PaperClipAI

Jannik/PaperClipAI

0 Fork 0

mirror of https://github.com/paperclipai/paperclip.git synced 2026-11-05 03:51:54 +01:00

Commit Graph

Author	SHA1	Message	Date
DottaandPaperclip	dabb8c5daf	test: remove repeated setup and strengthen regression proof (#16022 ) ## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Its tests protect company boundaries, approvals, durable tasks, credentials and the board UI. > - Several owner suites repeat large database fixtures and render frames. > - Some tests repeat stronger proof or pass when the behavior they describe is broken. > - This pull request shares exact setup, removes proven duplicate cases and repairs weak observations. > - It preserves distinct security, lifecycle, concurrency and user-facing contracts. > - The benefit is 5,027 fewer test lines and stronger regression checks. ## Linked Issues or Issue Description Refs #15978. This continues the merged test-audit cleanup in [PR #15978](https://github.com/paperclipai/paperclip/pull/15978), which removed 11,976 test/support lines. The savings in this pull request are additional. Related open work: [PR #13460](https://github.com/paperclipai/paperclip/pull/13460) changes legacy OAuth parent/version lock ordering. This pull request retains those production paths and the existing lock-order test. Its secrets changes compact fixtures and repair the two AWS wrapper exclusion observations. Scoped title searches found no open duplicate audit or fixture-consolidation PR and no matching audit issue. The problem is described here using the enhancement template. What existing behavior does this improve? The maintenance and regression value of the existing server, UI, CLI, shared and wiki-plugin test suites. Subsystem affected Cross-cutting test code. Two UI modules also lose unused test inspection exports. Current behavior Owner tests repeat database rows, service inputs, HTTP setup and render frames. Some cases repeat stronger assertions. Two AWS lock tests record only cleanup events, so an AWS provider write can happen inside the held lock without failing those observations. Several other negative tests accept or do not observe the wrong result. Worktree fixtures also use a temporary path alias that conflicts with the existing canonical-path checks on macOS. Proposed behavior Use small local helpers that preserve exact inputs, fresh allocations, callbacks, query order and await positions. Remove only cases with named stronger keepers. Tighten negative assertions and observe actual AWS provider entry while a confirmed lock holder remains parked. Resolve temporary fixture roots to their canonical paths. Reason and benefit Remove 5,027 net test lines and 15 redundant registered executions. Remove 11 lines of unused UI inspection exports. Keep every distinct company, actor, approval, credential, persistence, stale-write and concurrency contract. No wall-time speedup is claimed for setup compaction. Breaking changes None for application behavior, APIs or stored data. The removed exports expose private UI test state and have no remaining callers. ## What Changed - Consolidate repeated runtime, workspace, portability, route, tool-access, policy, gateway and state-service fixtures. - Share repeated UI render frames and CLI/wiki-plugin setup while retaining the original actions and expected results. - Remove 15 repeated case executions only where the audit records an exact stronger keeper. - Repair weak HTTP/UI negative assertions and two AWS account-home lock observations. Keep the existing durable results and cleanup behavior. - Remove the unused `__inflightGetCount` and `__secondTickInternals` inspection exports. Keep observable request coalescing and ticker lifecycle checks. - Canonicalize temporary worktree fixture paths without weakening production path checks. - Record the owner decisions, savings, independent preservation reviews and validation in [the dated audit report](https://github.com/paperclipai/paperclip/blob/58c5c4aa241c85f4665a3159c192888bbb2a5cfa/doc/plans/2026-10-12-test-audit-rounds.md) and [the compact control evidence](https://github.com/paperclipai/paperclip/blob/58c5c4aa241c85f4665a3159c192888bbb2a5cfa/doc/plans/2026-10-12-test-audit-controls.json). ## Verification - Complete owner and retained-sibling baselines passed before each audit cutover. Independent reviewers checked the full original tests, actual helper bodies, exact inputs, receipt provenance, async order, fresh allocations and case registrations. - All 14 UI/route/wiki mutations failed at their intended assertions: ten UI/API/ticker controls, one route control and three wiki persistence/receipt controls. Each mutated production source was restored to its exact original bytes. - Both old AWS lock tests passed with their respective AWS-only lock bypass. Both repaired tests failed at the intended assertions with the same bypasses. These give 16 intended failures and two passing weak controls across 18 executions. Each source was restored exactly. - Final complete state-owner/sibling run: 329/329 passed at code revision `64370831ebe191e5ce092f85f79b6fb661df7e16`. - Final `pnpm -r typecheck`, `pnpm check:token-gates` and `pnpm build`: passed at the same code revision. - Delivery head `58c5c4aa241c85f4665a3159c192888bbb2a5cfa` includes the evidence commit