Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/36547
This is part of a series of changes to minimize the number of spans committed to EditText, as a mitigation for platform issues on Samsung devices. See this [GitHub thread]( https://github.com/facebook/react-native/issues/35936#issuecomment-1411437789) for greater context on the platform behavior.
This adds `ReactBackgroundColorSpan` to the list of spans eligible to be stripped.
Changelog:
[Android][Fixed] - Minimize Spans 3/N: ReactBackgroundColorSpan
Reviewed By: javache
Differential Revision: D44240782
fbshipit-source-id: 2ded1a1687a41cf6d5f83e89ffadd2d932089969
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/36546
This is part of a series of changes to minimize the number of spans committed to EditText, as a mitigation for platform issues on Samsung devices. See this [GitHub thread]( https://github.com/facebook/react-native/issues/35936#issuecomment-1411437789) for greater context on the platform behavior.
This change generalizes `stripAttributeEquivalentSpans()` to allow plugging in different spans.
Changelog:
[Internal]
Reviewed By: rshest
Differential Revision: D44240781
fbshipit-source-id: 89005266020f216368e9ad9ce382699bd8db85a8
# Conflicts:
# ReactAndroid/src/main/java/com/facebook/react/views/textinput/ReactEditText.java
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/36543
This is part of a series of changes to minimize the number of spans committed to EditText, as a mitigation for platform issues on Samsung devices. See this [GitHub thread]( https://github.com/facebook/react-native/issues/35936#issuecomment-1411437789) for greater context on the platform behavior.
We cache the backing EditText span on text change to later measure. To measure outside of a TextInput we need to restore any spans we removed. Spans may overlap, so base attributes should be behind everything else.
The logic here for dealing with precedence is incorrect, and we should instead accomplish this by twiddling with the `SPAN_PRIORITY` bits.
Changelog:
[Android][Fixed] - Minimize Spans 1/N: Fix precedence
Reviewed By: javache
Differential Revision: D44240779
fbshipit-source-id: f731b353587888faad946b8cf1e868095cdeced3
Summary:
D42721684 (https://github.com/facebook/react-native/commit/be69c8b5a77ae60cced1b2af64e48b90d9955be5) left a pretty bad bug when using Fabric for Android. I missed that in Fabric specifically, on edit we will cache the Spannable backing the EditText for use in future measurement.
Because we've stripped the sizing spans, Spannable measurement has incorrect font size, and the TextInput size will change (collapsing) after the first edit. This effectively breaks any uncontrolled TextInput which does not have explicit dimensions set.
Changelog:
[Android][Fixed] - Fix measurement of uncontrolled TextInput after edit
Reviewed By: sammy-SC
Differential Revision: D43158407
fbshipit-source-id: 51602eab06c9a50e2b60ef0ed87bdb4df025e51e
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/35967
In https://github.com/facebook/react-native/issues/35936 we observed that the presence of AbsoluteSizeSpan may lead to hangs when using the Grammarly keyboard on Samsung.
This mitigation makes it so that we do not emit this span in any case where it is sufficient to rely on already set EditText textSize. In simple cases, tested on two devices, it causes typing into the TextInput to no longer hang.
This does not fully resolve the issue for TextInputs which meaningfully use layout-effecting spans (or at least font size), such as non-uniform text size within the input. We instead just try to reduce to minimum AbsoluteSizeSpan possible.
Testing the first commit was able to resolve hangs in some simpler inputs tested, by me and cortinico.
Changelog:
[Android][Fixed] - Mitigation for Samsung TextInput Hangs
Reviewed By: cortinico
Differential Revision: D42721684
fbshipit-source-id: e0388dfb4617f0217bc1d0b71752c733e10261dd
Summary:
`JNIEnv`'s `FindClass(..)` function takes the classes in the standard
`foo/bar/Baz` class specification (unless they're special, like arrays).
Specifying them with `Lfoo/bar/Baz;` results in a
`ClassNotFoundException` being raised -- which is especially unhelpful
when intending to re-throw an exception.
The docs for `JNIEnv#FindClass(..)` can be found [here][jnienv].
[jnienv]:
https://docs.oracle.com/javase/7/docs/technotes/guides/jni/spec/functions.html#:~:text=The%20name%20argument,java/lang/String%22
## Changelog
[Android] [Fixed] - Correctly resolve classes with FindClass(..)
