Repository navigation
fix(schedule): 冷段细化贴段首，避免窗口被夹到段尾 - #7
Merged
Merged
Conversation
_refine 里 target 由 seg_start 替换 hour 得到，段首晚于偏好时刻时 target 加一天会落到段外，再被 latest_start 夹到 f_end-need， 整段空闲只用到最后 run_estimate。改为 target = seg_start。 docs: README 去掉与当前实现不符的描述并精简，pyproject 下限对齐 3.11
Reviewer's Guide本 PR 修复了空闲窗口细化在偏好时刻早于冷段段首时的起点计算错误，使调度能够利用完整冷段而非退化为段尾短窗口；同时纠正文档与当前空闲窗口实现的偏差，并将项目 Python 支持下限和锁定依赖统一到 3.11+。 Flow diagram for corrected idle-window refinementflowchart TD
A["冷段开始 seg_start"] --> B["计算偏好时刻 target"]
B --> C{"target < seg_start"}
C -->|是| D["target = seg_start"]
C -->|否| E["保留 target"]
D --> F["计算 latest_start"]
E --> F
F --> G["计算 start_ts"]
G --> H["利用冷段并预留 buffer 与运行时间"]
File-Level Changes
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey - I've found 3 issues
Prompt for AI Agents
Please address the comments from this code review:
## Individual Comments
### Comment 1
<location path="amrita_plugin_memory/rethinking/schedule.py" line_range="533" />
<code_context>
)
if target < seg_start:
- target += timedelta(days=1)
+ target = seg_start
# 起点尽量靠近偏好时刻，但必须落在空闲段内且留够运行空间
latest_start = max(f_start, f_end - need * 60)
</code_context>
<issue_to_address>
**Preferred run starts too early**
When a cold segment crosses midnight and the preferred hour falls after midnight within that segment, `_refine` assigns `seg_start` to `target`, so `start_ts` schedules the run from the segment head (for example, 22:00) instead of the preferred time (03:00), even though that time falls within the segment.
Keep the next-day preferred target when it falls within the cold segment instead of unconditionally assigning `seg_start`.
</issue_to_address>
### Comment 2
<location path="README.md" line_range="498" />
<code_context>
-| 代码质量 | Ruff + Pyright |
+| 组件 | 技术 |
+| ---------- | ----------------------------------------------------------- |
+| 运行时 | Python 3.11+ / NoneBot2 / AmritaCore |
+| 向量库 | ChromaDB（PersistentClient 或 HttpClient） |
+| 嵌入 | OpenAI Embedding / Ollama Embedding |
</code_context>
<issue_to_address>
**Installation rejects stated Python versions**
When a user follows the README runtime requirement on Python 3.14 or later, the README says Python 3.11+, but `pyproject.toml` limits `requires-python` to `<3.14`, so package installation rejects a runtime the README says is supported.
Make the README runtime requirement match `pyproject.toml`, or update the package metadata to support the versions stated in the README.
</issue_to_address>
### Comment 3
<location path="README.md" line_range="573" />
<code_context>
| `decay.sweep_interval_hours` | `24` | 分层清理周期（小时） |
| `decay.sweep_batch_limit` | `5000` | 单次清理最多扫描的记忆条数 |
-| `per_session_memory_limit` | `50` | 单个分区（uni_id）的记忆条数上限 |
+| `per_session_memory_limit` | `100` | 单个分区（uni_id）的记忆条数上限 |
| `memo_max_chars` | `1000` | L1 备忘录字数硬上限 |
-| `enabled` | `false` | 是否启用离线巩固循环 |
</code_context>
<issue_to_address>
**Config metadata still says 50**
When a consumer displays the `ConfigFile` field description, `ConfigFile.per_session_memory_limit` has a default of 100, but its Pydantic field description still says the default is 50; consumers that display the config metadata are given a value that conflicts with the README and runtime default.
Update the field description to say the default is 100.
</issue_to_address>Sourcery assessment
Needs a human reviewer. 2 findings to address first, and when the preferred hour falls before a selected idle segment, the scheduler will now start at the segment’s beginning instead of clamping the run toward its end; if that timing is wrong, the agent could send a user-facing message earlier than intended. Reverting restores the prior scheduling behavior, but cannot retract a message already sent.
