Summary:
See also D36889794. This is a very similar idea, except the core problem is that BaseTextProps is accessing the same props as ViewProps; and for ParagraphProps parsing, it first defers to ViewProps' parser and then BaseTextProps. RawPropsParser is optimized to access the same props in the same order, *exactly once*, so if we access a prop out-of-order, or for a second time, that access and the next access are deoptimized. Paragraph/Text, in particular, were quite bad because we did this several times, and each out-of-order access requires scanning /all/ props.
This fixes the issue, at least partially, by (1) pulling all the duplicate accesses to the beginning of BaseTextProps, and (2) accessing them all in the same order as ViewProps, relatively (some props are skipped, but that matters less).
Practically what this means is that now, all of Props' accesses have a cost of O(1) for lookup, or a total of O(n) for all of them; each access is at the n+1 position in the internal RawPropsParser array, so each access is cheap. BaseTextProps' duplicate accesses, even though there are only 4 of them: (1) the first one scans the entire array until we reach the prop in question; (2) the next accesses require scans, but not whole-array scans, since they're in order. (3) The BaseTextProps accesses /after/ the duplicate accesses are all O(1).
tl;dr is that before we had something like O(n*6) cost for BaseTextProps parsing and now it is O(n*2).
Similar to my summary in the last diff: we may want to revisit the RawPropsParser API... but I want to tread gently there, and this gets us a large improvement without major, risky changes.
Empirically, based on a couple of systraces, average time for a single UIManager::createNode called from JS thread, before this stack: 17us. After: 667ns (3% as long). On average, for every 60 createNode calls, we will save 1ms on the UI thread. The savings will be greater for certain screens that use many Views or Text nodes, and lesser for screens that use fewer of these components.
Changelog: [Internal]
Reviewed By: mdvacca
Differential Revision: D36890072
fbshipit-source-id: 5d24b986c391d7bb158ed2f43d130a71960837d1
Summary:
Without getting into the weeds too much, RawPropParser "requires" that props be accessed in the same order every time a Props struct is parsed in order to most optimally fetch the values in a linear-ish fashion, basically ensuring that each rawProps.at() call is an O(1) operation, and overall getting all props for a particular component is O(n) in the number of props for a given struct. If props are called out of order, this breaks and worst-case we can end up with an O(n^2) operation.
Unfortunately, calling .at(x) twice with the same prop name triggers the deoptimized behavior. So as much as possible, always fetch exactly once and in the same order every time. In this case, we move initialization of two fields into the constructor body so that we can call .at() a single time instead of twice.
In the debug props of ViewProps I'm also reordering the fields to fetch them in the same order the constructor fetches them in, which will make this (debug-only) method slightly faster.
What's the impact of this? If you dig into the Tracery samples, the average/median RawPropsParser::at takes 1us or less. However, in /every single/ call to createNode for View components, there is at least one RawPropsParser::at call that takes 250+us. This was a huge red flag when analyzing traces, after which it was trivial (for View) to find the offending out-of-order calls. Since this is happening for every View and every type of component that derives from View, that's 1ms lost per every 4 View-type ShadowNodes created by ReactJS. After just 100 views created, that's 25ms. Etc.
There are other out-of-order calls lurking in the codebase that can be addressed separately. Impact scales with the size of the screen, the number of Views they render, etc.
Changelog: [Internal]
Reviewed By: mdvacca
Differential Revision: D36889794
fbshipit-source-id: 91e0a7ca39ed10778e60a0f0339a4b4dc8b14436
Summary: Creates a base TurboModule for Catalyst.
Reviewed By: ann-ss
Differential Revision: D36707915
fbshipit-source-id: 9a246f238efdcdde45e0332f7b868478deadf50e
Summary:
I'm backporting PR https://github.com/facebook/react-native/pull/33945 to main
as it was merged on the release branch to unblock 0.69.
Those changes are necessary as Hermes was not being donwloaded at all during `pod install`
on .69 and the app was failing to build with a missing `hermes/hermes.h` import.
Changelog:
[iOS] [Fixed] - Fix Hermes not being properly downloaded during pod install
Reviewed By: hramos
Differential Revision: D36810787
fbshipit-source-id: f898e61b6536dfcfc81feeff740703fbd697b000
Summary:
Right now, when we change the keyboardType on android between between default and email, the value keyboard type stays as email (specially noticeable with the key next to the spacebar, that changes between the comma (`,`) to the at sign (`@`)).
This is because the mask we are using when updating the input is only taking into account the class, and not the flags nor the variations.
