Merge progress updates from root agents/bgl_todo.md

- Mark completed items in Phase 2 (range-based algorithms, MST, connectivity)
- Mark completed items in Phase 3 (MPL removal, adjacency_list, containers)
- Add detailed container selector implementation section (3.2.1)
- Mark completed items in Phase 4 (coroutines, parallel execution, validation)
- Mark completed items in Phase 5 (all testing phases)
- Update Phase 3 effort and timeline estimates

Author: pratzl

agents/bgl_todo.md

@@ -71,18 +71,18 @@ libs/headers---
 - [x] Support direct lambda/invocable for single-event use case
 
 ### 2.4 Range-Based Algorithm Variants
-- [ ] `breadth_first_search(g, initial_vertices_range, visitor)`
-- [ ] `depth_first_search(g, initial_vertices_range, visitor)`
-- [ ] Ensure algorithms accept `std::views::filter` results as input
+- [x] `breadth_first_search(g, initial_vertices_range, visitor)`
+- [x] `depth_first_search(g, initial_vertices_range, visitor)`
+- [x] Ensure algorithms accept `std::views::filter` results as input
 
 ### 2.5 Minimum Spanning Tree Algorithms
-- [ ] Modernize `kruskal_minimum_spanning_tree` (return edge range)
-- [ ] Modernize `prim_minimum_spanning_tree` (return edge range)
+- [x] Modernize `kruskal_minimum_spanning_tree` (return edge range)
+- [x] Modernize `prim_minimum_spanning_tree` (return edge range)
 
 ### 2.6 Connectivity Algorithms
-- [ ] Modernize `connected_components` (return component map or range)
-- [ ] Modernize `strong_components`
-- [ ] Modernize `topological_sort` (return sorted vertex range)
+- [x] Modernize `connected_components` (return component map or range)
+- [x] Modernize `strong_components`
+- [x] Modernize `topological_sort` (return sorted vertex range)
 
 ---
 
@@ -91,28 +91,67 @@ libs/headers---
 **Goal:** Modernize `adjacency_list` and other containers; remove Boost.MPL.
 
 ### 3.1 Replace Boost.MPL
-- [ ] Replace all `boost::mpl::if_` with `std::conditional_t`
-- [ ] Replace `boost::mpl::and_`, `boost::mpl::or_` with `&&`, `||` on `::value`
-- [ ] Replace `boost::mpl::bool_` with `std::bool_constant`
-- [ ] Remove `#include <boost/mpl/*.hpp>` from all headers
+- [x] Replace all `boost::mpl::if_` with `std::conditional_t`
+- [x] Replace `boost::mpl::and_`, `boost::mpl::or_` with `&&`, `||` on `::value`
+- [x] Replace `boost::mpl::bool_` with `std::bool_constant`
+- [x] Remove `#include <boost/mpl/*.hpp>` from all headers
+- **Note:** The modern/ directory was designed from scratch using C++20 features,
+  so it has no Boost.MPL dependencies. Uses std::conditional_t, concepts, etc.
 
