docs(backlog): mark B1-B5 shipped with PR references

docs/backlog.md

@@ -1,55 +1,39 @@
 # ZeroAlloc.StateMachine — Backlog
 
-Items not in v1. Implement once the core is stable and real-world usage patterns are understood.
+The original post-v1 graduation backlog (B1–B5) has fully shipped. New items can be added below as real-world friction surfaces.
 
 ---
 
-## B1 — Hierarchical / nested states
+## B1 — Hierarchical / nested states — ✅ shipped (PR #25)
 
-Sub-machines embedded within a parent state. When entering a composite state, the sub-machine starts at its own initial state. When exiting, the sub-machine is reset.
-
-**Scope:**
-- `[CompositeState<TState>(State = X, SubMachine = typeof(SubFsm))]` attribute
-- Generator: when entering composite state, delegate TryFire to sub-machine first
-- Sub-machine exit propagates trigger back to parent
+**Shipped:** sub-machines embedded within a parent state via `[CompositeState<TState>(State = X, SubMachine = typeof(SubFsm))]`. When entering a composite state, the generator delegates `TryFire` to the sub-machine first; sub-machine exit propagates the trigger back to the parent. See `CompositeStateAttribute.cs`, `CompositeStateModel.cs`, and `CompositeStateTests.cs`.
 
 ---
 
-## B2 — History states
-
-Re-enter the last active sub-state when re-entering a composite state (shallow history).
+## B2 — History states — ✅ shipped (PR #25)
 
-**Scope:**
-- `[HistoryState<TState>(State = X)]` attribute
-- Generator stores last active sub-state before exit; restores on re-entry
+**Shipped:** shallow-history support via `[HistoryState<TState>(State = X)]`. The generator stores the last active sub-state before exit and restores it on re-entry. See `HistoryStateAttribute.cs` and `HistoryStateModel.cs`.
 
 ---
 
-## B3 — Timeout transitions
+## B3 — Timeout transitions — ✅ shipped (PR #27)
 
-Automatically fire a trigger after a configurable duration without external `TryFire`.
-
-**Scope:**
-- `[Transition<S,T>(From = X, On = Y, To = Z, AfterMs = 5000)]`
-- Generator emits a `System.Threading.Timer` field, started on `OnEnter{State}`, cancelled on `OnExit{State}`
-- Timer callback calls `TryFire` — concurrent-safe
+**Shipped:** automatic trigger firing after a configurable duration via `[Transition<S,T>(From = X, On = Y, To = Z, AfterMs = 5000)]`. The generator emits a race-safe `System.Threading.Timer` field with `Interlocked.CompareExchange` lazy init, started on `OnEnter{State}`, cancelled on `OnExit{State}`. See `TimedTransitionTests.cs` for coverage including concurrent firing and disposal semantics.
 
 ---
 
-## B4 — Visual diagram export
-
-Generate a Mermaid or PlantUML state diagram from the declared transitions as part of the build.
+## B4 — Visual diagram export — ✅ shipped (PR #29)
 
-**Scope:**
-- Additional generator output: `{TypeName}.mermaid` alongside `.g.cs`
-- Opt-in via `[StateMachine(Diagram = true)]`
+**Shipped:** Mermaid state diagram emitted alongside `.g.cs` as `{TypeName}.mermaid`, opt-in via `[StateMachine(Diagram = true)]`. The companion fix in the same PR closed an initial-state arm gap (initial state's `OnEnter` arm now runs on construction). See `MermaidDiagramWriter.cs` and `InitialStateArmTests.cs`.
 
 ---
 
-## B5 — Per-trigger granularity for concurrent mode
+## B5 — Per-trigger granularity for concurrent mode — ✅ shipped (PR #27)
+
+**Shipped:** multiple independent state variables within one machine via `[StateMachineGroup]` + `[StateMachinePart]`. The generator emits one `long` field per part with independent CAS loops, each carrying its own transition table. See `StateMachineGroupAttribute.cs`, `StateMachinePartAttribute.cs`, `StateMachineGroupWriter.cs`, and `StateMachineGroupTests.cs`.
+
+---
 
-Support multiple independent state variables within one machine (e.g., a component with both an operational state and a connection state that evolve concurrently).
+## Out of scope (for now)
 
-**Scope:**
-- `[StateMachineGroup]` containing multiple `[StateMachinePart]` fields
-- Generator emits one `long` field per part with independent CAS loops
+The original B1–B5 set covered the immediate post-v1 graduation candidates. Future entries graduate from this section once real-world friction surfaces a concrete value/cost tradeoff. Speculative additions are deliberately omitted — file a new backlog item with the friction narrative when one emerges.