We don't apply all masks because it may interfere with flags assigned by other props, like multiline or secure text entry. Therefore, we have created our own mask, taking into account all the variations and flags that the keyboardType prop may set. This may be hard to maintain, since whenever we add any other keyboard type, we have to take these flags into mind.
The error I was trying to fix was in particular regarding going back and forward from email, but this fix may solve other similar issues with other keyboard styles.
## Changelog
[Android] [Fixed] - Fix a bug where the keyboard, once set as email, won't change back to default.
Pull Request resolved: https://github.com/facebook/react-native/pull/33924
Test Plan: In order to test this PR, any test code with a TextInput, and a way to change the value of the keyboardType should work. We should be able to see how the keyboard changes to the correct type without staying, for example, on the email state.
Reviewed By: lunaleaps
Differential Revision: D36784563
Pulled By: makovkastar
fbshipit-source-id: 74d7b61b3c07feea4e4050d7a07603a68b98e835
Summary:
I was reading source code, and I noticed the _subscriptions attribute (and the _clearSubscriptions method) of the FileReader is not used.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://github.com/facebook/react-native/wiki/Changelog
-->
[Internal] [Changed] - Remove unused _subscriptions attribute in FileReader
Pull Request resolved: https://github.com/facebook/react-native/pull/33946
Test Plan: I checked that flow and jest are still green as I can't think about another way.
Reviewed By: lunaleaps
Differential Revision: D36807828
Pulled By: cortinico
fbshipit-source-id: 9bb599bbc7b79d2b4c010ba84cc8777e29b974ca
Summary:
tldr; `ReactEditText` and Android's emoji support in Android's AppCompat 1.4.0 / 1.4.x conflict in an odd way, causing NPEs.
`ReactEditText` defines an `InternalKeyListener`, `mKeyListener`, that it uses to make sure input from all keyboards works correctly. This listener is normally initialized at the end of the constructor.
Unfortunately, some versions of `AppCompatEditText`, most notably the version in the App Compat `1.4.0-alpha0x`, the [minimum version required for the Play Store's emoji compliance](https://developer.android.com/guide/topics/ui/look-and-feel/emoji2#appcompat) call `setInputType` from `AppCompatEditText`'s constructor. `ReactEditText` operates on the key listener inside of `setInputType` and since the `AppCompatEditText` constructor is called via call to `super` before the key listener is initialized, these versions of app compat can cause NPEs when used with React Native.
The fix is simple; check to see if `mKeyListener` is null, and initialize it if it is. This is necessary in both the constructor and inside of `setInputType`.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://github.com/facebook/react-native/wiki/Changelog
-->
https://github.com/facebook/react-native/wiki/Changelog
[Android] [Fixed] - NPE in `ReactEditText.setInputType` when React Native is used with some versions of a AppCompat 1.4.x. (and possibly others)
Pull Request resolved: https://github.com/facebook/react-native/pull/33920
Test Plan:
1. Build an app with both React Native and load it inside an app that is using AppCompat 1.4.x
2. Add a text field using React Native.
3. Open the screen of the app that contains the text field.
If you're working from this branch, you'll be fine. If you're working from main you'll get an NPE.
I can put together a test repo if needed.
Reviewed By: kacieb
Differential Revision: D36802622
Pulled By: cortinico
fbshipit-source-id: e7646da9a1ef0e0334152aecab0f972ca25092ec
Summary:
The sdks/.hermesversion file should be included inside the React Native NPM package.
While this file is available on the release branch, so it's effectively used during artifact preparation,
the file should also be included inside the react-native NPM package.
This commit addresses it.
Changelog:
[Internal] - Make sure sdks/.hermesversion is included inside the NPM package
Reviewed By: dmitryrykun
Differential Revision: D36785480
fbshipit-source-id: 1152de77818e92814b402a57ca5a05c235747eac
Summary:
Adding a flag to prepare for the phase 3 of the new architecture. This is still work in progress, not usable yet.
Changelog: [Internal]
Reviewed By: RSNara
Differential Revision: D36767843
fbshipit-source-id: 338d775681f2890461608b403749c3a7f05f84ff
Summary:
Followups to View Recycling diffs to improve things / clean up things a bit. This also fixes memory warnings which were not hooked up before.
Changelog: [Internal]
Reviewed By: mdvacca
Differential Revision: D36707792
fbshipit-source-id: 410e70bf0eeec5569566138af547e1601394d0e6
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33914
Add test coverage for the new hermesc functions in hermes-utils.js
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D36361612
fbshipit-source-id: 270f30917245c8d81baf136d9f58da4d77f7ad55
Summary:
Turns on the ssa env mode in Flow, which supercedes constrained writes with additional fixes
to refinement invalidation.
