[shivamhwp]

[GPT-6.1 Sol][Codex] on behalf of shivam.

Closes #31167

Open Azure DevOps PRs use their creation date as Updated, so comments and pushes leave active PRs sorted among quiet ones. This change derives activity from threads and available iteration dates, shares reads, and applies the date to PR lists, summaries, details and action replies. Generic branch probes keep main's metadata-only behavior so unrelated fork candidates do not trigger activity reads.

Cold pages return with pending dates before activity reads settle. Background activity reads run four at a time, then refresh the affected list. Detail answers and action replies update loaded rows, and a five-minute per-PR projection prevents late cached answers from rolling them back. Mobile revalidates loaded continuation pages; the shared query helper now waits for the new answer when refresh begins inside an atom batch.

Verification

Head 1fafa5479e64f6ee40e470e0cf0a129385dd61b4, main base e89abadaee70e0d73fb19a5ce40a686b1a92310e.

fleet-job check 31209 --head 1fafa5479e64f6ee40e470e0cf0a129385dd61b4 passed check:changed against this base. Boat removed VM bx_2yjk6a93 after completion. It passed format, lint, knip, GraphQL, 37 package typechecks, tests in 12 files and React compiler checks in 9 files. GitHub checks and test output pass at this head. Independent source reviews resolved the branch-read fan-out and batched-refresh findings, with no remaining source finding. The 5bc independent source review passed 49 focused tests; its author passed 1,851 client-runtime tests. The follow-up at ff7 independently passed all 69 pull-request state tests and resolved the stale-detail observation and tracing findings. A separate review at 1fafa accepted the reference-priority correction and independently passed all 30 focused Azure CLI/source-provider tests. Separate evidence audits accepted the web, Android and fresh native macOS captures, with no remaining evidence finding.

The captures below were made at 5bc, before the ff7 stale-detail correction and 1fafa reference-priority correction. Independent review confirmed that they remain applicable to the unchanged accepted-detail and action paths. The reference-priority correction lets an interactive reference lookup bypass queued background activity reads, without changing metadata-only branch lists. Its existing queue test failed first and now verifies that reference #6 starts while four background reads remain held. The added stale-detail regression failed first when an equal-date older detail restored the old title and draft flag; ff7 rejects that observation before cached rows or mobile continuation callbacks receive it.

These captures use the production app and server with synthetic Azure CLI responses. They do not establish authenticated Azure behavior or service latency.

Fresh web comparison

Main e89: ordinary panel Refresh loads the new comment, while Updated stays at 3d and #42 remains below #43.

main-refresh.mp4

Head 5bc: the same open list/panel starts with #42 at 40m below #1040 at 20m. Changing only #42's comment and using ordinary panel Refresh shows the new comment, both dates at just now, and #42 above #1040 after the explicit refresh. The backend stays connected. No list Refresh is clicked during this recording; automatic list requests occur.

web-refresh.mp4

Physical Android captured at 5bc

On a POCO F4 running Android 14, the compatible development client fingerprint is 832d41716ea2dfac7350605db3fd07914f87d6ef; Metro served source at 5bc81a579d668bca604985711417e54ea60f534c. In one unchanged All PRs scope, cold continuation #1090 becomes dated automatically and moves first, without manual list Refresh or navigation during settlement. The fixture records automatic re-reads of all four loaded pages. The gate has an eight-second per-process cap; this demonstrates settlement, not service latency.

cold1090.mp4

A second recording starts with #1030 still retained beyond the first 30 rows. Changing only its comment date, pulling to refresh detail and using Back moves it first with a fresh date. Accessibility scroll actions return to the top without triggering list Refresh.

known1030.mp4

Both raw Android recordings were decoded and their pending, settled and final frames inspected. They retain variable frame timestamps and decoder timestamp warnings; no edited timing or latency claim is made. Detail-menu draft/ready passed at 354207a4bac043e01b958f7f70fe5dbf81b65dd5 and uses an unchanged command path.

Fresh native macOS comparison

Built main e89abadaee70e0d73fb19a5ce40a686b1a92310e and capture head 5bc81a579d668bca604985711417e54ea60f534c on shiv, Apple M6 Mac mini, macOS 27.0.1, Electron 44.10.1. Both build:desktop runs passed 5/5; all 14 changed non-mobile runtime sources matched emitted source-map contents. Each app used a disposable profile and server home with mock keychain. Compiled artifacts stayed unchanged during the tests.

