| author |  Julius Marminge <julius0216@outlook.com> 2026-02-25 20:29:10-08:00 |
| committer |  Julius Marminge <jmarminge@gmail.com> 2026-02-26 13:20:36-08:00 |
| log | 96d610b1f82255acd0fc077917dde7a944033f23 |
| tree | eb19bb838d7f475b37dea15d35ab8b6b5ad2a460 |
| parent | 0c4c027e92bae27716d94d14e3531c390d9ecef1 |
more progress
5 files changed, 154 insertions(+), 6 deletions(-)
apps
server
.plans/16c-pr89-remediation-checklist.md+6-6
| ... | ... | @@ -53,8 +53,8 @@ Counts: active `53` (`valid=34`, `partially-valid=19`), closed-invalid `5` |
| 53 | 53 | - Threads: PRRT_kwDORLtfbc5wkPaL |
| 54 | 54 | - Audit note: SessionId-only routing can misassociate events under races/rebinds. |
| 55 | 55 | |
| 56 | - [ ] `C020` On `message.completed`, the message ID is added to the set and `thread.message.assistant.complete` is dispatched. On `turn.completed`, the same set is iterated and `thread.message.assistant.complete` is dispatched again for each ID—including already-completed ones. Consider removing message IDs from the set after dispatching on `message.completed`, or filtering out already-completed IDs before the `turn.completed` loop. | |
| 57 | - Status: `TODO` | |
| 56 | - [x] `C020` On `message.completed`, the message ID is added to the set and `thread.message.assistant.complete` is dispatched. On `turn.completed`, the same set is iterated and `thread.message.assistant.complete` is dispatched again for each ID—including already-completed ones. Consider removing message IDs from the set after dispatching on `message.completed`, or filtering out already-completed IDs before the `turn.completed` loop. | |
| 57 | - Status: `DONE` | |
| 58 | 58 | - Verdict: `partially-valid` |
| 59 | 59 | - Severity: `Medium` |
| 60 | 60 | - Area: `Runtime resilience and failure handling` |
| ... | ... | @@ -80,8 +80,8 @@ Counts: active `53` (`valid=34`, `partially-valid=19`), closed-invalid `5` |
| 80 | 80 | - Threads: PRRT_kwDORLtfbc5wyTaW, PRRT_kwDORLtfbc5wzli3 (+1 duplicate thread(s)) |
| 81 | 81 | - Audit note: Same unhandled rejection path remains in WS message handler. |
| 82 | 82 | |
| 83 | - [ ] `C042` Duplicated `resolveThreadWorkspaceCwd` across three files | |
| 84 | - Status: `TODO` | |
| 83 | - [x] `C042` Duplicated `resolveThreadWorkspaceCwd` across three files | |
| 84 | - Status: `DONE` | |
| 85 | 85 | - Verdict: `partially-valid` |
| 86 | 86 | - Severity: `Low` |
| 87 | 87 | - Area: `Runtime resilience and failure handling` |
| ... | ... | @@ -98,8 +98,8 @@ Counts: active `53` (`valid=34`, `partially-valid=19`), closed-invalid `5` |
| 98 | 98 | - Threads: PRRT_kwDORLtfbc5wnWwM, PRRT_kwDORLtfbc5w1C3-, PRRT_kwDORLtfbc5w1HGZ (+2 duplicate thread(s)) |
| 99 | 99 | - Audit note: Workspace CWD resolution duplication still present across modules. |
| 100 | 100 | |
| 101 | - [ ] `C044` Checkpoint reactor swallows diff errors silently for `turn.completed` | |
| 102 | - Status: `TODO` | |
| 101 | - [x] `C044` Checkpoint reactor swallows diff errors silently for `turn.completed` | |
| 102 | - Status: `DONE` | |
| 103 | 103 | - Verdict: `partially-valid` |
| 104 | 104 | - Severity: `Low` |
| 105 | 105 | - Area: `Runtime resilience and failure handling` |
apps/server/src/orchestration/Layers/CheckpointReactor.test.ts+48
| ... | ... | @@ -379,6 +379,54 @@ describe("CheckpointReactor", () => { |
| 379 | 379 | ).toBe("v2\n"); |
| 380 | 380 | }); |
| 381 | 381 | |
| 382 | it("appends capture failure activity when turn diff summary cannot be derived", async () => { | |
| 383 | const harness = await createHarness({ seedFilesystemCheckpoints: false }); | |
| 384 | const createdAt = new Date().toISOString(); | |
| 385 | ||