Reviewed By: SamChou19815
Differential Revision: D36717890
fbshipit-source-id: 20ee963591e8b454458273ac5aa86798e94a8bbd
Summary:
This diff upgrades xplat to 0.178.1 and pre-suppresses errors from turning on constrained writes.
To generate this diff I:
* Modified every `env_mode=constrain_writes` to `env_mode=ssa` and made a commit (this is so our upgrade script will work)
* Ran scripts/flow/upgrade.sh 0.178.1 to upgrade all the flowconfigs to 178.1 and suppress new-env errors
* Modified arvr/js/flowconfig.ejs to use 0.178.1 and ran `scripts/gen-flowconfig/gen-flowconfig --project arvr`
* Modified xplat/js/flowconfig.ejs to use 0.178.1 and ran `scripts/gen-flowconfig/gen-flowconfig --project xplat`
* Unstacked from the commit in point 1
Reviewed By: SamChou19815
Differential Revision: D36676019
fbshipit-source-id: c3032f18ed838afc327f00de563e7f20713bdc26
Summary:
`pod install --project-directory=ios` silently fails to prep Hermes:
```
% pod install --project-directory=ios
[Codegen] Generating ios/build/generated/ios/React-Codegen.podspec.json
[Hermes] Downloading Hermes source code for commit 515e112edc267ad58d3c70991b3d9a721cc66b19
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0
100 9478k 0 9478k 0 0 3939k 0 --:--:-- 0:00:02 --:--:-- 5591k
[Hermes] Expanding Hermes tarball for commit 515e112edc267ad58d3c70991b3d9a721cc66b19
[Hermes] Using pre-built HermesC
[!] One or more resources were not found and will not be included in the project. If they are found later and you want to include them, run `pod install` again.
warn Multiple Podfiles were found: ios/Podfile,macos/Podfile. Choosing ios/Podfile automatically. If you would like to select a different one, you can configure it via "project.ios.sourceDir". You can learn more about it here: https://github.com/react-native-community/cli/blob/master/docs/configuration.md
Auto-linking React Native module for target `ReactTestApp`: ReactTestApp-DevSupport
Analyzing dependencies
Fetching podspec for `DoubleConversion` from `../node_modules/react-native/third-party-podspecs/DoubleConversion.podspec`
[Codegen] Found FBReactNativeSpec
Fetching podspec for `RCT-Folly` from `../node_modules/react-native/third-party-podspecs/RCT-Folly.podspec`
Fetching podspec for `boost` from `../node_modules/react-native/third-party-podspecs/boost.podspec`
Fetching podspec for `glog` from `../node_modules/react-native/third-party-podspecs/glog.podspec`
[!] No podspec found for `hermes-engine` in `../node_modules/react-native/sdks/hermes/hermes-engine.podspec`
% ls -l node_modules/react-native/sdks
total 0
hermes-engine
hermesc
```
## Changelog
[iOS] [Fixed] - `pod install --project-directory=ios` fails when Hermes is enabled
Pull Request resolved: https://github.com/facebook/react-native/pull/33909
Test Plan: Instead of running `pod install` inside `ios` folder, run `pod install --project-directory=ios`.
Reviewed By: cortinico
Differential Revision: D36693625
Pulled By: hramos
fbshipit-source-id: 8757a9c388348276b07c785c211979ec8f2e2f84
Summary:
This target is not a good idea for a number of reasons:
1. It groups up multiple targets which breaks the dependency graph
2. It does not handle dependency remapping correctly
3. It has no mirror into fbcode
We should warn people this is a bad idea
Reviewed By: alexmalyshev
Differential Revision: D36519357
fbshipit-source-id: d60ca3237c7710118732578fecd1b2fc8903321b
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33897
Now the exception will display the class which caused the exception as well as helpful information as to why.
We've seen this happen a bunch due and have been very confused by the error message. It turns out that this processor runs before the classes listed are compiled. This means that if there's a compile error (or a missing import) the user will only see that this processor crashed, and not the compile error.
The additional information in the error is:
`java.lang.RuntimeException: Could not load classes set in ReactModuleList.nativeModules. Check that they exist and are imported correctly on class: com.meta.x.y.ReactPackage`
In this case, `com.meta.x.y.ReactPackage` is the class which needs to be fixed. Before, the error message made no mention of this class or the annotation.
Changelog: [Internal] This will change the way the annotation processor crashes. It will throw a RuntimeException instead of a ClassCastException.