 ### 3.2 Modernize adjacency_list
-- [ ] Reorder template parameters for better defaults
-- [ ] Support bundled properties via simple structs (designated initializer friendly)
-- [ ] Implement `g[v]` returning vertex property reference
-- [ ] Implement `g[e]` returning edge property reference
-- [ ] Ensure `add_vertex({.name = "A", .weight = 1.0})` works
+- [x] Reorder template parameters for better defaults
+- [x] Support bundled properties via simple structs (designated initializer friendly)
+- [x] Implement `g[v]` returning vertex property reference
+- [x] Implement `g[e]` returning edge property reference
+- [x] Ensure `add_vertex({.name = "A", .weight = 1.0})` works
+
+### 3.2.1 Container Selector Implementation ✅ COMPLETE
+**Status:** Full container selector infrastructure implemented with vecS, listS, setS support.
+
+**Completed:**
+- [x] Create `container_selectors.hpp` with all selector tags and `container_gen` mechanism
+- [x] Implement `vecS` selector with `std::vector` backing (index-based descriptors)
+- [x] Implement `listS` selector with `std::list` backing (iterator-based descriptors)
+- [x] Implement `setS` selector with `std::set` backing (iterator-based descriptors)
+- [x] Implement `container_gen` template specializations for all selectors
+- [x] Update adjacency_list template parameters: `adjacency_list<OutEdgeListS, VertexListS, DirectedS, VP, EP, GP>`
+- [x] Implement vertex removal for stable containers (listS, setS)
+- [x] Add `simple_adjacency_list` backward compatibility alias for old API
+- [x] Add comprehensive tests (`test_container_selectors.cpp`) covering:
+  - [x] Selector properties (is_sequence, is_associative, is_unique, etc.)
+  - [x] Container generation for all selector types
+  - [x] All selector combinations (vecS/listS/setS for both vertices and edges)
+  - [x] Direction variants (directed, undirected, bidirectional)
+  - [x] Iterator-based vs index-based descriptor handling
+
+**Selector Tags Implemented:**
+- `vecS` → `std::vector<T>` (sequence, not unique, index-based)
+- `listS` → `std::list<T>` (sequence, not unique, iterator-based)
+- `setS` → `std::set<T>` (associative, unique, iterator-based)
+- `mapS` → `std::map<size_t, T>` (mapped, unique)
+- `multisetS` → `std::multiset<T>` (associative, not unique)
+- `multimapS` → `std::multimap<size_t, T>` (mapped, not unique)
+- `hash_setS` → `std::unordered_set<T>` (unordered, unique)
+- `hash_mapS` → `std::unordered_map<size_t, T>` (unordered mapped, unique)
+- `hash_multisetS` → `std::unordered_multiset<T>` (unordered, not unique)
+- `hash_multimapS` → `std::unordered_multimap<size_t, T>` (unordered mapped, not unique)
+
+**Files Created/Modified:**
+- `include/bgl/modern/container_selectors.hpp` (NEW - 460 lines)
+- `include/bgl/modern/adjacency_list.hpp` (REWRITTEN - new template structure)
+- `test/test_container_selectors.cpp` (NEW - comprehensive tests)
+- All existing tests updated to use new API
 
 ### 3.3 Strong Typing for Descriptors
-- [ ] Implement `descriptor<Tag>` wrapper with `operator<=>` and hash
-- [ ] Define `vertex_tag`, `edge_tag`
-- [ ] Typedef `vertex_descriptor = descriptor<vertex_tag>` (opt-in or default)
-- [ ] Verify type safety prevents mixing vertex/edge descriptors
+- [x] Implement `descriptor<Tag>` wrapper with `operator<=>` and hash
+- [x] Define `vertex_tag`, `edge_tag`
+- [x] Typedef `vertex_descriptor = descriptor<vertex_tag>` (opt-in or default)
+- [x] Verify type safety prevents mixing vertex/edge descriptors
 
 ### 3.4 Modernize Other Containers
-- [ ] Review `adjacency_matrix` for range support
-- [ ] Review `compressed_sparse_row_graph` for `std::span` usage
-- [ ] Review `grid_graph`, `labeled_graph`, `subgraph`
+- [x] Implement `adjacency_matrix` with range support, O(1) edge lookup, bundled properties
+- [x] Implement `compressed_sparse_row_graph` with `std::span` for zero-copy adjacency access
+- [x] Implement `grid_graph` for N-dimensional implicit grid graphs
 
 ---
 
@@ -121,25 +160,25 @@ libs/headers---
 **Goal:** Add coroutines, parallel execution, algorithm composition.
 
 ### 4.1 Coroutine-Based Traversals (C++20/23)
-- [ ] Implement `generator<T>` (or use `std::generator` in C++23)
-- [ ] Implement `bfs_traverse(g, start)` as coroutine yielding vertices
-- [ ] Implement `dfs_traverse(g, start)` as coroutine yielding vertices
-- [ ] Support early termination via `co_return` / breaking out of range-for
+- [x] Implement `generator<T>` (or use `std::generator` in C++23)
+- [x] Implement `bfs_traverse(g, start)` as coroutine yielding vertices
+- [x] Implement `dfs_traverse(g, start)` as coroutine yielding vertices
+- [x] Support early termination via `co_return` / breaking out of range-for
 
 ### 4.2 Parallel Execution Policies
-- [ ] Identify parallelizable algorithms (BFS levels, independent component processing)
-- [ ] Add `std::execution::par` overloads for applicable algorithms
-- [ ] Benchmark parallel vs sequential performance
+- [x] Identify parallelizable algorithms (BFS levels, independent component processing)
+- [x] Add `std::execution::par` overloads for applicable algorithms
+- [x] Benchmark parallel vs sequential performance
 
 ### 4.3 Algorithm Composition
-- [ ] Design pipeline/composition API (e.g., `g | find_components() | filter_large()`)
-- [ ] Prototype with 2-3 composable operations
-- [ ] Evaluate integration with `std::ranges` pipelines
+- [x] Design pipeline/composition API (e.g., `g | find_components() | filter_large()`)
+- [x] Prototype with 2-3 composable operations
+- [x] Evaluate integration with `std::ranges` pipelines
 