and master `4b766b872993045a2e787404f0475503683e5ded`. All 27 audited test files and both application files match the locally verified code revision byte-for-byte. Current-head CI validates the integrated branch, including the incoming master changes. - All eight production-owner SHA guards for the 16 deliberate failing mutation controls still match the integrated delivery checkout. The mutation receipts therefore apply to those current owners. - Full default `pnpm test:run`: failed with exit 1 at frozen code revision `64370831ebe191e5ce092f85f79b6fb661df7e16`; HEAD and tracked source bytes remained unchanged. The first general-server group reported 878 passing files, four failing files and six skipped files; 17,846 passing cases, 24 failing cases and 105 skipped cases. Later groups did not run. Failures were `http-log-redaction` (socket hangup), `public-mcp` (initial 15-second timeout followed by hosted token “Refresh before subscribing” failures), `remove-test-home` (macOS `ENOTEMPTY` where permission errors were expected, plus cleanup failure), and `workspace-git-snapshot-streaming` (300-second timeout). All four failing test files and the relevant cleanup/MCP production owners are unchanged by this audit. Full local success is not claimed. - Focused complete-owner diagnosis at the same frozen revision: 149/152 passed, three failed, zero skipped. `public-mcp` passed 82/82 and `http-log-redaction` passed 66/66 unchanged. The two `remove-test-home` cases reproduced macOS `ENOTEMPTY` against an expected `EACCES`/`EPERM` contract; the 32 MiB workspace snapshot case reproduced its existing 300,000 ms timeout. HEAD and tracked files stayed clean. Exact blob comparison confirmed all four test files and the relevant cleanup/MCP owners match both inspected master `4b766b872993045a2e787404f0475503683e5ded` and the frozen local revision. The three remaining local failures are unresolved; no timeout, retry or assertion was weakened. - Current-head CI: [workflow 38299262149](https://github.com/paperclipai/paperclip/actions/runs/38299262149) has passed runner selection and created all downstream test/typecheck/build gates; those gates are queued, with no failures reported. Greptile completed successfully at `58c5c4aa241c85f4665a3159c192888bbb2a5cfa` with [5/5 and zero comments](https://github.com/paperclipai/paperclip/pull/16022#issuecomment-6131457592). All review threads are resolved or absent. This pull request will remain open after verification. Reviewers can reproduce the required workspace checks: ```sh pnpm -r typecheck pnpm check:token-gates pnpm test:run pnpm build ``` Browser suites were not run. Their sources and user journeys are unchanged by this audit. ## Risks - A fixture helper can hide a specialized input or change object freshness. Full-program expansion checks and independent reviews preserve the exact old inputs and execution order. - Removing duplicate proof can remove a distinct contract by mistake. The audit records the keeper for each deletion; all distinct security and lifecycle cases remain. - Concurrency checks can miss progress during a fixed delay. The AWS repairs record actual provider entry, confirm holder entry, use measured entry windows, require exact final event order and drain both operations in `finally`. - The full local default run failed. Focused diagnosis reproduced two macOS cleanup failures and one existing snapshot timeout in unchanged owners; these remain unresolved. Current-head CI is pending. No test deadline, retry policy, skip or CI shard was weakened to obtain a pass. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used OpenAI Codex with GPT-6, reasoning, repository inspection, code execution and delegated audit/review. The runtime does not expose the exact served model identifier or context-window size; those values are unavailable and are not inferred from the model catalog. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [ ] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing>	2026-10-12 22:23:11 +00:00
DottaandClaude Opus 4.8	3e348b96b9	perf(ui): cut live-updates churn that inflates tab memory (#9624 ) ## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Its web UI keeps live views (issue threads, run transcripts, dashboard) fresh via React Query polling plus a live-events websocket, coordinated across tabs with a `BroadcastChannel` layer > - Long-lived tabs viewing live agent runs grew to multi-GB memory footprints while their JS heap stayed ~60–200 MB — so the memory is off-heap (Blink/native + committed allocator arenas), not a classic JS leak > - Live profiling of a reproduced 2.7 GB / 66 MB tab showed 15–30% idle CPU, ~3 fetches/sec across overlapping poll loops, and ~8 `setInterval` create/clear cycles per second whose rate grew ~7× as the tab aged — relentless allocation churn that inflates committed memory the OS never reclaims, amplified across tabs by the cross-tab fan-out > - This pull request cuts that churn at its four largest sources (invalidation storm, per-instance 1 s timers, redundant polling, unbounded streamed-run set) > - The benefit is that idle tabs do far less periodic work, so their off-heap footprint stops ballooning over a long session ## Linked Issues or Issue Description No public GitHub issue exists; describing inline per CONTRIBUTING.md → "Link Issues or Describe Them In-PR", following the bug report template. What happened? Browser tabs viewing live agent runs grew to 8–16 GB memory footprint over a long session (multiple tabs open), while each tab's "live" JS heap stayed only ~150–250 MB. Every idle tab also burned 15–30% CPU. Tabs eventually approached the ~4 GB V8 heap ceiling / OS pressure and could crash. Expected behavior Tabs viewing live runs should hold a bounded footprint and do minimal work while idle, regardless of how long they stay open or how many tabs are open. Steps to reproduce Open several issue/run tabs that have agents actively streaming and leave them open for a while. Watch Chrome's Task Manager: Memory Footprint climbs into the GBs while "JavaScript Memory" stays small, and CPU stays high on idle tabs. Reproduced in ~90 minutes: a tab reached 2.7 GB footprint on a 66 MB JS heap, and the per-second timer-churn rate was ~7× higher on a 90-minute-old tab than a fresh one. Paperclip version or commit Branch `fix/live-updates-churn`, off `master`. Deployment mode Local dev (`pnpm dev`), web UI. Not adapter-specific — core UI live-updates plumbing (observed with `claude_local` / `codex_local` runs). ## What Changed - `ui/src/lib/query-invalidation-batcher.ts` (new) — `createInvalidationBatcher` throttles + de-dupes React Query invalidations into one flush per ~300 ms, and `createCoalescingQueryClient` wraps the client via a `Proxy` so only `invalidateQueries` is batched (optimistic `setQueryData` writes stay immediate). Wired into `LiveUpdatesProvider`, which previously invalidated synchronously on every websocket event. - `ui/src/hooks/useSecondTick.ts` (new) — one shared, ref-counted, page-wide 1 s ticker. `useLiveElapsed` in `IssueChatThread` now uses it instead of a per-instance `setInterval` that forced a full-thread re-render every second per live element. - `ui/src/components/transcript/useLiveRunTranscripts.ts` — when the realtime websocket is enabled, the recurring log poll backs off to a 30 s safety-net cadence instead of polling every 2 s on top of the live stream. Added a marker for the durable poll→push rearchitecture. - `ui/src/lib/issueChatTranscriptRuns.ts` — `resolveIssueChatTranscriptRuns` now caps the streamed run set (live/active runs always kept; most-recent linked runs fill up to 20) so a large run history can't open a live-transcript poll per historical run. - `ui/src/main.tsx` — explicit `gcTime` so cross-tab-published cache entries for unobserved resources are collected promptly. - Tests for the batcher, shared ticker, and run cap. ## Verification - `vitest`: new suites `query-invalidation-batcher.test.ts` (batcher collapses 20 invalidations → 1 flush; keeps distinct keys/variants; dispose cancels; proxy passes non-invalidate methods through), `useSecondTick.test.tsx` (single ref-counted timer, stops when idle), `issueChatTranscriptRuns.test.ts` (cap keeps newest + live). All pass. - Existing affected suites pass: `LiveUpdatesProvider` (23), `IssueChatThread` (), `useLiveRunTranscripts`, `AgentDetail.instructions` — 109 tests across affected files. - `tsc -b` clean. - Behavior confirmed by live profiling before the change (2.7 GB / 66 MB tab, ~8 interval churns/sec growing 7× with age). Runtime churn reduction should be re-measured against a rebuilt bundle with the same instrumentation. ## Risks Low-to-moderate; all changes reduce work rather than add features. - Invalidation batching delays live-driven refetches by up to ~300 ms. Optimistic `setQueryData` writes (e.g. the visible issue's new comment) remain immediate, so foreground updates still feel instant; only the safety-net refetch is throttled. Non-live invalidations (user actions, mutations) are unaffected — they use the real client. - Poll back-off relies on the websocket as the live source when realtime is enabled; a 30 s fallback poll still covers gaps/reconnects (both the transcript hook and `LiveUpdatesProvider` also auto-reconnect). - Run cap (20) means an issue with a very large run history streams live transcripts only for its live/active + 20 most-recent runs; older runs still open normally via their run pages. - Downstream test fallout (timing-sensitive tests around invalidation/polling) may need adjustment — flagged intentionally for follow-up. Durable follow-up (out of scope, marked in code): replace transcript/run polling with server push (SSE/websocket deltas) so idle tabs do no periodic work at all. ## Model Used - Provider: Anthropic, via the Claude Code CLI. - Model: Claude Opus 4.8 (`claude-opus-4-8`). - Reasoning mode: Extended thinking enabled. - Capabilities used: tool use (shell, file editing), sub-agent fan-out for codebase analysis, and the Chrome DevTools MCP to reproduce and profile the memory/CPU churn on a live instance. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work (a bug/perf fix, not planned core feature work) - [x] I have searched GitHub for duplicate or related PRs and linked them above (none found) - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have considered and documented any risks above - [ ] I have updated relevant documentation to reflect my changes (N/A — no user-facing docs; rationale documented inline) - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>	2026-07-15 14:17:37 -05:00

