如 `test_script`），触发一次真实的 `claude -p` 自主规划——验证
  65 +最终不会因为 `node_id` 撞名而进入需要人工介入的状态。
  66 +
  67 +## 6. 风险
  68 +
  69 +如果选择"自动重命名"类方案，需要非常小心地同步改写这个新节点自己内部的 `depends_on`/
  70 +`on_failure` 引用（如果它引用了自己），改写不完整会引入比"直接拒绝"更隐蔽的错误（图结构损坏，
  71 +但没有任何报错，节点执行时才会诡异地找不到依赖）——这条风险比"要不要做"更需要在设计阶段认真
  72 +对待。
  73 +
  74 +## 7. 开放问题
  75 +
  76 +- [ ] **具体用哪种方案**：
  77 + - (a) 命名规范前缀——要求所有自规划节点的 `node_id` 强制带 org_node 前缀（如
  78 + `validator_test_script`），从命名规则上让跨 org_node 撞名在正常情况下不可能发生。最简单、
  79 + 风险最低，但需要确认这不会破坏已有的 `_normalize_graph_manifest`/`_extract_graph_manifest`
  80 + 对 `node_id` 格式的既有假设。
  81 + - (b) 撞名后带反馈重规划——检测到撞名时不直接失败，而是把撞名信息作为反馈重新调用一次规划
  82 + （类似 `_review_planner_output_with_retries` 的重试模式），要求换一个不同的 id。需要设计
  83 + 重试上限。
  84 + - (c) 自动重命名——检测到撞名时自动加后缀改名并同步改写这次新规划里所有对该 id 的内部引用。
  85 + 见 §6 的风险，代价最高。
  86 +- [ ] 跟 `docs/todo/boss-agent-autonomous-blocked-task-arbiter-requirements.md`/
  87 + `docs/todo/agent-call-liveness-not-fixed-duration-timeout-requirements.md` 的实现顺序：
  88 + 三份文档都在处理"什么情况不该变成需要人工介入的 blocked_pending_human"，建议一起规划实现
  89 + 顺序,不要三个 Goal 各自独立设计出三套不一致的"是否需要人工介入"判断逻辑。
... ...
frontend/app.js
  View file @4cd87c4
... ... @@ -939,7 +939,7 @@ function renderTaskBoard() {
939 939 </div>
940 940 <table class="data-table">
941 941 <thead>
942   - <tr><th>task_id</th><th>goal_text</th><th>status</th><th>assigned_org_node</th><th>loop_round</th><th>updated_at</th></tr>
  942 + <tr><th>task_id</th><th>goal_text</th><th>status</th><th>assigned_org_node</th><th>loop_round</th><th>updated_at</th><th>执行顺序</th></tr>
943 943 </thead>
944 944 <tbody id="tb_rows"></tbody>
945 945 </table>
... ... @@ -984,7 +984,7 @@ function mountTaskBoard(queryParams) {
984 984 const resp = await fetch("/api/tasks?" + params.toString());
985 985 const tasks = await resp.json();
986 986 if (tasks.length === 0) {
987   - rowsEl.innerHTML = '<tr><td colspan="6">暂无任务</td></tr>';
  987 + rowsEl.innerHTML = '<tr><td colspan="7">暂无任务</td></tr>';
988 988 return;
989 989 }
990 990 rowsEl.innerHTML = tasks.map((t) => (
... ... @@ -999,6 +999,15 @@ function mountTaskBoard(queryParams) {
999 999 + "<td>" + t.assigned_org_node + "</td>"
1000 1000 + "<td>" + t.loop_round + "</td>"
1001 1001 + "<td>" + escapeHtml(t.updated_at) + "</td>"
  1002 + // 2026-08-13 (real incident retrospective): manually-assignable
  1003 + // execution-order hint -- serial steps are whole numbers (1, 2, 3),
  1004 + // parallel siblings of one step are decimals sharing the same
  1005 + // integer part (2.1, 2.2). Purely advisory (see task.execution_priority's
  1006 + // own schema comment) -- editing this never blocks/reorders real
  1007 + // execution, it's a planning aid rendered here for visibility.