Pull Request resolved: https://github.com/facebook/react-native/pull/34533
Test Plan:
No exact test plan. However, if an exception is thrown during the
rendering process, a fatal `ClassNotFoundException` is raised, rather
than having the error be passed up to the `ReactNativeManager`.
Fixes: facebook/yoga#1120
Reviewed By: amir-shalem
Differential Revision: D39133326
Pulled By: jacdebug
fbshipit-source-id: 86283b7d21aed49ed0e9027b2aef85f0108cdf9a
Summary:
With react-native 0.70-rc.3 and new arch, codegen may fail if it encounters `.d.ts` files because specs may appear to be unused.
## Changelog
[General] [Fixed] - Codegen should ignore `.d.ts` files
Pull Request resolved: https://github.com/facebook/react-native/pull/34439
Test Plan:
See repro in https://github.com/microsoft/react-native-test-app/pull/1052. The build will fail without manually patching this in.
If you prefer to use your own test app, try adding [react-native-safe-area-context](https://github.com/th3rdwave/react-native-safe-area-context) as a dependency.
Reviewed By: cipolleschi
Differential Revision: D38826388
Pulled By: cortinico
fbshipit-source-id: eb7c9be2d49286bae86b2428862fbf20f6f32ca5
Summary:
This fix solves a problem very well evaluated [here](https://github.com/Expensify/App/issues/2727) as well as this [one](https://github.com/facebook/react-native/issues/29290).
The issue is that when the app goes to background, in landscape mode, the RCTDeviceInfo code triggers an orientation change event that did not physically happen. Due to that, we get swapped values returned when going back to the app.
I debugged the react-native code, and to me it seems that react native publishes the orientation change event one extra time when switching the state of the app to 'inactive'. Here is what is happening:
1. iPad is in landscape.
2. We move the app to inactive state.
3. Native Code queues portrait orientation change (no such change happened physically), and immediately after it triggers landscape change (same as we had in point 1).
4. We restore the app to active state.
5. The app receives two queued orientation change events, one after another.
6. The quick transition between portrait and landscape happens even though it never went back to portrait.
Fresh `react-native init` app repro case can be found here: https://github.com/lbaldy/issue-34014-repro
Video presenting the issue (recorded while working on: https://github.com/Expensify/App/issues/2727 ): https://www.youtube.com/watch?v=nFDOml9M8w4
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://github.com/facebook/react-native/wiki/Changelog
-->
[iOS] [Fixed] - Fix the way the orientation events are published, to avoid false publish on orientation change when app changes state to inactive
Pull Request resolved: https://github.com/facebook/react-native/pull/34014
Test Plan:
1. Make sure you have a working version of E/App.
2. Open App/src/components/withWindowDimensions.js and update the constructor by changing this line:
`this.onDimensionChange = _.debounce(this.onDimensionChange.bind(this), 100);`
to
`this.onDimensionChange = this.onDimensionChange.bind(this);`
3. Open the NewExpensify.xcodeproj in xCode.
4. Open the RCTDeviceInfo.mm file and replace it's contents with the file from this PR.
5. Select your device of choice (I suggest starting with iPad mini) and run the app though xCode.
6. From this point you can move to the test scenarios described below.
Reproduction + Fix test video (Test 1): https://youtu.be/jyzoNHLYHPo
Reproduction + Fix test video (Test 2): https://youtu.be/CLimE-Fba-g
**Test 1:**
1. Launch app in portrait, open chat - no sidebar visible.
7. Switch to landscape - sidebar shows.
8. Put app to background.
9. Put app back to foreground - make sure the side menu doesn't flicker.
**Test 2:**
1. Launch app in portrait, open chat - no sidebar visible.
2. Switch to landscape - sidebar shows.
3. Put app to background. Switch orientation back to portrait.
4. Put app back to foreground - make sure the side menu hides again as it should be in portrait.
Reproduction + Fix test video (Test 3, Test 4): https://youtu.be/EJkUUQCiLRg
iPad mini test 1 applies.
Scenario 2 does not as the screen is too wide in both orientations and iPad pro shows sidebar always.
**Test 3:**
1. launch the app.
2. Make sure you're in landscape mode.
3. See split screen with some other app. Make sure the side bar is visible.
4. Play with the size of the view, resize it a bit. When the view shrinks it should hide the sidebar, when it grows it should show it.