Author

SHA1

Message

Date

DottaandPaperclip

dabb8c5daf

test: remove repeated setup and strengthen regression proof (#16022 )

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - Its tests protect company boundaries, approvals, durable tasks,
credentials and the board UI.
> - Several owner suites repeat large database fixtures and render
frames.
> - Some tests repeat stronger proof or pass when the behavior they
describe is broken.
> - This pull request shares exact setup, removes proven duplicate cases
and repairs weak observations.
> - It preserves distinct security, lifecycle, concurrency and
user-facing contracts.
> - The benefit is 5,027 fewer test lines and stronger regression
checks.

## Linked Issues or Issue Description

Refs #15978. This continues the merged test-audit cleanup in [PR
#15978](https://github.com/paperclipai/paperclip/pull/15978), which
removed 11,976 test/support lines. The savings in this pull request are
additional.

Related open work: [PR
#13460](https://github.com/paperclipai/paperclip/pull/13460) changes
legacy OAuth parent/version lock ordering. This pull request retains
those production paths and the existing lock-order test. Its secrets
changes compact fixtures and repair the two AWS wrapper exclusion
observations.

Scoped title searches found no open duplicate audit or
fixture-consolidation PR and no matching audit issue. The problem is
described here using the enhancement template.

**What existing behavior does this improve?**

The maintenance and regression value of the existing server, UI, CLI,
shared and wiki-plugin test suites.

**Subsystem affected**

Cross-cutting test code. Two UI modules also lose unused test inspection
exports.

**Current behavior**

Owner tests repeat database rows, service inputs, HTTP setup and render
frames. Some cases repeat stronger assertions. Two AWS lock tests record
only cleanup events, so an AWS provider write can happen inside the held
lock without failing those observations. Several other negative tests
accept or do not observe the wrong result. Worktree fixtures also use a
temporary path alias that conflicts with the existing canonical-path
checks on macOS.

**Proposed behavior**

Use small local helpers that preserve exact inputs, fresh allocations,
callbacks, query order and await positions. Remove only cases with named
stronger keepers. Tighten negative assertions and observe actual AWS
provider entry while a confirmed lock holder remains parked. Resolve
temporary fixture roots to their canonical paths.

**Reason and benefit**

Remove 5,027 net test lines and 15 redundant registered executions.
Remove 11 lines of unused UI inspection exports. Keep every distinct
company, actor, approval, credential, persistence, stale-write and
concurrency contract. No wall-time speedup is claimed for setup
compaction.