  1008 + + '<td><input type="number" step="0.1" class="tb-priority-input" '
  1009 + + 'data-task-id="' + t.task_id + '" value="' + (t.execution_priority == null ? "" : t.execution_priority) + '" '
  1010 + + 'placeholder="—" style="width:4.5em;" onclick="event.stopPropagation()"></td>'
1002 1011 + "</tr>"
1003 1012 )).join("");
1004 1013 for (const row of rowsEl.querySelectorAll("tr[data-task-id]")) {
... ... @@ -1006,6 +1015,17 @@ function mountTaskBoard(queryParams) {
1006 1015 window.location.hash = "#/tasks/" + row.dataset.taskId;
1007 1016 });
1008 1017 }
  1018 + for (const input of rowsEl.querySelectorAll(".tb-priority-input")) {
  1019 + input.addEventListener("change", async () => {
  1020 + const raw = input.value.trim();
  1021 + const execution_priority = raw === "" ? null : Number(raw);
  1022 + await fetch("/api/tasks/" + encodeURIComponent(input.dataset.taskId), {
  1023 + method: "PUT",
  1024 + headers: { "Content-Type": "application/json" },
  1025 + body: JSON.stringify({ execution_priority }),
  1026 + });
  1027 + });
  1028 + }
1009 1029 }
1010 1030  
1011 1031 keywordEl.addEventListener("input", refresh);
... ...
src/backend_server.py
  View file @4cd87c4
... ... @@ -459,16 +459,29 @@ class _Handler(BaseHTTPRequestHandler):
459 459 self._send_json(400, error)
460 460 return
461 461 goal_text = payload.get("goal_text")
462   - if not goal_text:
463   - self._send_json(400, {"error": "INVALID_PARAM", "detail": "goal_text is required"})
  462 + # 2026-08-13: execution_priority is an independent, optional field on
  463 + # this same PUT -- unlike goal_text it's never gated to status=
  464 + # 'queued' (see update_task_execution_priority's own docstring), so
  465 + # it's handled as its own branch rather than folded into the
  466 + # goal_text-only contract this endpoint originally had.
  467 + has_priority = "execution_priority" in payload
  468 + if not goal_text and not has_priority:
  469 + self._send_json(400, {"error": "INVALID_PARAM", "detail": "goal_text or execution_priority is required"})
464 470 return
465 471 try:
466 472 with self.server.conn_lock:
467   - result = task_query.update_task_goal_text(self.server.conn, task_id, goal_text)
  473 + result = None
  474 + if goal_text:
  475 + result = task_query.update_task_goal_text(self.server.conn, task_id, goal_text)
  476 + if has_priority:
  477 + result = task_query.update_task_execution_priority(
  478 + self.server.conn, task_id, payload.get("execution_priority")
  479 + )
468 480 except task_query.TaskNotFoundError:
469 481 self._send_json(404, {"error": "NOT_FOUND"})
470   - except task_query.TaskNotEditableError as e:
471   - self._send_json(409, {"error": "TASK_NOT_EDITABLE", "detail": str(e)})
  482 + except (task_query.TaskNotEditableError, task_query.InvalidParamError) as e:
  483 + status = 409 if isinstance(e, task_query.TaskNotEditableError) else 400
  484 + self._send_json(status, {"error": "TASK_NOT_EDITABLE" if status == 409 else "INVALID_PARAM", "detail": str(e)})
472 485 else:
473 486 self._send_json(200, result)
474 487  
... ... @@ -1043,6 +1056,7 @@ def _ensure_schema(conn):
1043 1056 schema.migrate_relax_agent_profile_migrated_from_profile(conn)
1044 1057 schema.migrate_drop_work_graph_run_org_node(conn)
1045 1058 schema.migrate_add_concurrency_wait_event_type(conn)
  1059 + schema.migrate_add_task_execution_priority(conn)
1046 1060  
1047 1061  
1048 1062 class _Server(ThreadingHTTPServer):
... ...
src/schema.py
  View file @4cd87c4
... ... @@ -116,6 +116,15 @@ CREATE TABLE task (
116 116 blocked_reason TEXT,
117 117 -- H (2026-08-06): see skill.owner_org_node's comment above -- same relaxation.