10. Move the app to background and back to foreground, please observe there are no flickers.
**Test 4:**
1. Launch the app.
2. Make sure you're in landscape mode.
3. Make the multitasking view and make Expensify app a slide over app.
4. Move back to fullscreen/split screen. Make sure the menu is shown accordingly
5. Move the app to background and back to foreground, please observe there are no flickers.
Non reg with and without the fix video: https://youtu.be/kuv9in8vtbk
Please perform standard smoke tests on transformation changes.
Reviewed By: cipolleschi
Differential Revision: D37239891
Pulled By: jacdebug
fbshipit-source-id: e6090153820e921dcfb0d823e0377abd25225bdf
Summary:
The current implementation of **getDefaultJSExecutorFactory** relies solely on try catch to load the correct .so file for jsc or hermes based on the project configuration.
Relying solely on try catch block and loading jsc even when project is using hermes can lead to launch time crashes especially in monorepo architectures and hybrid apps using both native android and react native.
So we can make use of an additional **ReactInstanceManager :: setJsEngineAsHermes** method that accepts a Boolean argument from the host app while building ReactInstanceManager which can tell which library to load at startup in **ReactInstanceManagerBuilder** which will now have an enhanced getDefaultJSExecutorFactory method that will combine the old logic with the new one to load the dso files.
The code snippet in **ReactInstanceManager** for adding a new setter method:
```
/**
* Sets the jsEngine as JSC or HERMES as per the setJsEngineAsHermes call
* Uses the enum {link JSInterpreter}
* param jsEngine
*/
private void setJSEngine(JSInterpreter jsEngine){
this.jsEngine = jsEngine;
}
/**
* Utility setter to set the required JSEngine as HERMES or JSC
* Defaults to OLD_LOGIC if not called by the host app
* param hermesEnabled
* hermesEnabled = true sets the JS Engine as HERMES and JSC otherwise
*/
public ReactInstanceManagerBuilder setJsEngineAsHermes(boolean hermesEnabled){
if(hermesEnabled){
setJSEngine(JSInterpreter.HERMES);
}
else{
setJSEngine(JSInterpreter.JSC);
}
return this;
}
```
The code snippet for the new logic in **ReactInstanceManagerBuilder**:
1) Setting up the new logic:
Adding a new enum class :
```
public enum JSInterpreter {
OLD_LOGIC,
JSC,
HERMES
}
```
A setter getting boolean value telling whether to use hermes or not and calling a private setter to update the enum variable.
```
/**
* Sets the jsEngine as JSC or HERMES as per the setJsEngineAsHermes call
* Uses the enum {link JSInterpreter}
* param jsEngine
*/
private void setJSEngine(JSInterpreter jsEngine){
this.jsEngine = jsEngine;
}
/**
* Utility setter to set the required JSEngine as HERMES or JSC
* Defaults to OLD_LOGIC if not called by the host app
* param hermesEnabled
* hermesEnabled = true sets the JS Engine as HERMES and JSC otherwise
*/
public ReactInstanceManagerBuilder setJsEngineAsHermes(boolean hermesEnabled){
if(hermesEnabled){
setJSEngine(JSInterpreter.HERMES);
}
else{
setJSEngine(JSInterpreter.JSC);
}
return this;
}
```
2) Modifying the getDefaultJSExecutorFactory method to incorporate the new logic with the old one:
```
private JavaScriptExecutorFactory getDefaultJSExecutorFactory(
String appName, String deviceName, Context applicationContext) {
// Relying solely on try catch block and loading jsc even when
// project is using hermes can lead to launch-time crashes especially in
// monorepo architectures and hybrid apps using both native android
// and react native.