Reviewed By: javache
Differential Revision: D36606279
fbshipit-source-id: aedf9682286fba49e23716b7eda16b2dd3b13422
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/33915
We don't need to import shelljs as a dependency anymore,
plus we had a duplicated entry in the files array for package.json
Changelog:
[Internal] [Changed] - Remove shelljs dependency and duplicated scripts in files
Reviewed By: dmitryrykun
Differential Revision: D36698750
fbshipit-source-id: 94f449f2c3c5d73d0f9ffd29df6b26f5fd6ef129
Summary:
See previous diff for details on general approach and benchmarks.
Changelog: [Internal]
Reviewed By: mdvacca
Differential Revision: D36676887
fbshipit-source-id: b177dcd19f1ea687bf7d2d4f2f637d2924723340
Summary:
Followup task for S271544.
There were no tests covering RTL on FlatLists, which would have prevented the above SEV. Adding an RNTester example for this.
Changelog:
[Internal] - Added RNTester example FlatList with RTL
Reviewed By: lunaleaps, javache
Differential Revision: D36601583
fbshipit-source-id: 054ac4eee876a514152f83ecc0522c62337cfea5
Summary:
When a JavaScript module is not registered as callable, the bridge will throw an error. Let's clean up the error message so that it's more helpful.
Now, it logs the number of registered callable JavaScript modules, as well as the list of callable JavaScript modules.
Changelog: [Internal]
Reviewed By: sshic
Differential Revision: D36646936
fbshipit-source-id: 62d091cda35e296ef9e569b444cd5b6f09b8bc54
Summary:
Changelog: [Internal][Changed] - Make the same optimization on enter/leave/move pointer events being dispatched by a mouse input.
If any ancestor view is listening to enter/leave events (just capture) then we dispatch the enter/leave event.
Reviewed By: vincentriemer
Differential Revision: D36601638
fbshipit-source-id: d6b5c32ae50bcf000100bcb878ca2ca89bd5c02e
Summary:
Upgrade the version of Xcode used in Sandcastle to Xcode 13.3.1, and the version used on Circle CI to Xcode 13.3.1.
Added a reminder to the Circle CI config to ensure we keep these versions in sync in future updates.
Circle CI software manifest: https://circle-macos-docs.s3.amazonaws.com/image-manifest/v7555/index.html
Changelog: [Internal]
Reviewed By: makovkastar
Differential Revision: D36671468
fbshipit-source-id: 8c028915d38738c92b5f759186b0fb95a9274211
Summary:
In AnimatedComponent.render, we get the initial values of any styles backed by AnimatedNode, but ONLY for non-native animations. Thus, if convert an animated node to native before the call to render to mount the component, we will end up not applying the initial value until after the component is mounted. This may result in a visible flicker as the expected value is applied some time after the component is mounted and visible.
- Without native driver: BaseViewManager.setTransform called during view preallocation (ViewManager.createViewInstance)
- With native driver: BaseViewManager.setTransform called during MountingManager.updateProps (called from PropsAnimatedNode.updateView)
This diff removes the isNative check in AnimatedStyle and AnimatedProps when traversing style props. This shouldn't be a problem:
- Created as non-native, animated with JS driver
- This diff does not affect this scenario
- Created as non-native, animated with native driver
- Initial value is applied correctly on render/mount. On subsequent renders, the outdated value from JS side will not apply on the platform view as it has not changed.
- Created as native, animated with native driver
- Initial value is applied correctly on render/mount. On subsequent renders, the outdated value from JS side will not apply on the platform view as it has not changed.
Changelog:
[Internal][Fixed] - Set correct initial value for AnimatedComponent for styles backed by native animated nodes
Reviewed By: JoshuaGross, javache
Differential Revision: D36612758
fbshipit-source-id: 922d6534c605b3eb0a1d9476753111b726f138f2
Summary:
RN-Tester is currently crashing at startup time due to an NPE.
This PR fixes it.
## Changelog
[Android] [Fixed] - Fix NPE on `ReactEditText` due to null mFabricViewStateManager
Pull Request resolved: https://github.com/facebook/react-native/pull/33910
Test Plan: Tested locally that RN Tester runs both in Debug and in Release
Reviewed By: cipolleschi
Differential Revision: D36666440
Pulled By: cortinico
fbshipit-source-id: f004ff228fb4f9ff339aac606858d47de3706426
Summary:
There are two methods in ReactRootView to handle touch events "onInterceptTouchEvent" and "onTouchEvent", for Venice we have ReactSurfaceView inherits ReactRootView but the implementation for above 2 touch handling methods still calls into it's super implementation which uses the bridge.