 ### 4.4 Validation Framework
-- [ ] Implement `validate_graph(g)` runtime checks
-- [ ] Check for invalid descriptors, dangling edges, etc.
-- [ ] Return structured error information
+- [x] Implement `validate_graph(g)` runtime checks
+- [x] Check for invalid descriptors, dangling edges, etc.
+- [x] Return structured error information
 
 ---
 
@@ -148,26 +187,28 @@ libs/headers---
 **Goal:** Comprehensive test coverage and migration documentation.
 
 ### 5.1 Concept Tests
-- [ ] `static_assert` tests for all concepts against `adjacency_list`
-- [ ] `static_assert` tests for user-defined graph types
-- [ ] Negative tests (types that should *not* satisfy concepts)
+- [x] `static_assert` tests for all concepts against `adjacency_list`
+- [x] `static_assert` tests for user-defined graph types
+- [x] Negative tests (types that should *not* satisfy concepts)
 
 ### 5.2 Range Tests
-- [ ] Test `vertices(g)` with `std::ranges::for_each`
-- [ ] Test `out_edges(v, g)` with `std::views::filter`
-- [ ] Test algorithm outputs with `std::ranges::sort`, `std::ranges::copy`
+- [x] Test `vertices(g)` with `std::ranges::for_each`
+- [x] Test `out_edges(v, g)` with `std::views::filter`
+- [x] Test algorithm outputs with `std::ranges::sort`, `std::ranges::copy`
+- [x] Test complex range pipelines with multiple views
+- [x] Test range property concepts (forward_range, etc.)
 
 ### 5.3 Algorithm Tests
-- [ ] Port existing Boost.Graph test cases to new API
-- [ ] Add tests for structured returns
-- [ ] Add tests for lambda visitors
-- [ ] Add tests for named parameter structs
+- [x] Port existing Boost.Graph test cases to new API
+- [x] Add tests for structured returns
+- [x] Add tests for lambda visitors
+- [x] Add tests for named parameter structs
 
 ### 5.4 Performance Tests
-- [ ] Benchmark Dijkstra (old vs new)
-- [ ] Benchmark BFS/DFS (old vs new)
-- [ ] Benchmark graph construction
-- [ ] Ensure no performance regression
+- [x] Benchmark Dijkstra (old vs new)
+- [x] Benchmark BFS/DFS (old vs new)
+- [x] Benchmark graph construction
+- [x] Ensure no performance regression
 
 ### 5.5 Documentation
 - [ ] Write migration guide: old API → new API
@@ -212,7 +253,7 @@ libs/headers---
 |-------|----------|--------|------------------|
 | **Phase 1** | HIGH | Medium | Concepts, ranges, property map foundation |
 | **Phase 2** | HIGH | High | Modern algorithm APIs, visitors, named params |
-| **Phase 3** | MEDIUM-HIGH | Medium | Container cleanup, strong typing, MPL removal |
+| **Phase 3** | MEDIUM-HIGH | High | Container selectors (listS/setS/mapS), strong typing, MPL removal |
 | **Phase 4** | MEDIUM | High | Coroutines, parallelism, composition |
 | **Phase 5** | HIGH | Medium | Tests, docs, migration guide |
 | **Phase 6** | MEDIUM | Low | Cleanup, release |
@@ -242,9 +283,11 @@ Phase 6 (Release) after all others complete
 |-------|----------|------------|
 | Phase 1 | 4-6 weeks | 4-6 weeks |
 | Phase 2 | 6-8 weeks | 10-14 weeks |
-| Phase 3 | 4-6 weeks | 14-20 weeks |
-| Phase 4 | 4-6 weeks | 18-26 weeks |
-| Phase 5 | 3-4 weeks | 21-30 weeks |
-| Phase 6 | 1-2 weeks | 22-32 weeks |
+| Phase 3 | 6-8 weeks | 16-22 weeks |
+| Phase 4 | 4-6 weeks | 20-28 weeks |
+| Phase 5 | 3-4 weeks | 23-32 weeks |
+| Phase 6 | 1-2 weeks | 24-34 weeks |
 
-**Total estimated duration: 5-8 months**
+**Total estimated duration: 6-8.5 months**
+
+*Note: Phase 3 duration increased to account for container selector implementation (listS, setS, mapS, multisetS, hash_* variants).*