Summary: This diff initializes Fabric C++ UIManager code from java, it also registers android RootViews inside C++.
Reviewed By: shergin
Differential Revision: D8878148
fbshipit-source-id: 8b7924f715b135eda37bb2683206d3e321a2b7b2
Summary: This diff creates the C++ base classes for Fabric and it integrates them into the starting process of Fabric inside Catalyst App
Reviewed By: shergin
Differential Revision: D8878123
fbshipit-source-id: ce41ad6addfdfb58e602b2dbafebd8f05847e69f
Summary:
These changes will fix executing javascript with any special characters, by making use of the `evaluateJavascript` function on Android 4.4+, and by properly escaping the URI on Android <4.4.
Fixes#19611 • Fixes#20365 • Fixes#9749 • Closes#19655 • Closes#12321
This PR supersedes #19655 by patching the same problem in all the places, and fixing it for Android <4.4 as well.
Pull Request resolved: https://github.com/facebook/react-native/pull/20366
Differential Revision: D9242968
Pulled By: hramos
fbshipit-source-id: f2e1abc786ba333dbd8aaa8922e716fd99ec26e0
Summary:
ag -L --ignore __snapshots__ 'flow strict|noflow|generated|The controller you requested could not be found.' | ag '\.js$' | xargs ag -l 'flow' | sort > ~/temp
cat ~/temp | xargs ag -L 'flow strict' | xargs sed -i '' 's/flow$/flow strict-local/'
until flow check; do flow check --json | jq -r '.errors[].message[0].path' | sort | uniq | xargs hg revert; done
allow_many_files
The controller you requested could not be found.
The controller you requested could not be found.
Reviewed By: TheSavior
Differential Revision: D9004573
fbshipit-source-id: 936bd5741706b781be06bf08b6ad805a69407dfd
Summary:
Currently, we scan all native modules to see if they implement the OnBatchCompleteListerner. If they do, we add those modules to a list, and when C++ calls OnBactchComplete is called, we execute the callback on each of the modules.
The only native module implementing this callback today is the UIManagerModule. With Fabric, UIManager will also not be a native module anymore. This diff removes all the work done for creating the list and assumes that UIManagerModule is the only place that is interested in OnBatchComplete call - and calls it directly.
Reviewed By: achen1
Differential Revision: D9186651
fbshipit-source-id: 473586b37c2465ccd041985dcdd56132026f34f1
Summary:
This is a small missing piece from b6f2aad9c0 and will complete fresco version bump.
Pull Request resolved: https://github.com/facebook/react-native/pull/20580
Differential Revision: D9228658
Pulled By: hramos
fbshipit-source-id: 9cf1509a590cedfe4c3358006cfb62533058e946
Summary:
Implemented a version of accessibility Hints on android by adding hint text to the end of contentDescription. There is already an existing prop on js and iOS implementation.
Changes involve:
* adding a prop on native android code for accessibilityHints
* setting the accessibilityDelegate after the props are all loaded.
* Instead of directly updating the accessibility delegate, the prop setters now update state by setting the values of the variables. Once all props are set, the accessibility delegate is set based on the props
* BaseViewManager keeps track of whether or not accessibility props have been set
* AccessibilityDelegateUtil keeps track of what the props have been set to. (Renamed AccessibilityRoleUtil to AccessibilityDelegateUtil)
Currently, this is the easiest way of emulating the way accessibility hints work on iOS, and I think having an android counter part is better than having nothing.
It's different from iOS in that it will announce the hint before the role, and also cannot be turned off.
Ex:
if I set the accessibility like this:
```
<View
style={styles.container}
accessible={true}
accessibilityLabel="accessibility label"
accessibilityRole="button"
accessibilityStates={['selected']}
accessibilityHint="accessibility Hint">
<Text> Tester </Text>
</View>
```
Talk back will read:
`accessibility label, accessibility Hint, button, selected`
In the future for next steps, I plan on investigating the process of making a second announcement after the first
Reviewed By: achen1
Differential Revision: D9037226
fbshipit-source-id: 8d484e1114eb69aa5f5314b3755b351b8ea80b09
Summary:
@public
This reverts the Yoga/Java storage experiment. I will follow up with any learnings.
Reviewed By: pasqualeanatriello
Differential Revision: D9168405
fbshipit-source-id: fb227fb9353bd4c4e3bebbe9b04eec1132e532e8
Summary:
@public
Adds another version of property storage for `YogaNode`, using `sun.misc.Unsafe`.
Adopts the stub concept from Litho for `Unsafe`, as it is hidden by the Android SDK.