// So we can use the value of enableHermes received by the constructor
// to decide which library to load at launch
// if nothing is specified, use old loading method
// else load the required engine
if (jsEngine == JSInterpreter.OLD_LOGIC) {
try {
// If JSC is included, use it as normal
initializeSoLoaderIfNecessary(applicationContext);
JSCExecutor.loadLibrary();
return new JSCExecutorFactory(appName, deviceName);
} catch (UnsatisfiedLinkError jscE) {
if (jscE.getMessage().contains("__cxa_bad_typeid")) {
throw jscE;
}
HermesExecutor.loadLibrary();
return new HermesExecutorFactory();
}
} else if (jsEngine == JSInterpreter.HERMES) {
HermesExecutor.loadLibrary();
return new HermesExecutorFactory();
} else {
JSCExecutor.loadLibrary();
return new JSCExecutorFactory(appName, deviceName);
}
}
```
### **Suggested changes in any Android App's MainApplication that extends ReactApplication to take advantage of this fix**
```
builder = ReactInstanceManager.builder()
.setApplication(this)
.setJsEngineAsHermes(BuildConfig.HERMES_ENABLED)
.setBundleAssetName("index.android.bundle")
.setJSMainModulePath("index")
```
where HERMES_ENABLED is a buildConfigField based on the enableHermes flag in build.gradle:
`def enableHermes = project.ext.react.get("enableHermes", true)
`
and then
```
defaultConfig{
if(enableHermes) {
buildConfigField("boolean", "HERMES_ENABLED", "true")
}
else{
buildConfigField("boolean", "HERMES_ENABLED", "false")
}
}
```
Our app was facing a similar issue as listed in this list: **https://github.com/facebook/react-native/issues?q=is%3Aissue+is%3Aopen+DSO**. Which was react-native trying to load jsc even when our project used hermes when a debug build was deployed on a device using android studio play button.
This change can possibly solve many of the issues listed in the list as it solved ours.
## Changelog
[GENERAL] [ADDED] - An enum JSInterpreter in com.facebook.react package:
```
/**
* An enum that specifies the JS Engine to be used in the app
* Old Logic uses the legacy code
* JSC/HERMES loads the respective engine using the revamped logic
*/
public enum JSInterpreter {
OLD_LOGIC,
JSC,
HERMES
}
```
[GENERAL] [ADDED] - An enum variable storing the default value of Js Engine loading mechanism in ReactInstanceManagerBuilder:
```
private JSInterpreter jsEngine = JSInterpreter.OLD_LOGIC;
```
[GENERAL] [ADDED] - A new setter method and a helper method to set the js engine in ReactInstanceManagerBuilder:
```
/**
* Sets the jsEngine as JSC or HERMES as per the setJsEngineAsHermes call
* Uses the enum {link JSInterpreter}
* param jsEngine
*/
private void setJSEngine(JSInterpreter jsEngine){
this.jsEngine = jsEngine;
}
/**
* Utility setter to set the required JSEngine as HERMES or JSC
* Defaults to OLD_LOGIC if not called by the host app
* param hermesEnabled
* hermesEnabled = true sets the JS Engine as HERMES and JSC otherwise
*/
public ReactInstanceManagerBuilder setJsEngineAsHermes(boolean hermesEnabled){
if(hermesEnabled){
setJSEngine(JSInterpreter.HERMES);
}
else{
setJSEngine(JSInterpreter.JSC);
}
return this;
}
```
[GENERAL] [ADDED] - Modified **getDefaultJSExecutorFactory** method
```
private JavaScriptExecutorFactory getDefaultJSExecutorFactory(
String appName, String deviceName, Context applicationContext) {
// Relying solely on try catch block and loading jsc even when
// project is using hermes can lead to launch-time crashes especially in
// monorepo architectures and hybrid apps using both native android
// and react native.
// So we can use the value of enableHermes received by the constructor
// to decide which library to load at launch
// if nothing is specified, use old loading method
// else load the required engine
if (jsEngine == JSInterpreter.OLD_LOGIC) {
try {
// If JSC is included, use it as normal
initializeSoLoaderIfNecessary(applicationContext);
JSCExecutor.loadLibrary();
return new JSCExecutorFactory(appName, deviceName);
} catch (UnsatisfiedLinkError jscE) {
if (jscE.getMessage().contains("__cxa_bad_typeid")) {
throw jscE;
}
HermesExecutor.loadLibrary();
return new HermesExecutorFactory();
}
} else if (jsEngine == JSInterpreter.HERMES) {
HermesExecutor.loadLibrary();
return new HermesExecutorFactory();
} else {
JSCExecutor.loadLibrary();
return new JSCExecutorFactory(appName, deviceName);
}
}
```
Pull Request resolved: https://github.com/facebook/react-native/pull/33952
Test Plan:
The testing for this change might be tricky but can be done by following the reproduction steps in the issues related to DSO loading here: https://github.com/facebook/react-native/issues?q=is%3Aissue+is%3Aopen+DSO
Generally, the app will not crash anymore on deploying debug using android studio if we are removing libjsc and its related libraries in **packagingOptions** in build.gradle and using hermes in the project.