**Breaking changes**

None for application behavior, APIs or stored data. The removed exports
expose private UI test state and have no remaining callers.

## What Changed

- Consolidate repeated runtime, workspace, portability, route,
tool-access, policy, gateway and state-service fixtures.
- Share repeated UI render frames and CLI/wiki-plugin setup while
retaining the original actions and expected results.
- Remove 15 repeated case executions only where the audit records an
exact stronger keeper.
- Repair weak HTTP/UI negative assertions and two AWS account-home lock
observations. Keep the existing durable results and cleanup behavior.
- Remove the unused `__inflightGetCount` and `__secondTickInternals`
inspection exports. Keep observable request coalescing and ticker
lifecycle checks.
- Canonicalize temporary worktree fixture paths without weakening
production path checks.
- Record the owner decisions, savings, independent preservation reviews
and validation in [the dated audit
report](https://github.com/paperclipai/paperclip/blob/58c5c4aa241c85f4665a3159c192888bbb2a5cfa/doc/plans/2026-10-12-test-audit-rounds.md)
and [the compact control
evidence](https://github.com/paperclipai/paperclip/blob/58c5c4aa241c85f4665a3159c192888bbb2a5cfa/doc/plans/2026-10-12-test-audit-controls.json).

## Verification

- Complete owner and retained-sibling baselines passed before each audit
cutover. Independent reviewers checked the full original tests, actual
helper bodies, exact inputs, receipt provenance, async order, fresh
allocations and case registrations.
- All 14 UI/route/wiki mutations failed at their intended assertions:
ten UI/API/ticker controls, one route control and three wiki
persistence/receipt controls. Each mutated production source was
restored to its exact original bytes.
- Both old AWS lock tests passed with their respective AWS-only lock
bypass. Both repaired tests failed at the intended assertions with the
same bypasses. These give 16 intended failures and two passing weak
controls across 18 executions. Each source was restored exactly.
- Final complete state-owner/sibling run: 329/329 passed at code
revision `64370831ebe191e5ce092f85f79b6fb661df7e16`.
- Final `pnpm -r typecheck`, `pnpm check:token-gates` and `pnpm build`:
passed at the same code revision.
- Delivery head `58c5c4aa241c85f4665a3159c192888bbb2a5cfa` includes the
evidence commit and master `4b766b872993045a2e787404f0475503683e5ded`.
All 27 audited test files and both application files match the locally
verified code revision byte-for-byte. Current-head CI validates the
integrated branch, including the incoming master changes.
- All eight production-owner SHA guards for the 16 deliberate failing
mutation controls still match the integrated delivery checkout. The
mutation receipts therefore apply to those current owners.
- Full default `pnpm test:run`: **failed** with exit 1 at frozen code
revision `64370831ebe191e5ce092f85f79b6fb661df7e16`; HEAD and tracked
source bytes remained unchanged. The first general-server group reported
878 passing files, four failing files and six skipped files; 17,846
passing cases, 24 failing cases and 105 skipped cases. Later groups did
not run. Failures were `http-log-redaction` (socket hangup),
`public-mcp` (initial 15-second timeout followed by hosted token
“Refresh before subscribing” failures), `remove-test-home` (macOS
`ENOTEMPTY` where permission errors were expected, plus cleanup
failure), and `workspace-git-snapshot-streaming` (300-second timeout).
All four failing test files and the relevant cleanup/MCP production
owners are unchanged by this audit. Full local success is not claimed.
- Focused complete-owner diagnosis at the same frozen revision: 149/152
passed, three failed, zero skipped. `public-mcp` passed 82/82 and
`http-log-redaction` passed 66/66 unchanged. The two `remove-test-home`
cases reproduced macOS `ENOTEMPTY` against an expected `EACCES`/`EPERM`
contract; the 32 MiB workspace snapshot case reproduced its existing
300,000 ms timeout. HEAD and tracked files stayed clean. Exact blob
comparison confirmed all four test files and the relevant cleanup/MCP
owners match both inspected master
`4b766b872993045a2e787404f0475503683e5ded` and the frozen local
revision. The three remaining local failures are unresolved; no timeout,
retry or assertion was weakened.
- Current-head CI: [workflow
38299262149](https://github.com/paperclipai/paperclip/actions/runs/38299262149)
has passed runner selection and created all downstream
test/typecheck/build gates; those gates are queued, with no failures
reported. Greptile completed successfully at
`58c5c4aa241c85f4665a3159c192888bbb2a5cfa` with [5/5 and zero
comments](https://github.com/paperclipai/paperclip/pull/16022#issuecomment-6131457592).
All review threads are resolved or absent. This pull request will remain
open after verification.