118 118 assigned_org_node TEXT NOT NULL CHECK (assigned_org_node != ''),
  119 + -- 2026-08-13 (real incident retrospective): a manually-assignable
  120 + -- execution-order hint across a project's tasks -- NULLable (most
  121 + -- tasks never get one), REAL not INTEGER so parallel siblings of one
  122 + -- serial step can be expressed as decimals (e.g. step 2 splits into
  123 + -- 2.1/2.2, both meant to run concurrently, before step 3). Purely
  124 + -- advisory: nothing in the scheduler/dispatch loop reads this column
  125 + -- to gate execution -- it's a human/Boss-Agent-facing planning aid
  126 + -- shown on v_task_board, not an enforcement mechanism.
  127 + execution_priority REAL,
119 128 created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,
120 129 updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP
121 130 );
... ... @@ -494,11 +503,12 @@ def migrate_relax_org_node_constraints(conn):
494 503 can't be toggled mid-transaction), rebuild, PRAGMA foreign_key_check to
495 504 confirm nothing went dangling, PRAGMA foreign_keys=ON again.
496 505  
497   - Depends on migrate_add_work_graph_fork_columns() having already run
498   - (called here unconditionally, it's idempotent) -- work_graph_run's
499   - rebuild needs to know about D's parent_run_id/fork_of_node_exec_id
500   - columns whether or not this specific database has seen that migration
501   - yet.
  506 + Depends on migrate_add_work_graph_fork_columns() and (2026-08-13)
  507 + migrate_add_task_execution_priority() having already run (both called
  508 + here unconditionally, both idempotent) -- work_graph_run's rebuild
  509 + needs to know about D's parent_run_id/fork_of_node_exec_id columns, and
  510 + task's rebuild needs to know about execution_priority, whether or not
  511 + this specific database has seen either migration yet.
502 512  
503 513 Idempotent: each table's rebuild is skipped if its stored CREATE TABLE
504 514 text no longer contains the old enum CHECK, so a fresh
... ... @@ -506,6 +516,7 @@ def migrate_relax_org_node_constraints(conn):
506 516 same as migrate_expand_artifact_types's own convention.
507 517 """
508 518 migrate_add_work_graph_fork_columns(conn)
  519 + migrate_add_task_execution_priority(conn)
509 520 conn.commit()
510 521  
511 522 conn.execute("PRAGMA foreign_keys = OFF")
... ... @@ -548,6 +559,7 @@ def migrate_relax_org_node_constraints(conn):
548 559 loop_round INTEGER NOT NULL DEFAULT 0,
549 560 blocked_reason TEXT,
550 561 assigned_org_node TEXT NOT NULL CHECK (assigned_org_node != ''),
  562 + execution_priority REAL,
551 563 created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,
552 564 updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP
553 565 );
... ... @@ -736,10 +748,17 @@ def migrate_add_reviewing_planner_output_status(conn):
736 748 migrate_relax_org_node_constraints() precedent (that function's own
737 749 docstring explains the same PRAGMA foreign_keys=OFF/rebuild/
738 750 foreign_key_check/ON dance in more depth) instead of the leaf-table
739   - executescript() shortcut this originally (wrongly) copied."""
  751 + executescript() shortcut this originally (wrongly) copied.
  752 +
  753 + Depends on (2026-08-13) migrate_add_task_execution_priority() having
  754 + already run (called here unconditionally, it's idempotent) -- same
  755 + "rebuild needs to know about a later additive column" reasoning as
  756 + migrate_relax_org_node_constraints()'s own dependency on
  757 + migrate_add_work_graph_fork_columns()."""