It can be like:
```
packagingOptions {
if (enableHermes) {
exclude "**/libjsc*.so"
}
}
```
Reviewed By: lunaleaps
Differential Revision: D37191981
Pulled By: cortinico
fbshipit-source-id: c528ead126939f1d788af7523f3798ed2a14f36e
Summary:
In https://github.com/facebook/react-native/pull/32695, the `Performance.now()` implementation changed to use unix epoch timestamps instead of a monotonic clock.
This is problematic, because it means that performance measurements get skewed if the device clock changes between two measurements.
With this change, the clock is now monotonic (and the implementation stays consistent between platforms).
More details and repro steps can be found in [this issue](https://github.com/facebook/react-native/issues/33977)
Closes https://github.com/facebook/react-native/issues/33977
## Changelog
[General] [Fixed] - Use monotonic clock for performance.now()
Pull Request resolved: https://github.com/facebook/react-native/pull/33983
Test Plan:
Run on iOS and Android:
```
const now = global.performance.now()
console.log(`${Platform.OS}: ${now}`)
```
Reviewed By: JoshuaGross, cipolleschi
Differential Revision: D37066999
Pulled By: dmitryrykun
fbshipit-source-id: 298547bf39faea1b025c17ff2d2e1a03f929865b
Summary:
As per commit: https://github.com/facebook/react-native/commit/4f3b17412018a10f9293247c802598d2b94a844b which states that "React Native requires that the RootView id be managed entirely by React Native, and will crash in addRootView/startSurface if the native View id isn't set to NO_ID."
This behaviour can not be guaranteed in **hybrid** apps that have a native android layer over which ReactRootViews are added and the native views need to have ids on them in order to work. **The control of views can jump back and forth between native android and react-native (fabric). As the ReactRootView is added to ViewGroups (or layouts) in Android Fragments and Activities, they contain ids on their views which might get passed down to the reactRootView by features like DataBinding**
Hence this can cause unnecessary crashes at runtime for hybrid apps even when they are not changing the id of the reactRootView object they are adding to their ViewGroups.
Our app is a hybrid app that uses both native android and react-native on different screens and on one such screen that has a Fragment adding a ReactRootView to its FrameLayout to render native android views to render in ReactNative, this crash occurs on pressing the back button as well as on unlocking the screen while staying on the same screen.
The app was running fine on more than a 100 million devices on React Native 0.63.4 but after updating to 0.67.2, that features this commit, it crashes on the very first device it was tested on.
Refer to the issue: https://github.com/facebook/react-native/issues/33121 for more information on the crash
The fragment in which this issues arises is like this:
```binding.frameLayout.addView(getReactRootView())```
where getReactRootView() is like this:
```
private var mReactRootView: ReactRootView? = null
private var mReactInstanceManager: ReactInstanceManager? = null
mReactRootView = ReactRootView(context)
if (activity != null) {
val application = activity?.application
if (application is MainApplication) {
mReactInstanceManager = application.reactInstanceManager
}
}
fun getReactRootView():View?{
return mReactRootView
}
```
So converting this to a soft exception such that pure react-native devs can still see the error while hybrid apps continue to run without crashes.
### Snippet of the change:
```
if (getId() != View.NO_ID) {
ReactSoftExceptionLogger.logSoftException(
TAG,
new IllegalViewOperationException(
"Trying to attach a ReactRootView with an explicit id already set to ["
+ getId()
+ "]. React Native uses the id field to track react tags and will overwrite this"
+ " field. If that is fine, explicitly overwrite the id field to View.NO_ID."));
}
```
## Changelog
[GENERAL] [ADDED] - A ReactSoftException log instead of a direct exception being thrown:
```
if (getId() != View.NO_ID) {
ReactSoftExceptionLogger.logSoftException(
TAG,
new IllegalViewOperationException(
"Trying to attach a ReactRootView with an explicit id already set to ["
+ getId()
+ "]. React Native uses the id field to track react tags and will overwrite this"
+ " field. If that is fine, explicitly overwrite the id field to View.NO_ID."));
}
```
[GENERAL] [REMOVED]- Directly throwing an exception even when the code is not responsible for this issue:
```
if (getId() != View.NO_ID) {
throw new IllegalViewOperationException(
"Trying to attach a ReactRootView with an explicit id already set to ["
+ getId()
+ "]. React Native uses the id field to track react tags and will overwrite this"
+ " field. If that is fine, explicitly overwrite the id field to View.NO_ID.");
}
```
Pull Request resolved: https://github.com/facebook/react-native/pull/33133
Test Plan:
This crash is hard to reproduce but when it occurs, this is the only way to fix it.