Reviewers can reproduce the required workspace checks:

```sh
pnpm -r typecheck
pnpm check:token-gates
pnpm test:run
pnpm build
```

Browser suites were not run. Their sources and user journeys are
unchanged by this audit.

## Risks

- A fixture helper can hide a specialized input or change object
freshness. Full-program expansion checks and independent reviews
preserve the exact old inputs and execution order.
- Removing duplicate proof can remove a distinct contract by mistake.
The audit records the keeper for each deletion; all distinct security
and lifecycle cases remain.
- Concurrency checks can miss progress during a fixed delay. The AWS
repairs record actual provider entry, confirm holder entry, use measured
entry windows, require exact final event order and drain both operations
in `finally`.
- The full local default run failed. Focused diagnosis reproduced two
macOS cleanup failures and one existing snapshot timeout in unchanged
owners; these remain unresolved. Current-head CI is pending. No test
deadline, retry policy, skip or CI shard was weakened to obtain a pass.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

OpenAI Codex with GPT-6, reasoning, repository inspection, code
execution and delegated audit/review. The runtime does not expose the
exact served model identifier or context-window size; those values are
unavailable and are not inferred from the model catalog.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [ ] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>

2026-10-12 22:23:11 +00:00

DottaandClaude Opus 4.8

3e348b96b9

perf(ui): cut live-updates churn that inflates tab memory (#9624 )

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - Its web UI keeps live views (issue threads, run transcripts,
dashboard) fresh via React Query polling plus a live-events websocket,
coordinated across tabs with a `BroadcastChannel` layer
> - Long-lived tabs viewing live agent runs grew to multi-GB memory
footprints while their JS heap stayed ~60–200 MB — so the memory is
off-heap (Blink/native + committed allocator arenas), not a classic JS
leak
> - Live profiling of a reproduced 2.7 GB / 66 MB tab showed 15–30% idle
CPU, ~3 fetches/sec across overlapping poll loops, and ~8 `setInterval`
create/clear cycles per second whose rate grew ~7× as the tab aged —
relentless allocation churn that inflates committed memory the OS never
reclaims, amplified across tabs by the cross-tab fan-out
> - This pull request cuts that churn at its four largest sources
(invalidation storm, per-instance 1 s timers, redundant polling,
unbounded streamed-run set)
> - The benefit is that idle tabs do far less periodic work, so their
off-heap footprint stops ballooning over a long session

## Linked Issues or Issue Description

No public GitHub issue exists; describing inline per CONTRIBUTING.md →
"Link Issues or Describe Them In-PR", following the bug report template.

**What happened?**

Browser tabs viewing live agent runs grew to 8–16 GB memory footprint
over a long session (multiple tabs open), while each tab's "live" JS
heap stayed only ~150–250 MB. Every idle tab also burned 15–30% CPU.
Tabs eventually approached the ~4 GB V8 heap ceiling / OS pressure and
could crash.

**Expected behavior**

Tabs viewing live runs should hold a bounded footprint and do minimal
work while idle, regardless of how long they stay open or how many tabs
are open.

**Steps to reproduce**

Open several issue/run tabs that have agents actively streaming and
leave them open for a while. Watch Chrome's Task Manager: Memory
Footprint climbs into the GBs while "JavaScript Memory" stays small, and
CPU stays high on idle tabs. Reproduced in ~90 minutes: a tab reached
2.7 GB footprint on a 66 MB JS heap, and the per-second timer-churn rate
was ~7× higher on a 90-minute-old tab than a fresh one.

**Paperclip version or commit**

Branch `fix/live-updates-churn`, off `master`.