Main loads the changed comment on ordinary panel Refresh, but its header and row stay at 3d and #42 stays below quiet #43.

main-pr42.mp4

Capture head 5bc starts with #42 at 2h below #1040 at 51m. The same comment-only refresh updates the comment, header and loaded row to 2m and moves #42 first. No manual list Refresh or navigation occurs during the recording. The decoded 16-second frame has the corrected state, shortly after the Refresh RPC at about 14.58 seconds.

head-pr42.mp4

The retained desktop row #1030 also moves first after changing only its comment date and choosing panel Refresh. Draft/ready replies preserve the authoritative date; Drafts only changes from empty to one row and back to empty as those actions settle.

head-pr1030.mp4

head-actions.mp4

head-filter.mp4

All five original recordings decoded fully and their trigger and end-state frames were inspected. MP4s copy the recorder streams without cuts or speed edits. Automatic base-list processes occurred in each window: main #42 3, head #42 6, #1030 6, draft/ready 6, draft filter 15. The list's 30-second stale time is not a configured periodic poll; these results establish explicit-refresh and signal-driven updates. Apps, profiles, test state and the native lock were cleaned up.

Scope and approval

This addresses the existing Updated field and its sorting and refresh consumers in #31167. The defect is an open PR retaining its creation date after comments or pushes. The contract and client changes carry that corrected date, represent pending reads truthfully and keep loaded rows consistent. Asynchronous reads preserve list responsiveness after a synchronous version showed a measured cold-list regression. There is no new setting or workflow.

This submission relies on CONTRIBUTING.md's focused obvious-bug exception. The changes serve that one correction across the existing clients; the guide judges purpose and behavioral impact without a line-count cutoff. No explicit maintainer approval is claimed. Maintainers retain the eligibility decision.

Costs and limits

An uncached page of N open PRs needs up to N thread processes, four at a time. List dates may be five minutes old, summary/detail one minute old; explicit Refresh invalidates them. The activity-cache target of 1,024 entries is soft while records are fresh or in flight. Azure's creation-based paging still limits sorting to fetched rows. Push dating relies on RefUpdate system threads unless an iteration read supplies a newer time.

Unverified: actual Azure service comment/push behavior and latency, signed distribution and real keychain. Synthetic integration is accepted for this task; no test environment was signed into a real account.

Implementation: Claude Opus 5.5 through T3 Code, with the stale-detail review correction by GPT-6.1 Sol through Codex. Investigation, verification and independent review: GPT-6.1 Sol through Codex in T3 Code.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	21.5 KiB	21.5 KiB	0 B (0.0%)	24.4 KiB	✅
Codex	Live turn messages	1	1	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.9 KiB	21.9 KiB	0 B (0.0%)	24.4 KiB	✅
Claude	Live turn messages	1	1	0 (0.0%)	8	✅

Other scenarios

All 64 scenario measurements match main.

All 64 scenario measurements

Provider	Scenario	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	short turn resume	WebSocket decoded	10.5 KiB	10.5 KiB	0 B (0.0%)	12.7 KiB	✅
Codex	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn running, cold open	HTTP decoded	281.7 KiB	281.7 KiB	0 B (0.0%)	302.7 KiB	✅
Codex	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Codex	long turn running, history page	HTTP decoded	92.4 KiB	92.4 KiB	0 B (0.0%)	107.4 KiB	✅
Codex	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Codex	long turn finished, cold open	HTTP decoded	381.5 KiB	381.5 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, socket cold open	WebSocket decoded	381.6 KiB	381.6 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, work page	HTTP decoded	186.2 KiB	186.2 KiB	0 B (0.0%)	200.2 KiB	✅
Codex	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Codex	idle thread resume	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell snapshot	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Codex	fleet shell subscribe	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Codex	fleet shell resume check	rows	0	0	0	0	✅
Codex	stale shell reload	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Codex	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Codex	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	stale todo list cold open	HTTP decoded	72.4 KiB	72.4 KiB	0 B (0.0%)	83.0 KiB	✅
Codex	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Codex	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅
Claude	short turn resume	WebSocket decoded	10.6 KiB	10.6 KiB	0 B (0.0%)	12.7 KiB	✅
Claude	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn running, cold open	HTTP decoded	283.3 KiB	283.3 KiB	0 B (0.0%)	302.7 KiB	✅
Claude	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Claude	long turn running, history page	HTTP decoded	93.4 KiB	93.4 KiB	0 B (0.0%)	107.4 KiB	✅
Claude	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Claude	long turn finished, cold open	HTTP decoded	384.2 KiB	384.2 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, socket cold open	WebSocket decoded	384.3 KiB	384.3 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, work page	HTTP decoded	188.1 KiB	188.1 KiB	0 B (0.0%)	200.2 KiB	✅
Claude	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Claude	idle thread resume	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell snapshot	HTTP decoded	43.0 KiB	43.0 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Claude	fleet shell subscribe	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Claude	fleet shell resume check	rows	0	0	0	0	✅
Claude	stale shell reload	HTTP decoded	43.0 KiB	43.0 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Claude	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Claude	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	stale todo list cold open	HTTP decoded	73.4 KiB	73.4 KiB	0 B (0.0%)	83.0 KiB	✅
Claude	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Claude	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅

Baseline: fcf50ff · PR result: 5b1e05e · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at 5b1e05e

Macroscope's review found this PR approvable — Azure DevOps PR details, summaries, and action replies now reflect recent thread activity. The additional read retains existing dates on failure and does not change remote PR actions or permissions.