740 758 row = conn.execute("SELECT sql FROM sqlite_master WHERE type='table' AND name='task'").fetchone()
741 759 if row is None or "'reviewing_planner_output'" in row[0]:
742 760 return
  761 + migrate_add_task_execution_priority(conn)
743 762 conn.execute("PRAGMA foreign_keys = OFF")
744 763 try:
745 764 conn.executescript(
... ... @@ -761,6 +780,7 @@ def migrate_add_reviewing_planner_output_status(conn):
761 780 loop_round INTEGER NOT NULL DEFAULT 0,
762 781 blocked_reason TEXT,
763 782 assigned_org_node TEXT NOT NULL CHECK (assigned_org_node != ''),
  783 + execution_priority REAL,
764 784 created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,
765 785 updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP
766 786 );
... ... @@ -860,6 +880,19 @@ def migrate_add_concurrency_wait_event_type(conn):
860 880 conn.commit()
861 881  
862 882  
  883 +def migrate_add_task_execution_priority(conn):
  884 + """2026-08-13 (real incident retrospective): adds task.execution_priority
  885 + on an ALREADY-POPULATED database. Plain `ALTER TABLE ... ADD COLUMN` (no
  886 + CHECK constraint involved, no create-copy-drop-rename dance needed) --
  887 + same pattern as migrate_add_work_graph_fork_columns. Idempotent and safe
  888 + to call unconditionally, including against a freshly create_schema()'d
  889 + database (which already has the column)."""
  890 + cols = {row[1] for row in conn.execute("PRAGMA table_info(task)").fetchall()}
  891 + if "execution_priority" not in cols:
  892 + conn.execute("ALTER TABLE task ADD COLUMN execution_priority REAL")
  893 + conn.commit()
  894 +
  895 +
863 896 def migrate_drop_work_graph_run_org_node(conn):
864 897 """Org Graph unification Phase 2c §2.1/§2.2 (2026-08-12, user-confirmed
865 898 2026-08-08): drops work_graph_run.org_node on an ALREADY-POPULATED
... ...
src/task_query.py
  View file @4cd87c4
... ... @@ -30,7 +30,7 @@ def list_tasks(conn, project_id, status=None, org_node=None, keyword=None, sort=
30 30 reaches the SQL text itself, only a pre-approved column name."""
31 31 query = (
32 32 "SELECT task_id, goal_text, task_type, status, assigned_org_node, loop_round, "
33   - "created_at, updated_at FROM task WHERE project_id = ?"
  33 + "created_at, updated_at, execution_priority FROM task WHERE project_id = ?"
34 34 )
35 35 params = [project_id]
36 36 if status:
... ... @@ -54,6 +54,7 @@ def list_tasks(conn, project_id, status=None, org_node=None, keyword=None, sort=
54 54 {
55 55 "task_id": r[0], "goal_text": r[1], "task_type": r[2], "status": r[3],
56 56 "assigned_org_node": r[4], "loop_round": r[5], "created_at": r[6], "updated_at": r[7],
  57 + "execution_priority": r[8],
57 58 }
58 59 for r in rows
59 60 ]
... ... @@ -88,6 +89,26 @@ def update_task_goal_text(conn, task_id, goal_text):
88 89 return get_task(conn, task_id)
89 90  
90 91  
  92 +def update_task_execution_priority(conn, task_id, execution_priority):
  93 + """2026-08-13 (real incident retrospective): unlike goal_text, this is
  94 + NOT gated to status='queued' -- execution_priority is a scheduling/
  95 + display hint (v_task_board's new column), not a planning input that
  96 + invalidates work already done, so it should stay settable regardless of
  97 + a task's current status (including a task that's already blocked or
  98 + mid-flight). None clears it back to unset."""
  99 + row = conn.execute("SELECT status FROM task WHERE task_id = ?", (task_id,)).fetchone()
  100 + if row is None:
  101 + raise TaskNotFoundError(task_id)
  102 + if execution_priority is not None and not isinstance(execution_priority, (int, float)):
  103 + raise InvalidParamError("execution_priority must be a number or null, got {!r}".format(execution_priority))
  104 + conn.execute(
  105 + "UPDATE task SET execution_priority = ?, updated_at = CURRENT_TIMESTAMP WHERE task_id = ?",
  106 + (execution_priority, task_id),
  107 + )
  108 + conn.commit()
  109 + return get_task(conn, task_id)
  110 +
  111 +
91 112 def delete_task(conn, task_id):
92 113 row = conn.execute("SELECT status FROM task WHERE task_id = ?", (task_id,)).fetchone()
93 114 if row is None:
... ... @@ -115,7 +136,8 @@ def delete_task(conn, task_id):
115 136 def get_task(conn, task_id):
116 137 row = conn.execute(
117 138 "SELECT task_id, project_id, goal_text, task_type, origin_ref, created_via, status, route, "
118   - "loop_round, blocked_reason, assigned_org_node, created_at, updated_at FROM task WHERE task_id = ?",
  139 + "loop_round, blocked_reason, assigned_org_node, created_at, updated_at, execution_priority "
  140 + "FROM task WHERE task_id = ?",
119 141 (task_id,),
120 142 ).fetchone()
121 143 if row is None:
... ... @@ -125,7 +147,7 @@ def get_task(conn, task_id):
125 147 "task_id": row[0], "project_id": row[1], "goal_text": row[2], "task_type": row[3],
126 148 "origin_ref": row[4], "created_via": row[5], "status": row[6], "route": row[7],
127 149 "loop_round": row[8], "blocked_reason": row[9], "assigned_org_node": row[10],
128   - "created_at": row[11], "updated_at": row[12],
  150 + "created_at": row[11], "updated_at": row[12], "execution_priority": row[13],
129 151 }
130 152  
131 153 artifacts = conn.execute(
... ...
src/test_backend_server.py
  View file @4cd87c4
... ... @@ -507,6 +507,21 @@ class TaskCrudRouteTest(unittest.TestCase):
507 507 self.assertEqual(status, 409)
508 508 self.assertEqual(body["error"], "TASK_NOT_EDITABLE")
509 509  
  510 + def test_put_execution_priority_works_regardless_of_status(self):
  511 + # 2026-08-13 (real incident retrospective): unlike goal_text, this
  512 + # must succeed even on a non-queued task.
  513 + self.server.conn.execute("UPDATE task SET status = 'coding' WHERE task_id = 'task-1'")
  514 + self.server.conn.commit()
  515 + status, body = self._put("/api/tasks/task-1", {"execution_priority": 2.1})
  516 + self.assertEqual(status, 200)
  517 + self.assertEqual(body["execution_priority"], 2.1)
  518 +
  519 + def test_get_list_includes_execution_priority(self):
  520 + self._put("/api/tasks/task-1", {"execution_priority": 3})
  521 + status, body = self._get("/api/tasks?project_id=proj-1")
  522 + self.assertEqual(status, 200)
  523 + self.assertEqual(body[0]["execution_priority"], 3)
  524 +
510 525 def test_delete_removes_queued_task(self):
511 526 status, body = self._delete("/api/tasks/task-1")
512 527 self.assertEqual(status, 200)
... ...
src/test_schema.py
  View file @4cd87c4
... ... @@ -568,6 +568,58 @@ class MigrateAddWorkGraphForkColumnsTest(unittest.TestCase):
568 568 fresh.close()
569 569  
570 570  
  571 +class MigrateAddTaskExecutionPriorityTest(unittest.TestCase):
  572 + """2026-08-13 (real incident retrospective): real upgrade path for an
  573 + already-populated database whose task table predates
  574 + execution_priority."""
  575 +
  576 + def setUp(self):
  577 + self.conn = schema.connect(":memory:")
  578 + old_sql = re.sub(
  579 + r",\n( --[^\n]*\n)* execution_priority REAL,\n",
  580 + ",\n",
  581 + schema.SCHEMA_SQL,
  582 + )
  583 + self.assertNotIn("execution_priority", old_sql) # sanity: the substitution actually matched
  584 + self.conn.executescript(old_sql)
  585 + self.conn.commit()
  586 + self.conn.execute(
  587 + "INSERT INTO project (project_id, name, target_folder_path, has_git_repo, status) "
  588 + "VALUES ('proj-1', 'P', '/tmp/proj-1', 1, 'active')"
  589 + )
  590 + self.conn.execute(
  591 + "INSERT INTO task (task_id, project_id, goal_text, task_type, status, assigned_org_node) "
  592 + "VALUES ('task-1', 'proj-1', 'g', 'goal', 'queued', 'planner')"
  593 + )
  594 + self.conn.commit()
  595 +
  596 + def tearDown(self):
  597 + self.conn.close()
  598 +
  599 + def test_migration_preserves_existing_rows_and_adds_nullable_column(self):
  600 + schema.migrate_add_task_execution_priority(self.conn)
  601 + row = self.conn.execute(
  602 + "SELECT status, execution_priority FROM task WHERE task_id = 'task-1'"
  603 + ).fetchone()
  604 + self.assertEqual(row, ("queued", None))
  605 +
  606 + def test_idempotent_on_a_database_already_migrated(self):
  607 + schema.migrate_add_task_execution_priority(self.conn)
  608 + schema.migrate_add_task_execution_priority(self.conn) # must not raise
  609 + count = self.conn.execute("SELECT COUNT(*) FROM task").fetchone()[0]
  610 + self.assertEqual(count, 1)
  611 +
  612 + def test_noop_on_a_freshly_created_database(self):
  613 + fresh = schema.connect(":memory:")
  614 + try:
  615 + schema.create_schema(fresh)
  616 + schema.migrate_add_task_execution_priority(fresh) # must not raise
  617 + cols = {row[1] for row in fresh.execute("PRAGMA table_info(task)").fetchall()}
  618 + self.assertIn("execution_priority", cols)
  619 + finally:
  620 + fresh.close()
  621 +
  622 +
571 623 _OLD_WORK_GRAPH_RUN_BLOCK = """-- 1.3 work_graph_run
572 624 CREATE TABLE work_graph_run (
573 625 run_id TEXT PRIMARY KEY,
... ...
src/test_task_query.py
  View file @4cd87c4
... ... @@ -340,6 +340,37 @@ class UpdateTaskGoalTextTest(TaskQueryTestBase):
340 340 task_query.update_task_goal_text(self.conn, "task-1", "new text")
341 341  
342 342  
  343 +class UpdateTaskExecutionPriorityTest(TaskQueryTestBase):
  344 + """2026-08-13 (real incident retrospective): unlike goal_text, this is
  345 + NOT gated to status='queued' -- it's a scheduling/display hint, not a
  346 + planning input."""
  347 +
  348 + def test_in_progress_task_can_still_be_edited(self):
  349 + self._insert_task("task-1", status="coding")
  350 + result = task_query.update_task_execution_priority(self.conn, "task-1", 2.1)
  351 + self.assertEqual(result["execution_priority"], 2.1)
  352 +
  353 + def test_serial_integer_value(self):
  354 + self._insert_task("task-1", status="queued")
  355 + result = task_query.update_task_execution_priority(self.conn, "task-1", 1)
  356 + self.assertEqual(result["execution_priority"], 1)
  357 +
  358 + def test_none_clears_it(self):
  359 + self._insert_task("task-1", status="queued")
  360 + task_query.update_task_execution_priority(self.conn, "task-1", 3)
  361 + result = task_query.update_task_execution_priority(self.conn, "task-1", None)
  362 + self.assertIsNone(result["execution_priority"])
  363 +
  364 + def test_nonexistent_task_raises_not_found(self):
  365 + with self.assertRaises(task_query.TaskNotFoundError):
  366 + task_query.update_task_execution_priority(self.conn, "no-such-task", 1)
  367 +
  368 + def test_non_numeric_value_raises_invalid_param(self):
  369 + self._insert_task("task-1", status="queued")
  370 + with self.assertRaises(task_query.InvalidParamError):
  371 + task_query.update_task_execution_priority(self.conn, "task-1", "not a number")
  372 +
  373 +
343 374 class DeleteTaskTest(TaskQueryTestBase):
344 375 def test_queued_task_can_be_deleted(self):
345 376 self._insert_task("task-1", status="queued")
... ...