| 386 | await Effect.runPromise( | |
| 387 | harness.engine.dispatch({ | |
| 388 | type: "thread.session.set", | |
| 389 | commandId: CommandId.makeUnsafe("cmd-session-set-missing-baseline-diff"), | |
| 390 | threadId: ThreadId.makeUnsafe("thread-1"), | |
| 391 | session: { | |
| 392 | threadId: ThreadId.makeUnsafe("thread-1"), | |
| 393 | status: "ready", | |
| 394 | providerName: "codex", | |
| 395 | providerSessionId: asSessionId("sess-1"), | |
| 396 | providerThreadId: ProviderThreadId.makeUnsafe("provider-thread-1"), | |
| 397 | activeTurnId: null, | |
| 398 | lastError: null, | |
| 399 | updatedAt: createdAt, | |
| 400 | }, | |
| 401 | createdAt, | |
| 402 | }), | |
| 403 | ); | |
| 404 | ||
| 405 | harness.provider.emit({ | |
| 406 | type: "turn.completed", | |
| 407 | eventId: EventId.makeUnsafe("evt-turn-completed-missing-baseline"), | |
| 408 | provider: "codex", | |
| 409 | sessionId: asSessionId("sess-1"), | |
| 410 | createdAt: new Date().toISOString(), | |
| 411 | threadId: ProviderThreadId.makeUnsafe("provider-thread-1"), | |
| 412 | turnId: asTurnId("turn-missing-baseline"), | |
| 413 | status: "completed", | |
| 414 | }); | |
| 415 | ||
| 416 | await waitForEvent(harness.engine, (event) => event.type === "thread.turn-diff-completed"); | |
| 417 | const thread = await waitForThread( | |
| 418 | harness.engine, | |
| 419 | (entry) => | |
| 420 | entry.checkpoints.length === 1 && | |
| 421 | entry.activities.some((activity) => activity.kind === "checkpoint.capture.failed"), | |
| 422 | ); | |
| 423 | ||
| 424 | expect(thread.checkpoints[0]?.checkpointTurnCount).toBe(1); | |
| 425 | expect( | |
| 426 | thread.activities.some((activity) => activity.kind === "checkpoint.capture.failed"), | |
| 427 | ).toBe(true); | |
| 428 | }); | |
| 429 | ||
| 382 | 430 | it("captures pre-turn baseline from project workspace root when thread worktree is unset", async () => { |
| 383 | 431 | const harness = await createHarness({ |
| 384 | 432 | hasSession: false, |
apps/server/src/orchestration/Layers/CheckpointReactor.ts+8
| ... | ... | @@ -244,6 +244,14 @@ const make = Effect.gen(function* () { |
| 244 | 244 | deletions: file.deletions, |
| 245 | 245 | })), |
| 246 | 246 | ), |
| 247 | Effect.tapError((error) => | |
| 248 | appendCaptureFailureActivity({ | |
| 249 | sessionId: event.sessionId, | |
| 250 | turnId, | |
| 251 | detail: `Checkpoint captured, but turn diff summary is unavailable: ${error.message}`, | |
| 252 | createdAt: event.createdAt, | |
| 253 | }), | |
| 254 | ), | |
| 247 | 255 | Effect.catch((error) => |
| 248 | 256 | Effect.logWarning("failed to derive checkpoint file summary", { |
| 249 | 257 | threadId: thread.id, |
apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.test.ts+72
| ... | ... | @@ -258,6 +258,78 @@ describe("ProviderRuntimeIngestion", () => { |
| 258 | 258 | expect(message?.streaming).toBe(false); |
| 259 | 259 | }); |
| 260 | 260 | |
| 261 | it("does not duplicate assistant completion when message.completed is followed by turn.completed", async () => { | |
| 262 | const harness = await createHarness(); | |
| 263 | const now = new Date().toISOString(); | |
| 264 | ||
| 265 | harness.emit({ | |
| 266 | type: "turn.started", | |
| 267 | eventId: asEventId("evt-turn-started-for-complete-dedup"), | |
| 268 | provider: "codex", | |
| 269 | sessionId: asSessionId("sess-1"), | |
| 270 | createdAt: now, | |
| 271 | turnId: asProviderTurnId("turn-complete-dedup"), | |
| 272 | }); | |
| 273 | ||
| 274 | await waitForThread( | |
| 275 | harness.engine, | |
| 276 | (thread) => | |
| 277 | thread.session?.status === "running" && thread.session?.activeTurnId === "turn-complete-dedup", | |