^{You can add or adjust custom eligibility rules. Learn more.}

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

• Last reviewed commit: 5bc81a5
• Latest commit, not reviewed: 5b1e05e

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• Resume reviews
• Trigger review

Walkthrough

Azure DevOps records and decoders now provide pull-request activity dates. The CLI shares and caches activity reads. Server and client code carry date-read states through list refreshes and display pending or unavailable activity.

Changes: Azure DevOps activity dates

Layer / File(s)	Summary
Activity data and decoding packages/source-control-azure-devops/src/server/azureDevOpsPullRequests.ts, packages/source-control-azure-devops/src/server/azureDevOpsPullRequestJson.ts, related tests	Normalized records include repository location and revision data. Pull-request updatedAt uses the later available creation or closure time. Thread and iteration decoders return activity timestamps with comments or iterations.
Shared CLI activity reads packages/source-control-azure-devops/src/server/AzureDevOpsPullRequestCli.ts, packages/source-control-azure-devops/src/server/AzureDevOpsPullRequestCli.test.ts	The CLI caches activity, shares concurrent thread reads, limits background reads, and reports pending or unavailable dates with a settlement effect. Thread and push timestamps contribute to activity dates.
Provider dates and server settlement packages/source-control-azure-devops/src/server/AzureDevOpsPullRequestProvider.ts, packages/source-control-azure-devops/src/server/AzureDevOpsSourceControlProvider.ts, packages/source-control-core/src/server/PullRequestProvider.ts, apps/server/src/pullRequest/PullRequestService.ts, related tests	Providers apply activity dates to list, summary, detail, action, and individual change-request results. The service schedules deferred reads and advances list-cache revisions when activity changes.
Client refresh and date display packages/contracts/src/pullRequest.ts, packages/client-runtime/src/state/pullRequests.ts, packages/client-runtime/src/state/runtime.ts, apps/mobile/src/features/pull-requests/*, apps/mobile/src/state/pull-requests.ts, apps/web/src/components/pullRequest/*, apps/web/src/state/pullRequests.ts, related tests	Client state retains matching summary fields in lists and revalidates loaded continuation pages. Mobile and web rows display pending or unavailable indicators instead of relative dates when the date-read state is defined.

Estimated code review effort: 4 (Complex) | ~60 minutes

Change: Bug fix · Severity of issue fixed: Medium

Sequence Diagram(s)

sequenceDiagram
  participant ListClient as Pull-request list client
  participant PullRequestService
  participant AzureProvider as AzureDevOpsPullRequestProvider
  participant ActivityCLI as AzureDevOpsPullRequestCli
  ListClient->>PullRequestService: request pull-request list
  PullRequestService->>AzureProvider: list change requests
  AzureProvider->>ActivityCLI: request listed activity
  ActivityCLI-->>AzureProvider: return known date and settlement effect
  AzureProvider-->>PullRequestService: return rows and settlement effect
  PullRequestService-->>ListClient: return rows with pending state
  PullRequestService->>ActivityCLI: settle deferred activity reads
  PullRequestService-->>ListClient: notify when activity dates change

Loading

Possibly related issues

• pingdotgg/t3code#21888 — This change shares and limits Azure activity reads, while still adding thread and iteration reads for activity dating.

Possibly related PRs

• pingdotgg/t3code#29918 — Both changes affect Azure pull-request activity reads; this PR also adds activity timestamp caching.
• pingdotgg/t3code#28085 — Both changes use Azure thread-list behavior; this PR uses thread timestamps to date pull requests.
• pingdotgg/t3code#20056 — Both changes use Azure discussion and iteration activity timestamps.

