Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42265
Adds a snapshot test against the `react-native` package which emits the shape of all Flow-typed modules under `Libraries/`, as an approximation of the public JS API.
This provides:
- Visibility for maintainers on any PR which changes the shape of the public API.
- An at-a-glance diff of changed APIs between React Native versions (useful for library integrators and the Release Crew).
Note — **workflow change**: Maintainers modifying public files/function signatures under Libraries/ will need to run `yarn jest -u` and commit the updated snapshot changes.
Changelog: [Internal]
Reviewed By: TheSavior, philIip, mdvacca
Differential Revision: D52729777
fbshipit-source-id: 90ca2924b50205485b6d49e52a2889d8e00a43b9
Summary:
X-link: https://github.com/facebook/yoga/pull/1547
Pull Request resolved: https://github.com/facebook/react-native/pull/42251
Yoga has an odd behavior, where `start`/`end` edges under row-reverse are relative to flex-direction, instead of writing direction.
While Yoga doesn't actually document what this behavior is supposed to be, it goes against CK documentation, historic RN documentation, and the behavior valid on the web. It is also applied inconsistently (e.g. sometimes only on container, sometimes on child). It really is a bug, instead of an intended behavior.
We changed the default behavior for Yoga, but left the existing one behind an errata (so existing fbsource users got old behavior). We have previously seen this behavior show up in product code, including CK when running on FlexLayout.
`row-reverse` is surprisingly uncommon though:
1. Litho has <40 usages
2. RN has ~40 usages in `RKJSModules`,~30 in `arvr/js`, ~6 in `xplat/archon`
3. CK has ~80 usages
4. NT has ~40 usages
There are few enough, mostly simple components, that we can inspect through each of them, looking for signs they will hit the issue (at the potential chance of missing some).
CK accounts for 10/14 usages that I could tell would trigger the issue, since it only exposes start/end edge, and not left/right. It might make sense to make it preserve behavior instead, to reduce risk a bit.
FlexLayout is now separately powering Bloks, which wasn't surveyed, so I didn't touch CK behavior under Bloks.
There could also be other usages in other frameworks/bespoke usages, and this has implications for OSS users. But based on our own usage, of many, many components, this seems rare.
Changelog:
[General][Breaking] - Make `start/end` in styles always refer to writing direction
Reviewed By: pentiumao, joevilches
Differential Revision: D52698130
fbshipit-source-id: 2a9ac47e177469f30dc988d916b6c0ad95d53461
Summary:
Original commit changeset: 9305bc56ba6b
Original Phabricator Diff: D52642168
bypass-github-export-checks
changelog: [Android][Fix] Backout fix that prevented scroll event in nested scroll when scrollEnabled = false, due to causing bugs when interacting with keyboard events
Reviewed By: bvanderhoof, arushikesarwani94
Differential Revision: D52736596
fbshipit-source-id: fa8c5c598e049cc58410892813825852c431eee4
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41978
I'm revering the removal of ReactModule codegen.
We are postpoinging the removal of the codegen for the future, the reasons are:
- resources: the experiment that removes the codegen shows neutral metrics, but the codegen is shared between bridge and bridgeless, so we will need to implement and test the removal for bridge and we don't have the time to do this right now.
- reduce fragmentation: we don't want to fragment NativeModules configuration between bridge and bridgeless, doing so will bring a lot of confusion to developers
- we don't want to introduce a public APIs in 0.73 that we know they are not used in production for now, we better remove these "unstable" apis before 0.74 cut
Note: I'm updating ReactAndroid.api because this is an intended change of APIs which were not part of 0.73 and we don't want them to be part of 0.74.
changelog: [internal] internal
Reviewed By: RSNara
Differential Revision: D52223650
fbshipit-source-id: 681bf5e4aab776505f64b1972a6ace6340db4587
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42259
At the end of last year, we reduce build fragmentation in iOS making sure that we were always building both architecture.
In the process, we break the semantic od RCt_NEW_ARCH_ENABLED flag, making several libs stop working in one of the two archs.
