Make retired state a function of history alone by hardening CRDT logic

[TrueDoctor]

No description provided.

[cubic-dev-ai]

All reported issues were addressed across 10 files

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 5 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

2 issues found across 8 files (changes from recent commits).

Confidence score: 3/5

• In document/graph-storage/src/document.rs, marks that arrive before their delta can leave a stale hot operation behind; a later Session::merge may add the delta without retrying cleanup, allowing the operation to be retired again as a duplicate history delta. Retry cleanup when the delta arrives.
• In document/graph-storage/src/history.rs, drop_retired_hot_ops scans the history once per hot-log entry, so cleanup can become O(history × hot_log) and stall sessions as both grow. Build a timestamp set once per retain pass or index timestamps.

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="document/graph-storage/src/history.rs">

<violation number="1" location="document/graph-storage/src/history.rs:44">
P2: This scan runs once per hot-log entry in `drop_retired_hot_ops`, making cleanup O(history × hot_log) and potentially stalling sessions as both grow. Build a timestamp set once per retain pass or index timestamps.</violation>
</file>

<file name="document/graph-storage/src/document.rs">

<violation number="1" location="document/graph-storage/src/document.rs:156">
P2: When marks arrive before their delta, this filter retains the hot op; a later canonical `Session::merge` adds the delta without retrying cleanup. The stale op can then be retired again as a second history delta; retry this cleanup whenever history advances and refold after a drop.</violation>
</file>

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

P2: This scan runs once per hot-log entry in drop_retired_hot_ops, making cleanup O(history × hot_log) and potentially stalling sessions as both grow. Build a timestamp set once per retain pass or index timestamps.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At document/graph-storage/src/history.rs, line 44:

<comment>This scan runs once per hot-log entry in `drop_retired_hot_ops`, making cleanup O(history × hot_log) and potentially stalling sessions as both grow. Build a timestamp set once per retain pass or index timestamps.</comment>

<file context>
@@ -39,6 +39,11 @@ impl History {
 
+	/// Whether history holds a delta authored at `timestamp`.
+	pub fn contains_timestamp(&self, timestamp: TimeStamp) -> bool {
+		self.deltas.iter().any(|delta| delta.timestamp == timestamp)
+	}
+
</file context>

[cubic-dev-ai]

P2: When marks arrive before their delta, this filter retains the hot op; a later canonical Session::merge adds the delta without retrying cleanup. The stale op can then be retired again as a second history delta; retry this cleanup whenever history advances and refold after a drop.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At document/graph-storage/src/document.rs, line 156:

<comment>When marks arrive before their delta, this filter retains the hot op; a later canonical `Session::merge` adds the delta without retrying cleanup. The stale op can then be retired again as a second history delta; retry this cleanup whenever history advances and refold after a drop.</comment>

<file context>
@@ -142,18 +141,21 @@ impl Document {
+	/// ahead of the delta carrying the op into history, stranding its effect in the working registry.
+	fn drop_retired_hot_ops(&mut self) -> bool {
+		let before = self.hot_log.len();
+		self.hot_log.retain(|hot_op| !self.history.contains_timestamp(hot_op.timestamp));
+
+		before != self.hot_log.len()
</file context>