Repository navigation
Conversation
|
|
|
@eVoloshchak Please copy/paste the Reviewer Checklist from here into a new comment on this PR and complete it. If you have the K2 extension, you can simply click: [this button] |
screen-20260819-175449-1787154841427.mp4@KJ21-ENG, this doesn't look right, the views are clipped much more than in main, could you double check it on your end please? |
|
Strange, checking it. |
|
I rechecked this against current main and reproduced the same full-list blanking there, so it is not introduced by this PR. The exact PR-head run also produced no updateState, updatePadding, missing-ViewState, or RetryableMountingLayerException event while the visual symptom occurred. Screen.Recording.2026-08-20.at.3.27.49.AM.mov |
|
It seems to be there for a long time. It's quite weird that it wasn't noticed till now. Here is the recording of one of the older versions. Record_2026-08-20-13-43-05.mp4 |
|
The root cause is the existing Android FlashList render window used by Inbox and Reports > Expenses. Both lists use drawDistance={250}; FlashList turns that into only about 350 px of directional pre-render buffer on the tested 727 px viewport, while the captured fast flings moved 558–917 px in 114–178 ms. The viewport therefore outruns the rendered cells and briefly reaches zero viewable items, producing the blank list until FlashList catches up. This is the known tradeoff from PRs #82378 and #83277, which reduced the prior 1000 px buffer for performance, and #83277 explicitly noted that fast scrolling could show blank spaces. An Android-only drawDistance={1000} candidate kept the list continuously rendered through repeated Inbox and Reports > Expenses tests. We can handle this in a separate issue, as I don't think that it should be handled in this PR. Here is the test branch:- https://github.com/KJ21-ENG/App/tree/KJ21-ENG/android-flashlist-fast-scroll-blanking. POW :- Screen.Recording.2026-08-20.at.1.10.35.PM.compressed.mp4 |
|
@eVoloshchak if its unrelated to these changes, lets continue with the review |
@mountiny meanwhile, should we create a new issue for this particular bug and start working on it? You can refer to #98869 (comment) for a detailed explanation. Thanks! |
|
I asked QA for retest of that |
|
Scrolling issue repro in Pixel 4/13 v9.4.57-3 az_recorder_20260820_201835.mp4 |
cc: @mountiny |
|
Asking about something internally |
Reviewer Checklist
Screenshots/VideosAndroid: HybridAppscreen-20260821-142315-1787314975681.mp4 |
|
@MelvinBot can you create an issue for this #98869 (comment) and assign @mountiny and @chrispader We will migrate to LegendList so we dont need to add more work for this in FlashList, but just to make sure the issue is fixed when LegendList is used |
mountiny
left a comment
There was a problem hiding this comment.
Thanks for adding these
|
Created the follow-up issue: #99202 — assigned to I framed it around the LegendList migration: it captures the FlashList fast-scroll blanking on Inbox and Reports > Expenses, but the goal is to verify it's fixed once LegendList is in place rather than add more |
|
🚧 mountiny has triggered a test Expensify/App build. You can view the workflow run here. |
|
✋ This PR was not deployed to staging yet because QA is ongoing. It will be automatically deployed to staging after the next production release. |
|
🧪🧪 Use the links below to test this adhoc build on Android, iOS, and Web. Happy testing! 🧪🧪
|
|
🚀 Deployed to staging by https://github.com/mountiny in version: 9.4.59-0 🚀
|
|
🚀 Deployed to production by https://github.com/AndrewGable in version: 9.4.59-5 🚀
Bundle Size Analysis (Sentry): |
Explanation of Change
Extends the existing React Native 0.86 patch for missing Android Fabric
ViewStatevalues. The patch already guardsupdateOverflowInset; this follow-upbackports React Native commit
0e86a043soupdatePaddingandupdateStatealsouse
getNullableViewState(), soft-log a missing state, and return instead ofthrowing
RetryableMountingLayerExceptionfrom a non-retryableIntBufferBatchMountItem.This addresses the production
9.4.46-10fatal recurrence whose stack isgetViewState -> updateState -> IntBufferBatchMountItem. Search/FlashListclipping,
sendAccessibilityEvent, andsetJSResponderare intentionallyunchanged.
Fixed Issues
$ #93833
PROPOSAL: #93833 (comment)
Tests
The production crash is nondeterministic, so these steps provide regression
smoke coverage rather than deterministic proof of the fix.
Final effectiveness will be confirmed by monitoring APP-3E3 after deployment.
inside React Native and introduces no JS/TS execution path.
Offline tests
N/A. The guard runs in React Native's Android Fabric mounting layer and does not
depend on network, Onyx, account, or permission state.
QA Steps
Same as Tests. Treat this as regression smoke coverage only; monitor APP-3E3
after deployment to confirm effectiveness.
PR Author Checklist
### Fixed Issuessection aboveTestssectionOffline stepssectionQA stepssectionViewStateis the failure path handled by the guard; there is no input or user-facing error state.ViewStateguard pattern.mainwas merged after review, I tested again. N/A: the branch starts from currentupstream/mainand no later merge has occurred.Screenshots/Videos
Android: Native
N/A: no visual change.
Android: mWeb Chrome
N/A: Android native-only change.
iOS: Native
N/A: Android native-only change.
iOS: mWeb Safari
N/A: Android native-only change.
MacOS: Chrome / Safari
N/A: Android native-only change.