This change should restore the semantic, so libraries that were using RCT_NEW_ARCH_ENABLED to run conditional code will still work in the same way. While doing so, I also removed the new USE_NEW_ARCH as we don't want unnecessary flags
## CHANGELOG:
[iOS][Fixed] - Bring the old RCT_NEW_ARCH_ENABLED semantic back for compatibility
Reviewed By: cortinico
Differential Revision: D52727792
fbshipit-source-id: e211b10e7885eada83dd2886375575133ca76c8c
Summary:
Yesterday we landed a change that removed tests for the Old Architecture for RNTester.
That was the right call as there are no build differences in RNTester between the two architectures. But we do have runtime differences, and we had an integration test running on RNTester that we deleted with the previous PR.
This change restores that test, adding this only new job to run that test
## Changelog:
[Internal] - Add back an old arch integration test
Pull Request resolved: https://github.com/facebook/react-native/pull/42262
Test Plan: CircleCI is green
Reviewed By: cortinico
Differential Revision: D52730661
Pulled By: cipolleschi
fbshipit-source-id: 10fbc2540abeebc72f635451f6f650827cf20041
Summary:
This PR fixes issue in `RNTester` causing labels and image background to not be visible in dark mode in `SnapshotExample`
It also fixes issue with description in `Header` not being visible in other components examples when using dark mode
Before & After
<img width="505" alt="image" src="https://github.com/facebook/react-native/assets/56474758/ce87df69-4b79-48a0-b9be-4a7335329b78">
## Changelog:
[INTERNAL] [FIXED] - Fix dark mode in SnapshotExample in RNTester
Pull Request resolved: https://github.com/facebook/react-native/pull/41222
Test Plan:
1. Launch `RNTester` with dark mode enabled
2. Open `Snapshot / Screenshot` example
3. All labels should be visible, image background should have white color
Reviewed By: cortinico
Differential Revision: D52685754
Pulled By: NickGerleman
fbshipit-source-id: 72f79be45d9c65e307553832592563461a64ff1d
Summary:
As discussed with cipolleschi, RNTester shouldn't be tested for Old Arch. This PR removes those unnecessary pipeline runs
## Changelog:
[INTERNAL] [REMOVED] - remove old architecture pipeline for RNTester
Pull Request resolved: https://github.com/facebook/react-native/pull/42245
Test Plan: CI Green
Reviewed By: cortinico
Differential Revision: D52694176
Pulled By: cipolleschi
fbshipit-source-id: a607bac4659b0611d5f49b5e45134f896bb96a91
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42230
While developing Xcode 15, Apple reimplemented the linker.
In Xcode 15.0, the linker was making old iOS (< 15) crash when they were built using Xcode 15.
To fix that, we make Apple create new compiler flags (`-ld_classic`) to have a backward compatible linker.
In Xcode 15.1, Apple fixed that behavior, so the flags should not be required anymore.
But now, if we pass `-ld_classic` to the linker and we have an app that is using `use_framworks!`, that app crashes at startup.
This change remove the flags if the Xcode that is used is 15.1 or greater.
*Note:* The previous change added the flags to Hermes as well. I tested this fix in a configuration where Hermes has the flags and React Native does not, and it works. So we are removing the flags only from React Native.
This Fixes https://github.com/facebook/react-native/issues/39945
## Changelog:
[Internal] - Do not add the `-ld_classic` flag if the app is built with Xcode 15.1 or greater.
Reviewed By: cortinico
Differential Revision: D52658197
fbshipit-source-id: 37d6bc895921c0fc3661f301870477191e7e42b3
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42185
We've rolled out some changes to the behavior here, mainly that throttling no longer happens by default on iOS.
This updates the documentation, in concert with https://github.com/facebook/react-native-website/pull/3971
Changelog:
[General][Changed] - Update API docs for scrollEventThrottle
Reviewed By: javache
Differential Revision: D52516092
fbshipit-source-id: 7be1d6e1bc62f38c795b64ad4be5d5c1b23bb742
Summary:
## Stack
These can suss out some real bugs, and helps further avoid mismatch with downstream MSVC on /W4 as used by MSFT.