| 278 | ); | |
| 279 | ||
| 280 | harness.emit({ | |
| 281 | type: "message.delta", | |
| 282 | eventId: asEventId("evt-message-delta-for-complete-dedup"), | |
| 283 | provider: "codex", | |
| 284 | sessionId: asSessionId("sess-1"), | |
| 285 | createdAt: now, | |
| 286 | turnId: asProviderTurnId("turn-complete-dedup"), | |
| 287 | itemId: asItemId("item-complete-dedup"), | |
| 288 | delta: "done", | |
| 289 | }); | |
| 290 | harness.emit({ | |
| 291 | type: "message.completed", | |
| 292 | eventId: asEventId("evt-message-completed-for-complete-dedup"), | |
| 293 | provider: "codex", | |
| 294 | sessionId: asSessionId("sess-1"), | |
| 295 | createdAt: now, | |
| 296 | turnId: asProviderTurnId("turn-complete-dedup"), | |
| 297 | itemId: asItemId("item-complete-dedup"), | |
| 298 | }); | |
| 299 | harness.emit({ | |
| 300 | type: "turn.completed", | |
| 301 | eventId: asEventId("evt-turn-completed-for-complete-dedup"), | |
| 302 | provider: "codex", | |
| 303 | sessionId: asSessionId("sess-1"), | |
| 304 | createdAt: now, | |
| 305 | turnId: asProviderTurnId("turn-complete-dedup"), | |
| 306 | status: "completed", | |
| 307 | }); | |
| 308 | ||
| 309 | await waitForThread( | |
| 310 | harness.engine, | |
| 311 | (thread) => | |
| 312 | thread.session?.status === "ready" && | |
| 313 | thread.session?.activeTurnId === null && | |
| 314 | thread.messages.some( | |
| 315 | (message) => message.id === "assistant:item-complete-dedup" && !message.streaming, | |
| 316 | ), | |
| 317 | ); | |
| 318 | ||
| 319 | const events = await Effect.runPromise( | |
| 320 | Stream.runCollect(harness.engine.readEvents(0)).pipe(Effect.map((chunk) => Array.from(chunk))), | |
| 321 | ); | |
| 322 | const completionEvents = events.filter((event) => { | |
| 323 | if (event.type !== "thread.message-sent") { | |
| 324 | return false; | |
| 325 | } | |
| 326 | return ( | |
| 327 | event.payload.messageId === "assistant:item-complete-dedup" && event.payload.streaming === false | |
| 328 | ); | |
| 329 | }); | |
| 330 | expect(completionEvents).toHaveLength(1); | |
| 331 | }); | |
| 332 | ||
| 261 | 333 | it("maps runtime.error into errored session state", async () => { |
| 262 | 334 | const harness = await createHarness(); |
| 263 | 335 | const now = new Date().toISOString(); |
apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts+20
| ... | ... | @@ -145,6 +145,22 @@ const make = Effect.gen(function* () { |
| 145 | 145 | } |
| 146 | 146 | }; |
| 147 | 147 | |
| 148 | const forgetAssistantMessageId = ( | |
| 149 | sessionId: ProviderSessionId, | |
| 150 | turnId: TurnId, | |
| 151 | messageId: MessageId, | |
| 152 | ) => { | |
| 153 | const key = providerTurnKey(sessionId, turnId); | |
| 154 | const existingIds = turnMessageIdsByTurnKey.get(key); | |
| 155 | if (!existingIds) { | |
| 156 | return; | |
| 157 | } | |
| 158 | existingIds.delete(messageId); | |
| 159 | if (existingIds.size === 0) { | |
| 160 | turnMessageIdsByTurnKey.delete(key); | |
| 161 | } | |
| 162 | }; | |
| 163 | ||
| 148 | 164 | const getAssistantMessageIdsForTurn = (sessionId: ProviderSessionId, turnId: TurnId) => { |
| 149 | 165 | return turnMessageIdsByTurnKey.get(providerTurnKey(sessionId, turnId)) ?? new Set<MessageId>(); |
| 150 | 166 | }; |
| ... | ... | @@ -257,6 +273,10 @@ const make = Effect.gen(function* () { |
| 257 | 273 | ...(turnId ? { turnId } : {}), |
| 258 | 274 | createdAt: now, |
| 259 | 275 | }); |
| 276 | ||
| 277 | if (turnId) { | |
| 278 | forgetAssistantMessageId(event.sessionId, turnId, assistantMessageId); | |
| 279 | } | |
| 260 | 280 | } |
| 261 | 281 | |
| 262 | 282 | if (event.type === "turn.completed") { |