Suggested labels: bugfix

Suggested reviewers: juliusmarminge, maria-rcks

Merge Risk: Low · previous review 5bc81

Looking up or checking out an Azure PR by reference may be briefly delayed while list activity reads are running. Settled dates may also trigger extra host reads. Neither blocks merge, but the lookup delay is worth fixing.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title clearly summarizes the main change: dating Azure DevOps pull requests by recent activity.
Linked Issues check		Issue #31167 requires Azure DevOps Updated dates and sorting to reflect PR activity, and the app to notice changes soon after. The changes derive activity dates from thread and iteration data, share a…
Out of Scope Changes check		The changes support issue #31167 across the Azure DevOps provider, server and client state, web, and mobile consumers. The shared query-helper change supports the activity-settlement refresh path. Rel…
Description check		The description explains the problem and change, links the issue, gives detailed verification results and limits, addresses scope and approval, and identifies the AI tools used. It does not use separa…

Full details: Linked Issues check

Explanation

Issue #31167 requires Azure DevOps Updated dates and sorting to reflect PR activity, and the app to notice changes soon after. The changes derive activity dates from thread and iteration data, share and bound reads, and apply dates to list, summary, detail, and action results. Pending reads settle into refreshed lists, and the client updates loaded rows from newer answers. The reported tests cover activity decoding, caching and settlement, pagination, stale answers, and refresh behavior. Reported checks pass.

Full details: Out of Scope Changes check

Explanation

The changes support issue #31167 across the Azure DevOps provider, server and client state, web, and mobile consumers. The shared query-helper change supports the activity-settlement refresh path. Related contracts, UI states, and tests implement or verify that behavior. The reviewed summary provides no evidence of an unrelated change.

Full details: Description check

Explanation

The description explains the problem and change, links the issue, gives detailed verification results and limits, addresses scope and approval, and identifies the AI tools used. It does not use separate "## Problem" and "## Change" headings, but it provides the required information in the opening section.

Finishing Touches

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Dismissing prior approval to re-evaluate fd1155e

Dismissing prior approval to re-evaluate 1fafa54

Dismissing prior approval to re-evaluate 67d8d5a

[shivamhwp]

[GPT-6.1 Sol][Codex] on behalf of shivam.

Handing this back after Julius’s source change in 67d8d5a. Before that update, required checks, exact-head Boat verification and independent reviews passed at 1fafa5479e64f6ee40e470e0cf0a129385dd61b4 against e89abadaee70e0d73fb19a5ce40a686b1a92310e. All review findings were resolved.

The PR body contains the audited synthetic web, physical Android and native macOS recordings with their original build attribution. Actual Azure service behavior remains unverified. The new main integration and refactor have not been verified by this automation thread; its earlier ready assessment applies to the prior head/base only. Stopping automated work and leaving further changes and readiness with Julius.

[juliusmarminge]

The automated pipeline stopped on this pull request and will not merge it:

the proof has no capture anyone can open:

• Comment activity advancing Updated dates, row ordering, and summary activity classification: no authenticated Azure DevOps repository was available.
• Source-branch push activity and ordinary panel Refresh: the same Azure access prerequisite was unavailable.
• Mobile continuation-page retention, refreshed detail dates, ordering, and explicit pending activity states: the reproduction could not be seeded.
• No public capture URLs exist because no valid reproduction captures were produced; ghdrop upload was not attempted.

It stays open for a person to finish or close.

[shivamhwp]

[GPT-6.1 Sol][Codex] on behalf of shivam.

Final test handoff to Julius. Using the production apps/server with synthetic Azure CLI responses, main e89abadaee70e0d73fb19a5ce40a686b1a92310e reproduced stale Updated dates and order after a comment-only panel Refresh in both web and native macOS. The refreshed comment appeared, but the header/list kept the creation date.

At capture head 5bc81a579d668bca604985711417e54ea60f534c, web and native macOS passed ordinary panel Refresh with corrected header/row dates and order, including retained #1030. Native macOS on shiv, macOS 27.0.1 / Electron 44.10.1, also passed draft/ready and draft-filter transitions. On a physical POCO F4 / Android 14, cold continuation #1090 automatically settled from pending to dated and reordered without manual refresh; retained #1030 passed detail Refresh, Back and reorder. Recordings were decoded, inspected and independently audited. The PR body retains all media and build attribution.

Evidence: main native reproduction, native corrected Refresh, Android cold continuation, Android retained-row Refresh, web corrected Refresh.

Later fail-first regressions covered stale detail rollback and reference lookups blocked behind background reads. Independent runs passed 69 client state tests and 30 Azure CLI/source-provider tests. At our last tested head 1fafa5479e64f6ee40e470e0cf0a129385dd61b4, exact-head Boat check:changed and required CI passed; all findings were resolved. These results do not verify Julius’s later commits or the current base. Actual authenticated Azure comment/push behavior and latency, signed distribution and real keychain remain unverified.

No tests, servers or devices remain running from this job. Temporary test artifacts were cleaned. Stopping automation here; Julius owns further testing and readiness.

Dismissing prior approval to re-evaluate 73e96f1

Dismissing prior approval to re-evaluate faa6918

Dismissing prior approval to re-evaluate 90d00b7

Dismissing prior approval to re-evaluate 5b1e05e