I enabled the families of warnings, but suppressed some major individual warnings that weren't clean. But I did clean some up, notably, missing initializer, and shortening 64 bit to 32 bit. We can do some of the rest incrementally (e.g. `-Wunused-parameter` has a fixit).
This change illuminates that MapBuffer is missing 64 bit integer support, but we often pass 64 bit counters to it, which is a bug. For now I just left TODOs around those.
`rn_xplat_cxx_library` is used for external libraries interfacing with RN, which we probably don't want to police, so I structured these stricter warnings as an opt-in flag, only enabled for our own rules.
## Diff
This fixes up source code to avoid emitting the extra warnings now enforced. Of what is enabled, this is mostly shortening 64 to 32, or missing field in initializer.
Changelog: [Internal]
Reviewed By: christophpurrer
Differential Revision: D52589303
fbshipit-source-id: 11cb778d065799fd0ead3ae706934146d13500bb
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42214
These methods are only used in the legacy React renderer. The Fabric renderer calls into different methods. So, let's just leave these unimplemented.
NOTE: I introduced the warning back into clearJSResponder, because I couldn't find the internal call-site to it.
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D52449787
fbshipit-source-id: bd001c0ca4a3e64aaaf6328b3322025b09ee6da9
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42218
## Changes
This diff makes the native view config interop layer on Android lazy.
As in: ViewManagers that were registered lazily with React Native will no longer be eagerly initialized by the nvc interop layer.
Changes to UIManager apis:
- UIManager.getConstants() now **only** contains the view configs for the eager view managers.
- UIManager.getConstantsForViewManager(name): lazily load view configs for lazy components
- UIManager.getDefaultEventTypes(): load default event types
- UIManager.getConstants().LazyViewManagersEnabled: true, if there are lazy view managers
- UIManager.getConstants().ViewManagerNames: a list of the lazy view managers
Changelog: [Internal]
Reviewed By: dmytrorykun
Differential Revision: D52399280
fbshipit-source-id: d9cd46de0507ecfe6cca5595a237e1063f60fa62
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42213
This improves resiliency of the native view config interop layer.
In open source, some packages might provide ViewManagers eagerly, while others might them lazily.
This also fixes another problem: Prior, eager view managers would be created **then destroyed *wastefully*** by the native view config interop layer.
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D52399003
fbshipit-source-id: 3345c82789f1ed8e613139a8323dac4b4a01d173
Summary:
This diff should not change any behaviour.
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D52399000
fbshipit-source-id: 976f5740c53d58ceead7d2bc4c9e0eb3f97ebb4e
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42215
This diff should not change any behaviour.
**Why:** This logic is only used once from the UIConstantsProviderManager. So, let's just inline it. Inlining this method will make ReactInstance.java have fewer private methods, which'll make ReactInstance.java easier to read.
**Concern:** Inlining this method into ReactInstance's constructor will make the constructor too hard to read.
- I think it'll be fine: we will simplify this method significantly in D52399003.
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D52399002
fbshipit-source-id: 8c0dc69af86109da8144546347eecd2e01c0e0be
Summary:
When a FlatList is in side a scroll view (think Netflix style navigation), the DPAD up/down fires on the scroll view, despite scrollEnabled={false} being set. This additiontially conflicts with any custom scroll event that has been created.
## Changelog:
[Android] [Fixed] - fix: prevent scroll event in nested scroll when scrollEnabled={false}
Pull Request resolved: https://github.com/facebook/react-native/pull/42219
Test Plan:
I tested this by making a ScrollView with FlatList of opposite scrolling direction inside with basic card layouts.