Reviewed By: pasqualeanatriello
Differential Revision: D9140103
fbshipit-source-id: a4b376eca341b724a00f873467ae8bf8eaac69f4
Summary: The method removes all settings for mLazyReactModules since Lazy and non lazy modules can exist in a single application now
Reviewed By: achen1
Differential Revision: D9012312
fbshipit-source-id: 0420149654f8146453250d83d4de4b4c2fd31e9f
Summary:
In OSS, during gradle build, the ReactModuleSpecProcess annotation-processor does not run. As a result, the `ReactModuleInfo` that we need for CoreReactPackage is not generated, resulting in a runtime error.
The fix is to make LazyReactPackage revert to what it was doing earlier. In `NativeModuleRegistryBuilder`, if we dont find `ReactModuleInfo` for any `ModuleSpec`, we eagerly instantiate the module and get all the `ReactModuleInfo` from it. By returning an emoy collection if the class is not available, we force the modules in `CoreReactPackage` to use this codepath instead.
The alternate fix would be to ensure that the annotation processor runs in gradle/OSS. However, the annotation processor will be removed eventually in the future, and we will also be move to generating them for JS, so that work will soon be irrelevant.
Reviewed By: fkgozali, achen1
Differential Revision: D9130517
fbshipit-source-id: 469cf0e32a2f3650f098547667b3cd09a63eb1a0
Summary:
/cc janicduplessis mdvacca
This addresses the same issue as #18830 which was reverted since it didn’t handle `removeClippedSubviews` properly.
When subview clipping is on, ReactViewGroup keeps track of its own children directly and accounts for the offset introduced by clipped views when calling ViewGroup methods that modify children.
Instead of accounting for just clipped children (views with no parent), it should account for any children that aren’t in the ViewGroup which also includes children that are being transitioned. If you look at the ViewGroup source code, [it explicitly retains the view parent until the transition finishes](https://github.com/aosp-mirror/platform_frameworks_base/blob/oreo-release/core/java/android/view/ViewGroup.java#L5034-L5036) which caused the `getParent()` checks to pass, even though those views should be ignored. I added a new `isChildInViewGroup` method that handles both clipped and transitioning views to fix this.
I reproduced the [earlier crash](https://github.com/facebook/react-native/pull/18830#issuecomment-382798628) by enabling clipping in [this test app](https://github.com/facebook/react-native/pull/18830#pullrequestreview-111758886) and adding a “clear views” button that resets the state to an empty items array with an animation.
- #18830
[ANDROID] [BUGFIX] [LayoutAnimation] - Removal LayoutAnimations no longer remove adjacent views as well in certain cases.
Pull Request resolved: https://github.com/facebook/react-native/pull/19775
Differential Revision: D9105838
Pulled By: hramos
fbshipit-source-id: 5ccb0957d1f46c36add960c0e4ef2a545cb03cbe
Summary:
- adds more property assignments
- reduces the number of layout roots that exist simultanously
Reviewed By: pasqualeanatriello
Differential Revision: D8989389
fbshipit-source-id: 6a0ac800a4caad61a2f4bf98caa314855b70875f
Summary: If source is null , source uri is null or source is not an array should respect style like in iOS
Reviewed By: yungsters
Differential Revision: D9018005
fbshipit-source-id: 5f695e8e3007c96e6004973e7fcbc6b57cc15249
Summary:
This patch is a bit of a hack job, but I'd argue it's necessary to dramatically improve the dev UX on Android devices. Somewhere in react-native, there's a shared SerialExecutor which AsyncStorage uses that is getting blocked, causing remote debugging to occasionally hang indefinitely for folks making AsyncStorage requests. This is frustrating from a dev UX perspective, and has persisted across several versions as far back as RN 0.44, and still remains on RN 0.54.
The issue seems to only happen on Android > 7+, which is likely because the ThreadPoolExecutor behavior changed in this version:
https://stackoverflow.com/questions/9654148/android-asynctask-threads-limitsFixes#14101
We've been using this patch in production for the past 4 months on our team by overriding the AsyncStorage native module. We use AsyncStorage extensively for offline data and caching.
You can test by compiling this commit version into a test react native repository that is set to build from source:
```sh
git clone https://github.com/dannycochran/react-native rnAsyncStorage
cd rnAsyncStorage
git checkout asyncStorage
cd ..
git clone https://github.com/dannycochran/asyncStorageTest
yarn install
cp -r ../rnAsyncStorage node_modules/react-native
react-native run-android
```
No documentation change is required.
https://github.com/facebook/react-native/pull/16905
[Android] [BUGFIX] [AsyncStorage] - Fix AsyncStorage causing remote debugger to hang indefinitely.