**Deployment mode**

Local dev (`pnpm dev`), web UI. Not adapter-specific — core UI
live-updates plumbing (observed with `claude_local` / `codex_local`
runs).

## What Changed

- **`ui/src/lib/query-invalidation-batcher.ts` (new)** —
`createInvalidationBatcher` throttles + de-dupes React Query
invalidations into one flush per ~300 ms, and
`createCoalescingQueryClient` wraps the client via a `Proxy` so only
`invalidateQueries` is batched (optimistic `setQueryData` writes stay
immediate). Wired into `LiveUpdatesProvider`, which previously
invalidated synchronously on every websocket event.
- **`ui/src/hooks/useSecondTick.ts` (new)** — one shared, ref-counted,
page-wide 1 s ticker. `useLiveElapsed` in `IssueChatThread` now uses it
instead of a per-instance `setInterval` that forced a full-thread
re-render every second per live element.
- **`ui/src/components/transcript/useLiveRunTranscripts.ts`** — when the
realtime websocket is enabled, the recurring log poll backs off to a 30
s safety-net cadence instead of polling every 2 s on top of the live
stream. Added a marker for the durable poll→push rearchitecture.
- **`ui/src/lib/issueChatTranscriptRuns.ts`** —
`resolveIssueChatTranscriptRuns` now caps the streamed run set
(live/active runs always kept; most-recent linked runs fill up to 20) so
a large run history can't open a live-transcript poll per historical
run.
- **`ui/src/main.tsx`** — explicit `gcTime` so cross-tab-published cache
entries for unobserved resources are collected promptly.
- Tests for the batcher, shared ticker, and run cap.

## Verification

- `vitest`: new suites `query-invalidation-batcher.test.ts` (batcher
collapses 20 invalidations → 1 flush; keeps distinct keys/variants;
dispose cancels; proxy passes non-invalidate methods through),
`useSecondTick.test.tsx` (single ref-counted timer, stops when idle),
`issueChatTranscriptRuns.test.ts` (cap keeps newest + live). All pass.
- Existing affected suites pass: `LiveUpdatesProvider` (23),
`IssueChatThread` (), `useLiveRunTranscripts`,
`AgentDetail.instructions` — 109 tests across affected files.
- `tsc -b` clean.
- Behavior confirmed by live profiling before the change (2.7 GB / 66 MB
tab, ~8 interval churns/sec growing 7× with age). Runtime churn
reduction should be re-measured against a rebuilt bundle with the same
instrumentation.

## Risks

Low-to-moderate; all changes reduce work rather than add features.
- **Invalidation batching** delays live-driven refetches by up to ~300
ms. Optimistic `setQueryData` writes (e.g. the visible issue's new
comment) remain immediate, so foreground updates still feel instant;
only the safety-net refetch is throttled. Non-live invalidations (user
actions, mutations) are unaffected — they use the real client.
- **Poll back-off** relies on the websocket as the live source when
realtime is enabled; a 30 s fallback poll still covers gaps/reconnects
(both the transcript hook and `LiveUpdatesProvider` also
auto-reconnect).
- **Run cap (20)** means an issue with a very large run history streams
live transcripts only for its live/active + 20 most-recent runs; older
runs still open normally via their run pages.
- Downstream test fallout (timing-sensitive tests around
invalidation/polling) may need adjustment — flagged intentionally for
follow-up.

Durable follow-up (out of scope, marked in code): replace transcript/run
polling with server push (SSE/websocket deltas) so idle tabs do no
periodic work at all.

## Model Used

- **Provider:** Anthropic, via the Claude Code CLI.
- **Model:** Claude Opus 4.8 (`claude-opus-4-8`).
- **Reasoning mode:** Extended thinking enabled.
- **Capabilities used:** tool use (shell, file editing), sub-agent
fan-out for codebase analysis, and the Chrome DevTools MCP to reproduce
and profile the memory/CPU churn on a live instance.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work (a bug/perf fix, not planned core feature work)
- [x] I have searched GitHub for duplicate or related PRs and linked
them above (none found)
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have considered and documented any risks above
- [ ] I have updated relevant documentation to reflect my changes (N/A —
no user-facing docs; rationale documented inline)
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

2026-07-15 14:17:37 -05:00