Both had scrollEnabled={false}
I scrolled the ScrollView myself as it has multiple rows using:
```
const scrollToItem = React.useCallback(
(itemIndex: number): void => {
const targetScrollY = itemIndex * height
scrollViewRef.current?.scrollTo({ y: targetScrollY, animated: true })
},
[height]
)
React.useEffect(() => {
// Row 0, is global nav, but it's also the first row of cards
// when we scroll to "1" what we mean is global nav is hidden
// we should still be showing the first row of items.
scrollToItem(rowIndex <= 1 ? 0 : rowIndex - 1)
}, [rowIndex, scrollToItem])
```
Reviewed By: NickGerleman
Differential Revision: D52642168
Pulled By: mdvacca
fbshipit-source-id: 9305bc56ba6b03b04b9f69a14d433593cab2025e
Summary:
`compose-source-maps.js` fails if `-o` is not specified when it should output the composed source map.
## Changelog:
[GENERAL] [FIXED] - Fix `compose-source-maps.js` failing if `-o` is not specified when it should output the composed source map
Pull Request resolved: https://github.com/facebook/react-native/pull/42203
Test Plan:
Tested this in an internal repo. This was the output before this fix:
```
% node node_modules/react-native/scripts/compose-source-maps.js dist/main.jsbundle.map dist/main.jsbundle.hbc.map
node:internal/streams/writable:472
throw new ERR_INVALID_ARG_TYPE(
^
TypeError [ERR_INVALID_ARG_TYPE]: The "chunk" argument must be of type string or an instance of Buffer or Uint8Array. Received undefined
at _write (node:internal/streams/writable:472:13)
at Writable.write (node:internal/streams/writable:494:10)
at Object.<anonymous> (/~/node_modules/.store/react-native-virtual-c8e66dddc1/node_modules/react-native/scripts/compose-source-maps.js:64:20)
at Module._compile (node:internal/modules/cjs/loader:1376:14)
at Module._extensions..js (node:internal/modules/cjs/loader:1435:10)
at Module.load (node:internal/modules/cjs/loader:1207:32)
at Module._load (node:internal/modules/cjs/loader:1023:12)
at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:135:12)
at node:internal/main/run_main_module:28:49 {
code: 'ERR_INVALID_ARG_TYPE'
}
Node.js v20.10.0
```
Reviewed By: christophpurrer
Differential Revision: D52650438
Pulled By: arushikesarwani94
fbshipit-source-id: b8f8f01fb6d843887d874a7283a1a6807c7762e7
Summary:
Dependency on `chalk` was introduced in https://github.com/facebook/react-native/pull/37510, but was never declared. In pnpm setups, the CLI fails to run because of this.
This needs to be picked to 0.73.
## Changelog:
[GENERAL] [FIXED] - Declare missing dependency `chalk`
Pull Request resolved: https://github.com/facebook/react-native/pull/42235
Test Plan: n/a
Reviewed By: huntie
Differential Revision: D52660337
Pulled By: cortinico
fbshipit-source-id: 1cd45fcff72045c127773566a27103f1b38262b3
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42233
This diff removes the need for providing the `ios_folder` argument to `use_react_native`. We no longer do any manual path tranformations to get the iOS project root.
Instead we use `Pod::Config.instance.installation_root` which always points to the correct directory.
Changelog: [iOS][Breaking] - CocoaPods: remove the `ios_folder` argument from the `use_react_native` function.
Reviewed By: cipolleschi
Differential Revision: D52659429
fbshipit-source-id: 67c79cd9d74a0351ad2c242b74cbd48b6bd2dc94
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42207
In open source, with the new architecture, layout animations are **always** enabled.
They cannot be disabled.
Therefore, when this UIManagerModule method is called with false, just report an error. That way, if layout animations were explicitly disabled on Android, the developer will know, when they try to enable the new architecture.
Changelog: [Internal]
Reviewed By: sammy-SC
Differential Revision: D52349297
fbshipit-source-id: 7969bd7294ce7369643004e5ff7e0c1ed4a59cd6
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41998
Now the error message propts people to turn on the interop layer.
