Repository navigation
Conversation
Codecov Report✅ Changes either increased or maintained existing code coverage, great job!
|
|
This PR adds a new |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0984cb4040
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
I’ll review this PR within the next 4 hours |
| // Full account requests capture this generation when they are created. Their success handoff only | ||
| // marks that captured generation ready, so a response that finishes after another reset stays stale. | ||
| let productMarketingWindowDataResetID: string | undefined; | ||
| Onyx.connectWithoutView({ |
There was a problem hiding this comment.
we shouldnt be using connectWithoutView
| willAlertModalBecomeVisible: false, | ||
| }, | ||
| [ONYXKEYS.RAM_ONLY_IS_PRODUCT_MARKETING_WINDOW_COVERED]: false, | ||
| // PRODUCT_MARKETING_WINDOW_DATA_STATE intentionally has no default. Onyx initial defaults |
|
I don't think that's the right solution. We can fix this by simply adding Line 162 in ac9126f add: ONYXKEYS.NVP_LAST_DISMISSED_MARKETING_WINDOW,
I think we can simply add NVP_LAST_DISMISSED_MARKETING_WINDOW to KEYS_TO_PRESERVE_DELEGATE_ACCESS. // Otherwise it's wiped and refetched by OpenApp, letting the product-marketing window flash back in
// before the dismissal value round-trips back from the server.
ONYXKEYS.NVP_LAST_DISMISSED_MARKETING_WINDOW, |
|
@huult The current implementation uses a persisted reset generation because both issues happen while account-scoped Onyx data is being cleared and rehydrated. Each full OpenApp/ReconnectApp captures the generation it starts under and only marks that same generation ready after its complete response data has been applied. Until then, the product marketing window stays hidden. This prevents an older in-flight request from unblocking a newer reset, and lets a later full reconnect complete a pending reset after a failed request or app restart. That is the reason I used the shared rehydration gate instead of relying only on |
|
https://github.com/huult/App/pull/new/97846-test-branch Screen.Recording.2026-08-05.at.22.38.13.mov@KJ21-ENG I opened a draft solution. Could you take a look? |
|
@KJ21-ENG Are you still working on the update? |
|
@huu1t Yes, I found an additional case during testing. If we preserves the existing dismissal value across both clears without checking which account it belongs to. I have doubt that this could carry an agent’s dismissal state over to an owner who never dismissed the window, So checking that |
|
@KJ21-ENG We should fix the deploy blocker first. We can investigate and fix the other issues later. |
|
@KJ21-ENG any updates here? These are the last 2 blockers |
|
@KJ21-ENG so this is ready for review? |
|
@JS00001 Yes |
Reviewer Checklist
Screenshots/VideosAndroid: HybridAppAndroid: mWeb ChromeiOS: HybridAppiOS: mWeb SafariMacOS: Chrome / Safari |
|
This is fixed, tested locally Screen.Recording.2026-08-05.at.5.33.24.PM.mov |
…-rehydration-gate [cp stag] Fix product marketing window flicker during account rehydration (cherry picked from commit f86bc34) (cherry-picked to staging by arosiclair)
|
✋ This PR was not deployed to staging yet because QA is ongoing. It will be automatically deployed to staging after the next production release. |
|
🚧 JS00001 has triggered a test Expensify/App build. You can view the workflow run here. |
|
🧪🧪 Use the links below to test this adhoc build on Android, iOS, and Web. Happy testing! 🧪🧪
|
|
🚀 Cherry-picked to staging by https://github.com/arosiclair in version: 9.4.50-3 🚀
|
|
🤖 No help site changes required. I reviewed the changes in this PR against the help site articles under There's no user-facing feature, UI, tab, setting, or button change here — it fixes a transient flicker so the existing (correct) behavior stays correct. Nothing in the public help articles documents the product marketing window's rehydration timing, so no article needs updating. No draft PR was created. @KJ21-ENG, please confirm you agree no help site update is needed here. If you believe this behavior should be documented, let me know and I'll open a draft help site PR. |
|
🚀 Deployed to production by https://github.com/arosiclair in version: 9.4.50-3 🚀
Bundle Size Analysis (Sentry): |
|
🚀 Cherry-picked to staging by https://github.com/arosiclair in version: 9.4.51-0 🚀
|
|
🤖 No help site changes required (reconfirmed). As I noted in my earlier review, this PR is a purely internal fix — it adds There's no user-facing feature, UI, tab, setting, or button change — the existing behavior is unchanged; the fix just removes a transient flicker. No public help article documents the marketing window's rehydration timing, so nothing under @KJ21-ENG, please confirm you agree no help site update is needed here. If you believe this behavior should be documented, let me know and I'll open a draft help site PR. |
|
i am back |
|
🚀 Deployed to production by https://github.com/roryabraham in version: 9.4.51-1 🚀
Bundle Size Analysis (Sentry): |
Explanation of Change
This is a follow-up to #96141 that fixes one account-rehydration race exposed through two paths: returning from copilot access and using Clear cache and restart.
Account-scoped resets temporarily remove the product-marketing dismissal NVP. During that interval, an absent dismissal could be interpreted as eligibility and briefly remount an already-dismissed window. This change introduces a persisted reset-generation handshake shared by every browser tab. The window stays hidden until a successful OpenApp or full ReconnectApp adds the current generation to a ready-generation map after applying the complete account data, including the dismissal NVP. Queued requests wait for the queue to drain, while full side-effect reconnects mark readiness through success data after their HTTPS payload is applied. Because each request can only add the generation it began under, a stale response cannot open a newer unfinished generation or close a newer finished one. An unfinished reset also remains hidden across a process restart. Imported state explicitly marks the current generation ready because it intentionally bypasses OpenApp.
Fixed Issues
$ #97834
$ #97846
PROPOSAL: #95791 (comment)
Tests
Offline tests
Automated coverage verifies that a failed OpenApp, stale pre-reset response, and queued or side-effect partial ReconnectApp do not mark the current generation ready, while current-generation queued and side-effect full ReconnectApp do.
QA Steps
Same as Tests.
PR Author Checklist
### Fixed Issuessection aboveTestssectionOffline stepssectionQA stepssectionAvatar, I verified the components usingAvatarare working as expected)StyleUtils.getBackgroundAndBorderStyle(theme.componentBG))npm run compress-svg)Avataris modified, I verified thatAvataris working as expected in all cases)Designlabel and/or tagged@Expensify/designso the design team can review the changes.mainbranch was merged into this PR after a review, I tested again and verified the outcome was still expected according to theTeststeps.Screenshots/Videos
Android: Native
Android: mWeb Chrome
iOS: Native
iOS: mWeb Safari
MacOS: Chrome / Safari
Screen.Recording.2026-08-05.at.12.48.22.PM.compressed.mp4