In this diff we make ReactSurfaceView inherits ReactRootView's "dispatchJSTouchEvent" and "dispatchJSPointerEvent". So that Venice has separate implementation for touch events handling.
Changelog:
[Android][Changed] - Revamp touch event dispatching methods
Reviewed By: fkgozali, JoshuaGross
Differential Revision: D36629466
fbshipit-source-id: fb7c5950afe6249d22edd3fac3fa5d3b83b3af84
Summary:
This op code was incorrectly configured to take two args, while it only takes one.
Changelog: [Internal]
Differential Revision: D36664590
fbshipit-source-id: 6e1fdb9f64bbd32fbe05bbd174f94ae57292bcf9
Summary:
This diff cleans up several Android Makefiles which we're not using anymore
as they've been replaced by CMake files.
There are still 3 Makefiles left, which I'm aiming to remove in the near future.
Changelog:
[Internal] [Changed] - Remove unused Makefiles from React Native core
Reviewed By: javache
Differential Revision: D36660902
fbshipit-source-id: 8afffac74d493616b0f9414567821cd69f4ef803
Summary:
There's no need for this to be a setter/getter, as there are no side-effects, and it means we can use the same helper method to read it as other feature flags.
Changelog: [Internal]
Reviewed By: cortinico
Differential Revision: D36595602
fbshipit-source-id: d27c01bd39d03606a6f8c17ba181a8cd0bf58dbb
Summary:
Enables two new experiments (and the current behaviour as default) to speed up access to TurboModule methods from JS.
1) HostObject - Current behaviour
2) Prototype - Connect the TM HostObject via `__proto__`, and cache any methods accessed on the wrapper object.
3) Eager - Eagerly store all methods on the wrapper object, do not expose the HostObject to JS at all (TurboModules no longer need to be HostObjects in this scenario)
Changelog: [Internal]
Reviewed By: JoshuaGross, rubennorte, mdvacca
Differential Revision: D36590018
fbshipit-source-id: c9565eb239eb6aeee0f06b581ff8cd72a92073fc
Summary:
* Make constructor private, all access is through install()
* Use nullability of longLivedObjectCollection_ instead of separate bool disableGlobalLongLivedObjectCollection_
Changelog: [Internal]
Reviewed By: RSNara
Differential Revision: D36592492
fbshipit-source-id: d65e779e1ac9fbe121937c5a20763aefcd589795
Summary:
In D36345402 (https://github.com/facebook/react-native/commit/56e9aa369f5c13af38cf80ba47e9eb29d835ec89) I changed the behaviour for mount items to be skipped if they were just setting zero values. AndroidTextInput is the only component that I'm aware of that has non-zero padding by default, and we account for this when creating the native shadow node. This optimization broken TextInput use-cases that explicitly request zero-padding, since we end up ignoring it.
To keep this optimization, explicitly init ReactTextInput's padding to 0, but only in Fabric. `updateState` was the closest thing I could find to a Fabric-only callback, once it's fully rolled out, we can also move this to the constructor.
Changelog: [Internal]
Reviewed By: JoshuaGross
Differential Revision: D36545775
fbshipit-source-id: 07bb96032c69d7e350980b0b975e637b66c307ed
Summary:
Update `hermes-engine.podspec` to use pre-built Hermes artifacts from the corresponding React Native GitHub Release when targeting a specific React Native release.
Otherwise, fallback to building Hermes from source.
Changelog: [Internal]
Reviewed By: cortinico, cipolleschi
Differential Revision: D36609257
fbshipit-source-id: 6179c9e255558c7eaf1417ff46a2e7db120295f0
Summary:
Due to invalid utf-8 data NSString may return nil during initialisation and this leads to crash when nil goes to NSDictionary or NSArray
Changelog: [Internal]
Differential Revision: D36623628
fbshipit-source-id: 23efad87cc9fe4a23c5e90dad63e0d74036b62fa
Joshua Gross
Shubho Sadhu
Jake Holland
Nicola Corti
Evan Yeung
Daniel Espino García
Antoine Doubovetzky
Jérémy Barbet
vincent-paing
Andy Zolyak
Moti Zilberman
Kevin Gozali
LeoTM
fortmarek
Héctor Ramos
Jordan Brown
Tommy Nguyen
Michael Lee (Engineering)
Daniel Abramowitz
Eric Hartzog
Genki Kondo
Ramanpreet Nara
David Vacca
Luna Wei
Lulu Wu
Riccardo Cipolleschi
Pieter De Baets
Franco Meloni
Rubén Norte
Vlad Hnatiuk
