Split scheduler commit and flush delegate methods #44188
Closed
+82
−8
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D56414689 |
Base commit: d59de54 |
Summary: The current approach used for `batchRenderingUpdatesInEventLoop` is not compatible with Android due to limitations in its props processing model. The raw props changeset is passed through to Android, and must be available for the Android mounting layer to correctly apply changes. We have some logic to merge these payloads when multiple ShadowNode clones take place but were previously assuming that a ShadowTree commit was a safe state to synchronize. In the current implementation this means that two commits driven from layout effects (triggering states A → B → C) may cause Android to observe only the B → C props change, and miss out on any props changed in A → B. Changelog: [Android][Fixed] Cascading renders were not mounting correctly when `batchRenderingUpdatesInEventLoop` is enabled. Differential Revision: D56414689
javache
force-pushed
the
export-D56414689
branch
from
a046bda to
101f13a
Compare
|
This pull request was exported from Phabricator. Differential Revision: D56414689 |
|
This pull request has been merged in 849da21. |
|
This pull request was successfully merged by @javache in 849da21. When will my fix make it into a release? | How to file a pick request? |
cipolleschi
pushed a commit
that referenced
this pull request
Apr 30, 2024
Summary: Pull Request resolved: #44188 The current approach used for `batchRenderingUpdatesInEventLoop` is not compatible with Android due to limitations in its props processing model. The raw props changeset is passed through to Android, and must be available for the Android mounting layer to correctly apply changes. We have some logic to merge these payloads when multiple ShadowNode clones take place but were previously assuming that a ShadowTree commit was a safe state to synchronize. In the current implementation this means that two commits driven from layout effects (triggering states A → B → C) may cause Android to observe only the B → C props change, and miss out on any props changed in A → B. Changelog: [Android][Fixed] Cascading renders were not mounting correctly when `batchRenderingUpdatesInEventLoop` is enabled. Reviewed By: rubennorte Differential Revision: D56414689 fbshipit-source-id: 7c74d81620db0f8b7bd67e640168afc795c7a1d7
kosmydel
pushed a commit
to kosmydel/react-native
that referenced
this pull request
May 6, 2024
Summary: Pull Request resolved: facebook#44188 The current approach used for `batchRenderingUpdatesInEventLoop` is not compatible with Android due to limitations in its props processing model. The raw props changeset is passed through to Android, and must be available for the Android mounting layer to correctly apply changes. We have some logic to merge these payloads when multiple ShadowNode clones take place but were previously assuming that a ShadowTree commit was a safe state to synchronize. In the current implementation this means that two commits driven from layout effects (triggering states A → B → C) may cause Android to observe only the B → C props change, and miss out on any props changed in A → B. Changelog: [Android][Fixed] Cascading renders were not mounting correctly when `batchRenderingUpdatesInEventLoop` is enabled. Reviewed By: rubennorte Differential Revision: D56414689 fbshipit-source-id: 7c74d81620db0f8b7bd67e640168afc795c7a1d7
kosmydel
pushed a commit
to kosmydel/react-native
that referenced
this pull request
Jun 11, 2024
Summary: Pull Request resolved: facebook#44188 The current approach used for `batchRenderingUpdatesInEventLoop` is not compatible with Android due to limitations in its props processing model. The raw props changeset is passed through to Android, and must be available for the Android mounting layer to correctly apply changes. We have some logic to merge these payloads when multiple ShadowNode clones take place but were previously assuming that a ShadowTree commit was a safe state to synchronize. In the current implementation this means that two commits driven from layout effects (triggering states A → B → C) may cause Android to observe only the B → C props change, and miss out on any props changed in A → B. Changelog: [Android][Fixed] Cascading renders were not mounting correctly when `batchRenderingUpdatesInEventLoop` is enabled. Reviewed By: rubennorte Differential Revision: D56414689 fbshipit-source-id: 7c74d81620db0f8b7bd67e640168afc795c7a1d7
This was referenced Jun 28, 2024
Merged
javache
added a commit
to javache/react-native
that referenced
this pull request
Sep 27, 2024
Summary: In facebook#44188, we've started combining multiple transactions in a single transaction, to meet React's atomicity requirements, while also dealing with the constraints of Android's Fabric implementation. This revealed a bug where in some scenarios (especially when using transitions), a node may be deleted and created during the same transaction. The current implementation of FabricMountingManager assumes it can safely reorder some operations, which it does to optimize the size of IntBufferBatch mount items. This is however incorrect and unsafe when multiple transactions are merged. **Example:** Differentiator output: ``` # Transaction 1 Remove facebook#100 from facebook#11 Delete facebook#100 # Transaction 2 Create facebook#100 Insert facebook#100 into facebook#11 ``` FabricMountingManager output ``` Remove facebook#100 from facebook#11 Insert facebook#100 into facebook#11 Delete facebook#100 ``` Note that the create action is also skipped, because we only update `allocatedViewTags` after processing all mutations, leading FabricMountingManager to assume creation is not required. This leads to an invalid state in SurfaceMountingManager, which will be surfaced as a crash in `getViewState` on the next mutation that interacts with these views. Changelog: [Android][Fixed] Fix crash in getViewState when using suspense fallbacks. Differential Revision: D63148523