If any app used to crash with this exact error, it will no longer crash but show an error log in Logcat for developers to be informed about the issue.
Reviewed By: ShikaSD
Differential Revision: D34304212
Pulled By: cortinico
fbshipit-source-id: f0eaeef2e905a6e0587df088b43cc49cabda397a
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33611
Bumping the NDK to 23 to prevent build failures due to the NDK
using longer paths.
Changelog:
[Android] [Fixed] - Use NDK 23 only for Windows users.
Reviewed By: motiz88
Differential Revision: D35547459
fbshipit-source-id: 4e44c0b5fd8d1c559b04fb43eb4eeadc13943394
Summary:
This adds an unnecessary dependency between two NPM package which can be avoided. See https://github.com/reactwg/react-native-releases/discussions/17#discussioncomment-2452813
for context.
Changelog:
[Internal] [Changed] - react-native-gradle-plugin should not depend on react-native-codegen NPM package
Reviewed By: dmitryrykun
Differential Revision: D35279729
fbshipit-source-id: f18f79809f115f28203ac0a843fafead63528904
Summary:
It is not necessary to call node via yarn. Instead with this commit node is called directly (windows aware). This enables builds on systems that don't have yarn installed.
Fixes https://github.com/facebook/react-native/issues/33525
## Changelog
[Android] [Fixed] - Don't require yarn for codegen tasks
Pull Request resolved: https://github.com/facebook/react-native/pull/33530
Test Plan:
1. react-native init test
2. cd test
3. enable newArchEnabled=true (gradle.properties)
4. enable enableHermes: true (build.gradle)
5. react-native run-android (when the yarn is not installed on the system)
(I have not tested or verified if this works on windows build machines)
Reviewed By: sshic
Differential Revision: D35279376
Pulled By: cortinico
fbshipit-source-id: 430e4a7bcdec7d5377efac747f6b935d634451cc
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33581
We currently have some code on the RN Gradle Plugin that we need to ship.
There are both bugfixes needed for RN 0.68.1 and for the current nightly
(therefore needed for RN 0.69).
I've verified that this works on a fresh RN 0.68.0 install with `npm pack`
and triggering a build on a fresh setup from app template, with newArchEnabled set to true.
Changelog:
[Android] [Changed] - Bump React Native Gradle plugin to 0.0.6
Reviewed By: rubennorte
Differential Revision: D35439444
fbshipit-source-id: 662b2211c44b261e3e3c9cddc946107cfb495c79
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33588
Currently users on M1 machine can't use the New Architecture correctly as they will get build failures when building the native code.
This Diff fixes it by automatically recognizing the host architecture and switching to NDK 24 if user is runnign on `aarch64`
Changelog:
[Android] [Fixed] - Improve support for Android users on M1 machine
Reviewed By: mdvacca
Differential Revision: D35468252
fbshipit-source-id: b73f5262b9408f04f3ae4fd26458a4d17c1ec29a
Summary:
As users can toggle `enableSeparateBuildPerCPUArchitecture` to create a split APK, once that is off, the `-PreactNativeArchitecture` is not correctly considered when building the local module.
This will make sure that, if users have `enableSeparateBuildPerCPUArchitecture` set to `false`, their
app is building the local `app_modules` only for the requested architectures.
Practically, users invoking with `--active-arch-only` might experience a build failure if they have a fully clean environment (would be forced to do a full build before using `--active-arch-only`). This addresses this scenario.
Changelog:
[Android] [Fixed] - Template: Specify abiFilters if enableSeparateBuildPerCPUArchitecture is not set.
Reviewed By: ShikaSD
Differential Revision: D35250700
fbshipit-source-id: 4e555888636cf182495fab2b4a562d93a70b9e66
Summary:
This PR fixes RTTI (run-time type information) for ShadowNodeWrapper and ShadowNodeListWrapper classes, i.e., calls to dynamic_cast and dynamic_pointer_cast that are called via JSI's getHostObject calls.
