fix flaky cookieAccess tests in slow CI

[watson]

Motivation

The document.cookie fallback variant of cookieAccess.spec is occasionally flaky on CI. The failure mode is should notify when cookie is changed externally reporting:

Expected spy observer to have been called once. It was called 0 times.

The test is unrelated to any specific feature work — it can be reproduced from main on slow CI runs.

Changes

The root cause is a real-time / mock-time mismatch. When mockClock() is installed, new Date() returns the mocked time. setCookie() computes the cookie's expiry from new Date() + expireDelay, but since mockClock() initializes the mocked time to the current real time, the absolute expiry the browser sees is REAL_T_install + expireDelay. The browser then expires cookies against real wall-clock time — it doesn't know anything about the mock.

With the previous expireDelay = 1000 ms, any CI run slow enough for >1 s of real time to elapse between mockClock() and the polling interval reading document.cookie would see the cookie already expired. getCookies() returned [], matched previousCookieValues = [] captured at construction, and the observable never notified.

Two complementary fixes in packages/core/src/browser/cookieAccess.spec.ts:

• Bump the test cookie expireDelay from 1000 ms to ONE_MINUTE at every call site, leaving real time comfortable headroom to elapse during a slow test without the cookie disappearing.
• Register a generic deleteCookie(COOKIE_NAME) cleanup in each setup() variant. Three tests in the file (should pass empty array when cookie does not exist, should write the value returned by the callback, should notify the observable after writing) write the test cookie via getAllAndSet without registering their own cleanup. Previously the short expiry incidentally limited the blast radius of those leaks; under the longer expiry, the leaked cookies would otherwise persist into the next test and break should pass empty array when cookie does not exist under certain Jasmine random orderings. The generic cleanup is registered first so it runs last (LIFO) and is a no-op when a more specific cleanup already deleted the cookie.

No production code changes — this is test-only.

Test instructions

To stress the random ordering (including the originally failing seed 00743):

for seed in 00001 00100 00743 12345 99999 50000; do
  yarn test:unit --spec packages/core/src/browser/cookieAccess.spec.ts --seed $seed
done

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change
• Added e2e/integration tests for this change
• Updated documentation and/or relevant AGENTS.md file

[watson]

• fix flaky cookieAccess tests in slow CI #4613 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cit-pr-commenter-54b7da]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	169.51 KiB	169.51 KiB	0 B	0.00%	✅
Rum Profiler	5.97 KiB	5.97 KiB	0 B	0.00%	✅
Rum Recorder	21.23 KiB	21.23 KiB	0 B	0.00%	✅
Logs	54.70 KiB	54.70 KiB	0 B	0.00%	✅
Rum Slim	127.85 KiB	127.85 KiB	0 B	0.00%	✅
Worker	22.99 KiB	22.99 KiB	0 B	0.00%	✅

🚀 CPU Performance

Action Name	Base CPU Time (ms)	Local CPU Time (ms)	𝚫%
RUM - add global context	0.0031	0.003	-3.23%
RUM - add action	0.0164	0.0136	-17.07%
RUM - add error	0.0137	0.0116	-15.33%
RUM - add timing	0.0007	0.0005	-28.57%
RUM - start view	0.0151	0.011	-27.15%
RUM - start/stop session replay recording	0.0013	0.001	-23.08%
Logs - log message	0.0234	0.0207	-11.54%

🧠 Memory Performance

Action Name	Base Memory Consumption	Local Memory Consumption	𝚫
RUM - add global context	38.75 KiB	37.87 KiB	-904 B
RUM - add action	64.74 KiB	63.06 KiB	-1.68 KiB
RUM - add timing	36.82 KiB	41.44 KiB	+4.62 KiB
RUM - add error	70.81 KiB	73.59 KiB	+2.78 KiB
RUM - start/stop session replay recording	41.37 KiB	41.81 KiB	+459 B
RUM - start view	486.34 KiB	482.38 KiB	-3.96 KiB
Logs - log message	55.35 KiB	53.74 KiB	-1.61 KiB

🔗 RealWorld

[datadog-datadog-prod-us1-2]

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 76.96% (+0.00%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 203f5c1 | Docs | Datadog PR Page | Give us feedback!}

[thomas-lebeau]

Supposedly we already delete all cookies in the global forEach

If this is not working properly we should probably fix it there, no?