javache
added a commit
to javache/react-native
that referenced
this pull request
Oct 26, 2024
…acebook#46702) Summary: In facebook#44188, we've started combining multiple transactions in a single transaction, to meet React's atomicity requirements, while also dealing with the constraints of Android's Fabric implementation. This revealed a bug where in some scenarios (especially when using transitions), a node may be deleted and created during the same transaction. The current implementation of FabricMountingManager assumes it can safely reorder some operations, which it does to optimize the size of IntBufferBatch mount items. This is however incorrect and unsafe when multiple transactions are merged. **Example:** Differentiator output: ``` # Transaction 1 Remove facebook#100 from facebook#11 Delete facebook#100 # Transaction 2 Create facebook#100 Insert facebook#100 into facebook#11 ``` FabricMountingManager output ``` Remove facebook#100 from facebook#11 Insert facebook#100 into facebook#11 Delete facebook#100 ``` Note that the create action is also skipped, because we only update `allocatedViewTags` after processing all mutations, leading FabricMountingManager to assume creation is not required. This leads to an invalid state in SurfaceMountingManager, which will be surfaced as a crash in `getViewState` on the next mutation that interacts with these views. Changelog: [Android][Fixed] Fix crash in getViewState when using suspense fallbacks. Differential Revision: D63148523
javache
added a commit
to javache/react-native
that referenced
this pull request
Oct 28, 2024
…acebook#46702) Summary: Pull Request resolved: facebook#46702 In facebook#44188, we've started combining multiple transactions in a single transaction, to meet React's atomicity requirements, while also dealing with the constraints of Android's Fabric implementation. This revealed a bug where in some scenarios (especially when using transitions), a node may be deleted and created during the same transaction. The current implementation of FabricMountingManager assumes it can safely reorder some operations, which it does to optimize the size of IntBufferBatch mount items. This is however incorrect and unsafe when multiple transactions are merged. **Example:** Differentiator output: ``` # Transaction 1 Remove facebook#100 from facebook#11 Delete facebook#100 # Transaction 2 Create facebook#100 Insert facebook#100 into facebook#11 ``` FabricMountingManager output ``` Remove facebook#100 from facebook#11 Insert facebook#100 into facebook#11 Delete facebook#100 ``` Note that the create action is also skipped, because we only update `allocatedViewTags` after processing all mutations, leading FabricMountingManager to assume creation is not required. This leads to an invalid state in SurfaceMountingManager, which will be surfaced as a crash in `getViewState` on the next mutation that interacts with these views. Changelog: [Android][Fixed] Fix crash in getViewState when using suspense fallbacks. Reviewed By: sammy-SC Differential Revision: D63148523
facebook-github-bot
pushed a commit
that referenced
this pull request
Oct 28, 2024
…46702) Summary: Pull Request resolved: #46702 In #44188, we've started combining multiple transactions in a single transaction, to meet React's atomicity requirements, while also dealing with the constraints of Android's Fabric implementation. This revealed a bug where in some scenarios (especially when using transitions), a node may be deleted and created during the same transaction. The current implementation of FabricMountingManager assumes it can safely reorder some operations, which it does to optimize the size of IntBufferBatch mount items. This is however incorrect and unsafe when multiple transactions are merged. **Example:** Differentiator output: ``` # Transaction 1 Remove #100 from #11 Delete #100 # Transaction 2 Create #100 Insert #100 into #11 ``` FabricMountingManager output ``` Remove #100 from #11 Insert #100 into #11 Delete #100 ``` Note that the create action is also skipped, because we only update `allocatedViewTags` after processing all mutations, leading FabricMountingManager to assume creation is not required. This leads to an invalid state in SurfaceMountingManager, which will be surfaced as a crash in `getViewState` on the next mutation that interacts with these views. Changelog: [Android][Fixed] Fix crash in getViewState when using suspense fallbacks. Reviewed By: sammy-SC Differential Revision: D63148523 fbshipit-source-id: 07ae26b2f7b7eba1b9784041dd3059b0956c035e
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
The current approach used for
batchRenderingUpdatesInEventLoopis not compatible with Android due to limitations in its props processing model. The raw props changeset is passed through to Android, and must be available for the Android mounting layer to correctly apply changes.We have some logic to merge these payloads when multiple ShadowNode clones take place but were previously assuming that a ShadowTree commit was a safe state to synchronize.
In the current implementation this means that two commits driven from layout effects (triggering states A → B → C) may cause Android to observe only the B → C props change, and miss out on any props changed in A → B.
Changelog: [Android] Fixed cascading renders not mounting correctly when
batchRenderingUpdatesInEventLoopis enabled.Differential Revision: D56414689