And, it adds more details to the suggestion to use hasViewManager(viewManagerName).
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D52002909
fbshipit-source-id: 80ea60b4f6a5fe15d773bb1f3f41de5ce43d6652
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42067
These methods should not be implemented in the new architecture.
The **only** code that called these UIManagerModule methods was the paper renderer. And the New Architecture should instead use the Fabric renderer.
Changelog: [Internal]
Reviewed By: sammy-SC
Differential Revision: D52345416
fbshipit-source-id: 76511aa97e5dfa938aca658af03fb43122547df1
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41997
Many methods on PaperUIMangaer are iOS only.
Many methods on PaperUIManager are Android only.
This diff makes sure that BridgelessUIManager only exports Android methods on Android, and iOS methods on iOS.
Changelog: [Internal]
Reviewed By: sammy-SC
Differential Revision: D52012876
fbshipit-source-id: 6527048083eae93577a58d4b77f0645fab84217f
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42200
Changelog: [Internal]
Quick hack to make it easy to determine whether a given build of React Native is using the C++ implementation of InspectorPackagerConnection or the legacy platform-specific implementation.
For now, we just append this information to the `title` field. Ultimately, rather than polluting the title, this should be an inert capability flag that gets reported via `inspector-proxy`. I'm not doing that yet since we have work in the pipeline to set up a proper capability flag system soon.
Reviewed By: huntie
Differential Revision: D52629415
fbshipit-source-id: a4e873f4be78ae49b35b94fd5d41d0e2efc02dbe
Summary:
PR https://github.com/facebook/react-native/pull/42159 was working but it was the wrong fix.
The right fix is to use the `"PUBLIC_HEADERS_FOLDER_PATH"` Xcode build setting instead.
bypass-github-export-checks
## Changelog:
[iOS][Changed] - Revert "Update Yoga.podspec: fixes archiving for macos catalyst on react-native 0.73.1 in xcode"
## Facebook:
Original commit changeset: 21b9b3568986
Original Phabricator Diff: D52624342
Reviewed By: arushikesarwani94
Differential Revision: D52656133
fbshipit-source-id: 84a37fe3fca57d5e34139c17c6c1957fe8d40aaf
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42165
This will help avoid a name collision when we remove the new suffix from newGetOrCreateReactInstanceTask.
Changelog: [Internal]
Reviewed By: mdvacca
Differential Revision: D52495532
fbshipit-source-id: 79a04cff51eef07b91876a1351b8444654a79274
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42187
Updates the docblocks for `Pressability` and `usePressability`, as suggested in the code review for {D52388699}.
Changelog:
[General][Changed] Updated Pressability/usePressability Docblocks
Reviewed By: sammy-SC
Differential Revision: D52604388
fbshipit-source-id: e82dd6caa46fe69281e996cbdb8b8e5105b46955
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42194
Apps that have multiple concurrently running React instances may suffer from issues where tearing down one instance affects the bindings / LongLivedObjectCollection instance of another due to the use of static getter for LongLivedObjectCollection. This should allow host platforms, e.g., react-native-windows (which still forks the TurboModuleBinding C++ files [here](https://github.com/microsoft/react-native-windows/tree/main/vnext/ReactCommon/TEMP_UntilReactCommonUpdate/react/nativemodule/core/ReactCommon) for the reasons already mentioned) to manage per instance LongLivedObjectCollections.
## Changelog
[Internal]
Reviewed By: christophpurrer
Differential Revision: D52581170
fbshipit-source-id: 791e3baeefaf23f544eeddd5a216735535523a9d
Alex Hunt
Nick Gerleman
David Vacca
Riccardo Cipolleschi
Riccardo Cipolleschi
Pieter De Baets
Alex Taylor (alta)
Kacper Rozniata
Oskar Kwaśniewski
Ramanpreet Nara
Ryan Wilson
Tommy Nguyen
Paul Schreiber
Dmitry Rykun
Moti Zilberman
zhongwuzw
Tim Yung
Eric Rozell