Pull Request resolved: https://github.com/facebook/react-native/pull/18522
Differential Revision: D8624088
Pulled By: hramos
fbshipit-source-id: a1d2e3458d98467845cb34ac73f2aafaaa15ace2
Summary:
@public
Adds an implementation of `YogaNodeProperties` that sets style properties using a `ByteBuffer` rather than JNI calls.
We hope for a speed improvement.
Reviewed By: pasqualeanatriello
Differential Revision: D9042225
fbshipit-source-id: c7f2b24eaeddd1190755bec85a5034079bd2f492
Summary:
@public
Adds an implementation of `YogaNodeProperties` that accesses style and layout properties using a `ByteBuffer` rather than JNI calls.
We hope for a speed improvement.
This needs further cleanup after experimenting, e.g. to codegen the offsets.
Reviewed By: pasqualeanatriello
Differential Revision: D8911723
fbshipit-source-id: 3c24b57eb545155878896ebb5d64d4553eb6bedc
Summary:
Here we introduce an abstraction over node property storage, in order to experiment with different approaches for Java/C integration.
- interface `YogaNodeProperties` as abstraction
- current JNI code factored into `YogaNodePropertiesJNI.java`
- `YogaNode` delegates all calls, no API changes
Reviewed By: astreet
Differential Revision: D8769448
fbshipit-source-id: e67327ce41fa047a51a986c652b3d59992a510e2
Summary: Using strings as keys for module names instead of actuall `.class` prevents additional class loading
Reviewed By: achen1
Differential Revision: D8955006
fbshipit-source-id: 12e6ad35fc35c969c05ca081976cc05b054f6821
Summary:
When native modules use `LazyReactPackage`, the modules themselves are not initialized. However, they still use the class names, causing the classes to load. This diff removes the need to perform any class loads. Any properties of the classes that are required are now populated in the `ReactModuleInfo` of that class.
Note that this diff itself does not prevent class loading since any references to `*.class` in `LazyReactpackage` needs to be removed in a consequent diff
Reviewed By: achen1
Differential Revision: D8950025
fbshipit-source-id: 80ddf7e1f33bf2af0db1bd262069795de77ec611
Summary: `ReactModuleInfoProvider` was using `Class` as the keys. This would make the classes to load. This change makes the map use canonicalName Strings of the classes, removing the need for the classes to load at this point.
Reviewed By: achen1
Differential Revision: D8944078
fbshipit-source-id: 4aa562d74d6ad3ebb9962b581d4e2f5e89d47ca6
Summary:
Added a check for null on `accessibilityStates`.
Now, if a null value is passed in to the prop, it simply does not do anything.
Reviewed By: achen1
Differential Revision: D9034714
fbshipit-source-id: caffa41a1d8b08d5358d085e4e268e8e6f1c9344
Summary:
We were seeing a bug where embedded feed stories showed the loading spinner forever until a scroll event occurred. Embedded feed stories are built on top of our RN + Litho integration which requires additional logic to support remeasuring the shadow tree when the intrinsic size of the litho component changes (aka wrap_content). Previously, we simply overrode requestLayout() of ReactLithoView to inform the shadow tree that that node needs to be dirtied -- in which case a
subsequent layout pass would resolve these changes. This worked fine in dev builds, but it turns out that release builds never received a subsequent layout pass. Dev builds apparently have a loop that dispatches view updates every second or so while release doesn't (not sure why), but that was the reason why the dev version eventually invalidated the views. The solution here is to simply always invoke a view update when any embedded Litho component requires a relayout.
Reviewed By: mdvacca
Differential Revision: D8983995
fbshipit-source-id: 53c528b3c00aad2bbe8e7f6360dc8e1594da1298
Summary:
An application could either have lazy, or non-lazy modules. This diff simply lets the individual reactPackages decide if they should be lazy or not, based on the variable in `ReactInstanceManagerBuilder`.
This diff also removed the method `setLazyNativeModules()` since an app can now have both native and non-native modules.
Reviewed By: achen1
Differential Revision: D8940026
fbshipit-source-id: 0399f4f39ad57f2b03e4dce117a9e2c28c4ed2b1
Summary:
In D8884991, I added a strings.xml file to react-native-github UI Manager resources.
Because this gets deployed to open source, and not all apps have the same Locale support that Facebook has, we temporarily don't want these strings to be translated.
I added `translatable = false` to the strings in this diff.
Reviewed By: fkgozali
Differential Revision: D9018766
fbshipit-source-id: 04d06478d0ac665fa02387e6df035c31522fdea6