The fix is simply to add a so-called "key function" in a form of virtual destructor. Key functions needs to be a virtual non-pure and non-inlined functions that points the compiler as to which library contains the vtable/type information for a given class (see https://itanium-cxx-abi.github.io/cxx-abi/abi.html#vague-vtable and https://developer.android.com/ndk/guides/common-problems#rttiexceptions_not_working_across_library_boundaries)
Without the "key function", calls to dynamic_cast for ShadowNodeWrapper instances won't work across library boundaries because the class will have separate definitions in each separate library, therefore objects created in one of those libraries won't be recognized as the same type by the other library. This has been a problem in reanimated and gesture-handler libraries where we call `object.getHostObject<ShadowNodeWrapper>(rt)` (this is a method from JSI) in order to access ShadowNode instance from a handle we have in JS. I think, this issue is going to be relevant to more libraries that cope with view instances. In this scenario, we have a separate library, say "libreanimated.so" that calls to `getHostObject` which is an inline function that calls `dynamic_cast` for the `ShadowNodeWrapper` class. On the other hand, the instances of `ShadowNodeWrapper` are created by the code from `libreact_render_uimanager.so`. Because of that `dynamic_cast` fails even though it is called on instance of `ShadowNodeWrapper` because the class has separate vtable/type info: one in `libreanimated.so` and one in `libreact_render_uimanager.so` (by "fails" I mean that it actually returns `nullptr`).
This problem has been documented here: https://developer.android.com/ndk/guides/common-problems#rttiexceptions_not_working_across_library_boundaries where the solution is for the class to have a so-called "key function". The key function makes it so that compiler sees that one of the implementation for a given class is missing and therefore can safely assume that a vtable/type info for a given class is embedded into some library we link to.
This change adds a virtual destructor that is declared in the header file but defined in file that gets compiled as a part of `libreact_render_uimanager`. As a result, the compiler only creates one vtable/type info and calls to dynamic_cast works as expected in all libraries for `ShadowNodeWrapper` and `ShadowNodeListWrapper` classes.
This issue would only surface on Android, because on iOS all libraries by default are bundled together via Pods, whereas on Android each library is loaded separately using dynamic loading.
## Changelog
[Fabric][Android specific] - Fix dynamic_cast (RTTI) for ShadowNodeWrapper and similar classes when accessed by third-party libraries.
Pull Request resolved: https://github.com/facebook/react-native/pull/33500
Test Plan:
1. In order to test this you need to add a library that'd include `<react/renderer/uimanager/primitives.h>` (i.e. use this branch of reanimated library: https://github.com/software-mansion/react-native-reanimated/tree/fabric)
2. After compiling the app inspect libreact_render_uimanager.so and libreanimated.so artifacts with `nm` tool
3. Notice that symbols like `vtable for facebook::react::ShadowNodeWrapper` and `typeinfo for facebook::react::ShadowNodeWrapper` are only present in the former and not in the latter library (before this change you'd see them both)
Reviewed By: ShikaSD
Differential Revision: D35143600
Pulled By: javache
fbshipit-source-id: 5fb25a02365b99a515edc81e5485a77017c56eb8
On Windows there are limits on number of character in file paths and in command lines
Ref: https://android.googlesource.com/platform/ndk/+/master/docs/BuildSystemMaintainers.md#Path-Length-Limits
NDK allows circumventing command line limits using response(RSP) files as inputs using NDK_APP_SHORT_COMMANDS flag.
Windows can support long file paths if configured through registry or by prefixing all file paths with a special character sequence
The latter requires changes in NDK. And there are tools in NDK (AR) which is not able to handle long paths (>256) even after setting the registry key.
The new architecutre source tree is too deep, and the object file naming conventions in NDK makes the matters worse, by producing incredibly long file paths.
Other solutions such as symlinking source code etc. didn't work as expected, and makes the build scripts complicated and hard to manage.
This change temporarily works around the issue by placing the temporary build outputs as short a path as possible within the project path.
Changelog:
[Android] [Fixed] - Fix for building new architecture sources on Windows
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33499
This DIFF turns on the `FB_SONARKIT_ENABLED` flag when installing Flipper ina RN app. The flag is enabled only in Debug config, given that Flipper is installed only in this configuration.
This PR also fixes this issue: https://github.com/facebook/react-native/issues/33497
This PR is required because release 0.67 has the Flag in the app, while release 0.68 moved it in the React-Core pod.
We can't enable the flag at the `React-Core.podspec` level because we should not make assumptions on whether users want flipper or not.
## Changelog
[iOS][Changed] - Enable SonarKit in React-Core when the configuration is `'Debug'`
Reviewed By: cortinico
Differential Revision: D35141506
fbshipit-source-id: 171b7fa8ea7727c633ef963408e86b332c32e9fa
Summary:
Flipper-DoubleConversion and Flipper-Glog iOS pods received a build optimization
a few versions back that pre-compiled the pods and references the xcframework slices
Unfortunately, the pre-compile did not include macCatalyst slices, so this disabled support
for flipper on macOS for react-native >0.65
lblasa has re-compiled the pods with the macCatalyst slices added
See https://github.com/facebook/flipper/issues/3117
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://github.com/facebook/react-native/wiki/Changelog
-->
[iOS] [Fixed] - update Flipper pods to support re-enable macCatalyst
Pull Request resolved: https://github.com/facebook/react-native/pull/33406
Test Plan:
- [ ] The Flipper repo has a react-native test that appeared to work with these versions, CI should work here
- [ ] Prove there is no regression, a flipper-enabled build test should work for simulator iOS target on arm64
- [ ] Prove there is no regression, a flipper-enabled build test should work for simulator iOS target on x86_64 mac
- [ ] Prove there is no regression, a flipper-enabled build test should work for real device iOS target
- [ ] To prove the issue is resolved, a build should be attempted for a macCatalyst target, and it should work.
Reviewed By: cortinico
Differential Revision: D34789654
Pulled By: lblasa
fbshipit-source-id: 466803dd07b5820220512b7d7d760b94b8aa65f7
Summary:
This re-applies Consider relative to pwd installation root when looking for files in rn module via cocoapods by danilobuerger
The `:reactNativePath` provided by `use_native_modules!` is the rn module path relative to the installation root (usually `./ios`). However, when executing cocoapods from a dir thats not the installation root, packages that use the relative `:reactNativePath` variable in their path must also consider the relative to pwd installation root.
This fixes usage of cocoapods with the `--project-directory` flag like
```bash
bundle exec pod install --project-directory=ios
```
## Changelog
[iOS] [Fixed] - Fix usage of cocoapods with --project-directory flag and new arch
Pull Request resolved: https://github.com/facebook/react-native/pull/33427
Test Plan:
1) Enable the new arch
2) Execute from the projects root dir
```bash
bundle exec pod install --project-directory=ios
```
3) It will fail with
```
[!] Invalid `Podfile` file: [codegen] Couldn't not find react-native-codegen..
```
4) Apply the patch
5) Execute from the projects root dir
```bash
bundle exec pod install --project-directory=ios
```
6) It will succeed
Reviewed By: cortinico
Differential Revision: D34890926
Pulled By: dmitryrykun
fbshipit-source-id: f7adc6196874822d5ff38f275414f529d385f2ea
Summary:
Avoid breaking tools relying on absolute path for `cliPath`
## Changelog
[Android] [Fixed] - Enable cliPath to have an absolute path value
Pull Request resolved: https://github.com/facebook/react-native/pull/32983
Test Plan:
declare `cliPath` from `expo`:
```groovy
cliPath: new File(["node", "--print", "require.resolve('react-native/package.json')"].execute(null, rootDir).text.trim()).getParentFile().getAbsolutePath() + "/cli.js",
```
and run an android build
Reviewed By: ShikaSD
Differential Revision: D33843275
Pulled By: cortinico
fbshipit-source-id: 65f55a5e07a4ec0a6205d5f06f150377708c30cc
Nick Gerleman
Marek Fořt
Distiller
Nicola Corti
Distiller
Thibault Malbranche
Distiller
Dmitry Rykun
David
Evan Charlton
Tommy Nguyen
lbaldy
Distiller
Lorenzo Sciandra
Sparsha Saha
Kunal Farmah
Olivier Payen
Kunal Farmah
Lorenzo Sciandra
Distiller
Anandraj
Distiller
Danilo Bürger
Krzysztof Magiera
Distiller
Distiller
Riccardo Cipolleschi
Distiller
Andrei Shikov
Mike Hardy
Krisztiaan