Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/MODULE_GUIDE.md
Original file line number Diff line number Diff line change
Expand Up @@ -458,7 +458,7 @@ When in doubt, find the most specific relevant test first, then trace into imple

## Build and Verification Commands

Common commands from `CLAUDE.md`:
Common commands from `AGENTS.md`:

```bash
maturin develop
Expand Down
2 changes: 1 addition & 1 deletion docs/adr/0001-sequence-oriented-data-model.cn.md
Original file line number Diff line number Diff line change
Expand Up @@ -36,4 +36,4 @@ pandas/SQL 式的集合语义被否决为核心模型。`README.md` 的 FAQ 明
- `docs/USER_MODEL.cn.md`: “LTSeq 不是什么”
- `docs/DESIGN_SUMMARY.cn.md`: 概览
- `README.md`: Design Philosophy、FAQ
- `CLAUDE.md`: Architecture Overview
- `AGENTS.md`: Architecture overview
2 changes: 1 addition & 1 deletion docs/adr/0001-sequence-oriented-data-model.md
Original file line number Diff line number Diff line change
Expand Up @@ -36,4 +36,4 @@ Set-based semantics as in pandas/SQL were rejected as the core model. The `READM
- `docs/USER_MODEL.md`: "What LTSeq Is Not"
- `docs/DESIGN_SUMMARY.md`: §Overview
- `README.md`: Design Philosophy, FAQ
- `CLAUDE.md`: Architecture Overview
- `AGENTS.md`: Architecture overview
4 changes: 2 additions & 2 deletions docs/adr/0005-no-materialization-rule.cn.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@

该规则并不(也无法)覆盖所有返回表的 API。以下操作因设计或当前实现而物化:

**正确性所需的例外**(记录于 `CLAUDE.md`;不是捷径):
**正确性所需的例外**(记录于 `AGENTS.md`;不是捷径):

1. **物理位置操作**(`rvs`、`step`、带键 `distinct`)在分配行位置前,先把表快照为单个按序分区(collect → read_batch),因为对惰性多分区计划做无序/分区窗口不会保持输入顺序(`set_ops.rs::snapshot_single_partition`)。
2. **`fold()`** 逐行运行用户提供的 Python 回调 `fn(state, row)`;任意 Python 无法表达为 DataFusion 计划,逐行路径(`to_arrow()` → Python 行 → 累加 → 对同一 Arrow 表追加新列后 `from_arrow()`)是固有属性。docstring 标明这是非惰性慢路径(对比 Polars `cumulative_eval`)。
Expand All @@ -42,7 +42,7 @@

## 来源

- `CLAUDE.md`: No Materialization Rule(两个正确性例外)
- `AGENTS.md`: Key design patterns, No materialization rule(两个正确性例外)
- `docs/ARCHITECTURE.cn.md`: No-Materialization Rule、设计目标 #4
- `docs/DESIGN_SUMMARY.cn.md`: §5.4、§7.4
- `docs/USER_MODEL.cn.md`: 物化模型
Expand Down
4 changes: 2 additions & 2 deletions docs/adr/0005-no-materialization-rule.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ Benchmarking and design experience identified materialization as "the main archi

The rule does not (and cannot) cover every table-returning API. The following operations materialize by design or by current implementation:

**Correctness-required exceptions** (documented in `CLAUDE.md`; not shortcuts):
**Correctness-required exceptions** (documented in `AGENTS.md`; not shortcuts):

1. **Physical-position ops** (`rvs`, `step`, keyed `distinct`) snapshot the table into a single in-order partition (collect → read_batch) before assigning row positions, because an unordered/partitioned window over a lazy multi-partition plan does not preserve input order (`set_ops.rs::snapshot_single_partition`).
2. **`fold()`** runs a user-supplied Python callback `fn(state, row)` per row; arbitrary Python cannot be a DataFusion plan, so the row-wise path (`to_arrow()` → Python rows → accumulate → `from_arrow()` of the same Arrow table plus the new column) is inherent. Its docstring flags it as a non-lazy slow path (compare Polars `cumulative_eval`).
Expand All @@ -42,7 +42,7 @@ These live on the Rust side (Arrow batches, no Python row round-trips), but they

## Sources

- `CLAUDE.md`: No Materialization Rule (the two correctness exceptions)
- `AGENTS.md`: Key design patterns, No materialization rule (the two correctness exceptions)
- `docs/ARCHITECTURE.md`: No-Materialization Rule, Design Goals #4
- `docs/DESIGN_SUMMARY.md`: §5.4, §7.4
- `docs/USER_MODEL.md`: Materialization Model
Expand Down
2 changes: 1 addition & 1 deletion docs/adr/0007-lambda-dsl-schemaproxy.cn.md
Original file line number Diff line number Diff line change
Expand Up @@ -33,4 +33,4 @@ API 应当读起来像 Python(`t.filter(lambda r: r.age > 18)`),又不付
- `docs/USER_MODEL.cn.md`: 核心心智模型 #4、表达式模型
- `docs/MODULE_GUIDE.cn.md`: 表达式子系统导览
- `docs/api.cn.md`: §0、§11
- `CLAUDE.md`: Expression Transpilation
- `AGENTS.md`: Key design patterns, Expression transpilation
2 changes: 1 addition & 1 deletion docs/adr/0007-lambda-dsl-schemaproxy.md
Original file line number Diff line number Diff line change
Expand Up @@ -33,4 +33,4 @@ Capture, serialize, transpile:
- `docs/USER_MODEL.md`: Core Mental Model #4, Expression Model, Common Mistakes
- `docs/MODULE_GUIDE.md`: Expression Subsystem Tour, transpiler tour
- `docs/api.md`: §0, §11
- `CLAUDE.md`: Expression Transpilation
- `AGENTS.md`: Key design patterns, Expression transpilation
2 changes: 1 addition & 1 deletion docs/adr/0012-rust-thin-shell-python-mixins.cn.md
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@

## 来源

- `CLAUDE.md`: Key Design Patterns(PyO3 单 `#[pymethods]` 约束、mixin 组合)
- `AGENTS.md`: Key design patterns(PyO3 单 `#[pymethods]` 约束、mixin 组合)
- `docs/ARCHITECTURE.cn.md`: PyO3 边界设计、Rust/Python 包结构
- `docs/DESIGN_SUMMARY.cn.md`: §1.3、§1.4
- `docs/MODULE_GUIDE.cn.md`: `src/lib.rs`、模块导览
Expand Down
2 changes: 1 addition & 1 deletion docs/adr/0012-rust-thin-shell-python-mixins.md
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ Both sides of the boundary risk monolith files: PyO3 allows only **one `#[pymeth

## Sources

- `CLAUDE.md`: Key Design Patterns (PyO3 single `#[pymethods]` constraint, mixin composition)
- `AGENTS.md`: Key design patterns (PyO3 single `#[pymethods]` constraint, mixin composition)
- `docs/ARCHITECTURE.md`: PyO3 Boundary Design, Rust/Python Package Structure
- `docs/DESIGN_SUMMARY.md`: §1.3, §1.4
- `docs/MODULE_GUIDE.md`: `src/lib.rs`, module tours
Expand Down
Loading