Blocking findings: amrita_plugin_memory/rethinking/schedule.py:533, README.md:498
Sourcery review 指出 target=seg_start 会把 22:00-09:00 这种跨午夜段的起点 拉到段首 18:00/22:00，而 03:00 其实就在段内。改为次日偏好时刻仍在段内 （nxt < seg_end）时取 nxt，两次都不在段内才退到段首。 顺带修正 config 里 per_session_memory_limit 的描述（写的是 50，实际 100）， README 的 Python 版本改成 3.11-3.13 与 requires-python 的 <3.14 对齐。
Contributor
Author
|
三条都处理了，push 在 5c3a549。 Comment 1 — 跨午夜冷段起点被贴到段首：这条是对的，而且比我原来那个修改更严重。我复现了：只让 09:00–18:00 活跃，冷段是 18:00 到次日 09:00，原实现给出 修法不是回退成 if target < seg_start:
nxt = target + timedelta(days=1)
target = nxt if nxt < seg_end else seg_start三种情况都验过：跨午夜且 03:00 在段内 → 起点 03:00；冷段 04:00–08:00 且偏好时刻在段外 → 起点 04:00 用满 225 分钟；段内且不跨午夜 → 03:00 不变。Sourcery 提到的「可能比预期更早给用户发消息」正是修掉的行为，现在不会了。 Comment 2 — README 说 3.11+ 但 metadata 是 <3.14：改成 Comment 3 — ConfigFile 的字段描述写 50：改成 100。这个是我改文档时只改了 README、漏了代码里的 description，确实会误导看配置元信息的人。 回归验证：300 组随机不变量仍零失败，二分与暴力解一致，30 天密度下 200 次规划平均 3.1 ms，ruff 与 py_compile 干净。 |
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
概要
修
IdlePlanner._refine里的一处窗口起点夹取错误，顺带把 README 里跟当前实现不符的描述改掉并精简，requires-python下限对齐 3.11。调度修正
_refine先用seg_start.replace(hour=target_hour)算偏好时刻，段首晚于偏好时刻时target += timedelta(days=1)，于是 target 落到段外，随后被latest_start = max(f_start, f_end - need*60)夹到段尾 —— 整段空闲只用到最后那点。复现：只让 00:00–04:00 与 09:00–24:00 活跃，冷段是 04:00–08:00 一整块。
原实现给出
07:35–07:45（10 分钟），改后给出04:00–07:45（225 分钟）。影响不只在窗口长度：07:35 起跑意味着 DMN 顶在用户起床前，尾部 buffer 刚好被吃掉，
一旦超时就撞上用户第一句话，而这块本来有四个小时可以安排。
修法只有一行，
if target < seg_start: target = seg_start。验证
非强制时不落进活跃时段，零失败
merge_noise与二分结果对照暴力解，5 组随机样本一致ruff check/ruff format --check通过，全量py_compile通过文档
README 从 766 行压到 727 行，改动以删除和纠错为主：
per_session_memory_limit默认值 50 改成实际的 100SubconsciousState.payloadsubconscious_get_contextcurrent_date，不再是last_run/total_tokens那套.env表重复的嵌入配置表、三层模型的大图（改成三句话）、配置表里 18 行 idle 逐项（改成指向
SubconsciousConfig）依赖
requires-python从>=3.10提到>=3.11：chromadb 1.5.9 自身虽然写的是>=3.9，但它拉进来的 numpy 2.4 / onnxruntime 1.30 / sqlalchemy 2.1 都是
>=3.11，声明 3.10 会在解析阶段就失败。
uv.lock随之重解，移除 py3.10 分支（-782 行）。[tool.ruff] target-version保持py310未动 —— 提到 py311 会让 UP017 对timezone.utc批量报错，那是另一件事，不该混进来。Review 关注
有对应用例覆盖
uv.lock变动较大但都是 py3.10 分支清理，185 个包与本地环境一致Summary by Sourcery
修正冷段窗口起点规划并同步项目文档与 Python 依赖版本要求。
Bug Fixes:
Enhancements:
Build:
Documentation: