Summary:
The root cause comes down to the fact that historically `-DFOO=0` and `-DFOO=1` patterns have been used when it comes to preprocessor definitions and various engineers use `#ifdef FOO` and `#if FOO` interchangeably. That results in an incorrect pattern because the real evaluation is following:
- `FOO` not defined: `#if FOO` = `FALSE`, `#ifdef FOO` = `FALSE`.
- `-DFOO=0`: `#if FOO` = `FALSE`, `#ifdef FOO` = `TRUE`.
- `-DFOO=1`: `#if FOO` = `TRUE`, `#ifdef FOO` = `TRUE`.
Reviewed By: aditya7fb
Differential Revision: D14811733
fbshipit-source-id: a7fedc56ab0ce8eedcb48bda783739a4c1eecf81
Summary: This will simplify updating the JSI API from upstream in the future.
Reviewed By: mhorowitz
Differential Revision: D14762674
fbshipit-source-id: fa4a86f08425943e301da4ef3df9893ebaa1493e
Summary:
Main change is to the property diffing - we now use the last known props set on the view rather than the default props to compute the diff. This requires exposing a `getProps` method on all view components which should be fine I think.
I also realized that in more complex animations with multiple nodes, the node that the animation starts on might not be connected to a view, so we don't know if it's fabric just based on that, so we have to do a recursive search through the children to find if there are any that are associated with a fabric view to decide we should start the animation immediately. Unfortunately there can still be a timing gap here since the animated API is async and the uimanager API is sync - I'll need to change the animated API to be sync to completely fix this.
Reviewed By: shergin
Differential Revision: D14732028
fbshipit-source-id: 882c056b0b63aa576f8e42439be405cf7fb3147a
Summary: This moves the iOS WebView files to be fb internal which completes the removal of WebView from React Native open source as part of the Lean Core effort.
Reviewed By: TheSavior
Differential Revision: D14630076
fbshipit-source-id: 7bc11d6c1470bb748c823c86cbb8b5ee94b508ff
Summary:
ScrollView's scrollTo behavior on iOS was recently changed to limit the offset to the content size plus any content inset (see #23427). This departure from the old behavior created UI issues for anyone that is using the over-scroll ability for the purpose of positioning elements at specific coordinates on the screen. Examples include using this behavior to position TextInputs above the virtual keyboard programmatically when focused or moving drop down elements positioned near the bottom of the content toward the top of the screen when selected to show a larger absolutely positioned item list. Default behavior does not change and this is an "opt-in" type of prop to re-enable the old behavior.
[iOS] [Added] - Added scrollToOverflowEnabled prop to ScrollView
Pull Request resolved: https://github.com/facebook/react-native/pull/24296
Differential Revision: D14762619
Pulled By: cpojer
fbshipit-source-id: d2a552b5cb321d52e8ea4116327bf9ec647a3aae
Summary:
There is a timing issue when reloading the bridge (in dev mode) and the tear down of the TurboModules. This causes `Instance` to never get freed, hence the "bridge" isn't cleaning up properly. The side effect can be bogus error saying that it's unable to find a module.
To address this, JSCallInvoker should just take in weak_ptr<Instance>.
Reviewed By: RSNara
Differential Revision: D14739181
fbshipit-source-id: f9f2a55486debaeb28d3d293df3cf1d3f6b9a031
Summary: A retain-cycle regression was added by D14162776 and detected here: T41208740. Fix should be trivial.
Reviewed By: shergin
Differential Revision: D14709784
fbshipit-source-id: bad6ab31a5170e9320c4432cbd20d02ec6164f38
Summary:
It closes#24170.
I opened the issue yesterday, and I think I found the way how to resolve it. So, I'm opening this PR directly without any comment on the issue.
In the `endRefreshProgrammatically` of `RCTRefreshControl.m`, the comment says that " The contentOffset of the scrollview MUST be greater than 0". However, if the `contentInset` prop is set for the `FlatList`, I think `contentOffset` can be a negative value after refreshing and should be greater than `-contentInset.top`.
As I reported the bug in that issue, If I am using `contentInset` prop to the `FlatList`, making `contentOffset` value be always 0 when refreshing ends causes something wrong situation.
[iOS] [Fixed] - Fix setting the contnetOffset when refreshing ends
Pull Request resolved: https://github.com/facebook/react-native/pull/24191
Differential Revision: D14710987
Pulled By: sahrens
fbshipit-source-id: 03f06df9a93a2a46a7cc0b56269091778805917e
Summary:
Xcode 10.2 forbids creating categories for swift class that uses `+load` method. In react-native categories like this are used to register swift classes as modules (macro `RCT_EXTERN_MODULE`) This PR changes it to use `__attribute__((constructor))` instead of objc `+load` method.
I introduced new macro for this purpose, `RCT_EXPORT_MODULE_NO_LOAD`, it expands in something like:
```
void RCTRegisterModule(Class);
+ (NSString *)moduleName {
return @"jsNameFoo";
}
__attribute__((constructor)) static void initialize_ObjcClassFoo{
RCTRegisterModule([ObjcClassFoo class]);
}
```
Functions marked with `__attribute__((constructor))` are run before main and after all `+load` methods, so it seems like correct thing to do.
Fixes https://github.com/facebook/react-native/issues/24139
Doc about loading order https://developer.apple.com/documentation/objectivec/nsobject/1418815-load?language=objc
[iOS] [Fixed] - Fix runtime crash in xcode 10.2 when using RCT_EXTERN_MODULE for swift classes.
Pull Request resolved: https://github.com/facebook/react-native/pull/24155
Reviewed By: javache
Differential Revision: D14668235
Pulled By: shergin
fbshipit-source-id: 0c19e69ce2a68327387809773848d4ecd36d7461
Summary:
Otherwise reloading from metro deadlocks like this:
`[RCTSurfaceRegistry _stopAllSurfaces]` is calling `[RCTSurfaceRegistry enumerateWithBlock]` which locks `_mutex` and then we call `surfaceForRootTag` which then deadlocks on the same mutex:
https://pxl.cl/tmm1
Reviewed By: shergin
Differential Revision: D14679843
fbshipit-source-id: 9f4ecdfa7a79fcf7f3fc2ce437d4399b00910f26
Summary:
If size is zero, the `CGContext` would invalid, so we need to filter zero things.
cc. shergin
[iOS] [Fixed] - Fix invalid CGContext when invalidate layer
Pull Request resolved: https://github.com/facebook/react-native/pull/24195
Differential Revision: D14683396
Pulled By: shergin
fbshipit-source-id: b5b45fb86ca3158284281460cf1d95d1b22eab0d
Summary: When calling into JS (e.g. promise resolve/reject, callback) in TurboModule, we bypass the bridge's message queue. At times this causes race condition, where there are a bunch of pending UI operations (in RCTUImanager) waiting to be flushed, but nothing adds calls to the message queue. Usually tapping the screen will trigger the flush because we're sending down touch events to JS.
Reviewed By: JoshuaGross
Differential Revision: D14656466
fbshipit-source-id: cb3a174e97542bf80f0a37b4170b6a8e6780fa35
Summary:
This makes the RCTEvent protocol more generic to make it easier to use the event coalescing feature for type of events other than components. This does a few other improvements that will be useful in follow up PRs.
- Add `RCTComponentEvent` which is used instead of deprecated `[sendInputEventWithName:body:]` and remove that method completely (was only used at 2 places).
- Make `coalescingKey` optional for events that return NO from `canCoalesce`.
- Make `viewTag` optional for events that are not related to views.
- Fast path for events that return NO from `canCoalesce`.
- Add a missing test for event coalescing with different view tags.
Ended up making only one PR for all this since the changes are related and hard to separate.
**Migration**
Use a custom RCTEvent subclass with `[sendEvent:]` (preferred way to allow type safe events) or `RCTComponentEvent`.
**Test plan**
- Ran RCTEventDispatcher unit tests
- Tested manually in RNTester
Changelog:
[iOS] [Changed] - Remove deprecated RCTEvent method, sendInputEventWithName:body:
Pull Request resolved: https://github.com/facebook/react-native/pull/15894
Reviewed By: shergin
Differential Revision: D13726194
Pulled By: hramos
fbshipit-source-id: 11f63a99e08f46ec6b4f16f8d9949cdbf5c3fe13
Summary:
This PR is a follow-up to #21211 by request of hramos to incorporate some additional crash fixes / synchronization edge cases found in our production testing. What follows is largely a copy of the original PR description from #21211 - see that PR for original discussion thread as well as context on why this replacement PR was needed.
This PR is a minimalistic change to RCTTiming that causes it to switch exclusively to NSTimer (i.e., the 'sleep timer') in order to continue triggering timers when the app has moved to the background.
Many people have expressed a desire for background timer support on iOS. (See #1282, #167, and #16493). In our app — a podcast/audio player — we use background timers to ensure that we never lose track of the user's playback position, should the app crash or be terminated by the OS.
The RCTTimer module uses a RN-managed CADisplayLink if the next requested timer is less than a second away; otherwise, it switches to an NSTimer (which is refers to as a 'sleep timer' in source). The RN-managed CADisplayLink is always disabled when the app goes to the background (and thus cannot be used); however, the NSTimer will still issue its callbacks in the background.
This PR adds a flag to track whether the app is in the background, and if so, all timers are routed through NSTimer until the app returns to the foreground. vishnevskiy at Discord opened a similar PR (#16493) that implements a drop-in for CADisplayLink which falls back to NSTimer, but I decided to incorporate the background-NSTimer logic directly into RCTTimer, since NSTimer is already in use.
It's worth noting that the background NSTimer may not fire as often as requested — it may give the appearance of lagging depending upon your app's priority in the background. For our audio app, NSTimer fires exactly on schedule if there's an open AVAudioSession and audio is playing; if audio is not playing, it fires about half as often as requested, which is still adequate for networking polling and other tasks.
It's worth noting that background timers only function as long as an app is actually running in the background. Apple offers a variety of Background Modes (which can be toggled in the Capabilities section of the target inspector in Xcode), and the app will need to be legitimately using one of these modes in order for this change to provide any value — otherwise it will be terminated within a couple of seconds of moving to the background.
The good thing about this change is that for apps that do perform essential computation in support of their Background Mode, they can now use `setTimeout` and `setInterval` without problem — whereas in the past, neither would ever trigger their callback until the app returns to the foreground.
Pull Request resolved: https://github.com/facebook/react-native/pull/23674
Differential Revision: D14621326
Pulled By: shergin
fbshipit-source-id: c76e060ad2c662c140d7d2f4fb5aaa7094032515
Summary:
We will use it inside `core` module, so we have to decouple it from `view`.
As part of this, I added some comments, changed `const Float &` to just `Float` and put the implementation into `.cpp` file.
Reviewed By: mdvacca
Differential Revision: D14593952
fbshipit-source-id: 80f7746f4fc5b95febc8df9f5a9c0386a6425c88
Summary: This is based on the work done in D8686586. Removed the logger instance from JSIExecutor constructor and installed it into the runtimeInstaller at all call sites.
Reviewed By: mhorowitz
Differential Revision: D14444120
fbshipit-source-id: 0476fda4230c467573ea04102a12101bcdf36c53
Summary:
With our current infra, we support automatic conversion of method arguments using `RCTConvert`.
```
RCT_EXPORT_METHOD(foo:(RCTSound*) sound)
{
//...
}
```
```
interface RCTConvert (RCTSound)
+ (RCTSound *) RCTSound: (NSDictionary *) dict;
end
implementation RCTConvert (RCTSound)
+ (RCTSound *) RCTSound: (NSDictionary *) dict
{
//...
}
end
```
```
export interface Spec extends TurboModule {
+foo: (dict: Object) => void,
}
```
With this setup, when we call the foo method on the TurboModule in JS, we'd first convert `dict` from a JS Object to an `NSDictionary`. Then, because the `foo` method has an argument of type`RCTSound*`, and because `RCTConvert` has a method called `RCTSound`, before we invoke the `foo` NativeModule native method, we first convert the `NSDictionary` to `RCTSound` using `[RCTConvert RCTSound:obj]`. Essentially, if an argument type of a TurboModule method is neither a primitive type nor a struct (i.e: is an identifier), and it corresponds to a selector on `RCTConvert`, we call `[RCTConvert argumentType:obj]` to convert `obj` to the type `argumentType` before passing in `obj` as an argument to the NativeModule method call.
**Note:** I originally planned on using `NSMethodSignature` to get the argument types. Unfortunately, while the Objective C Runtime lets us know that the type is an identifier, it doesn't inform us which identifier it is. In other words, at runtime, we can't determine whether identifier represents `RCTSound *` or some other Objective C class. I figure this also the reason why the old code relies on the `RCT_EXPORT_METHOD` macros to implement this very same feature: https://git.io/fjJsC. It uses `NSMethodSignature` to switch on the argument type, and then uses the `RCTMethodInfo` struct to parse the argument type name, from which it constructs the RCTConvert selector.
One caveat of the current solution is that it won't work work unless we decorate our TurboModule methods with `RCT_EXPORT_METHOD`.
Reviewed By: fkgozali
Differential Revision: D14582661
fbshipit-source-id: 3c7dfb2059f031dba7495f12cbdf406b14f0b5b4
Summary:
The `RCTBridge` contains numerous definitions of notification names, which we can observe in order to get insights into the React Native performance.
The Android implementation is a little different, such that you can listen for any of the [following](https://github.com/facebook/react-native/blob/master/ReactAndroid/src/main/java/com/facebook/react/bridge/ReactMarkerConstants.java) marker constants, simply by including the following code:
```java
ReactMarker.addListener(new ReactMarker.MarkerListener() {
Override
public void logMarker(ReactMarkerConstants name, Nullable String tag, int instanceKey) {
Log.d("ReactNativeEvent", "name: "+ name.name() + " tag: " + tag);
}
});
```
This will allow you to perform the necessary processing, calculations as required.
---
This PR enables observing for the module setup event (`RCTDidSetupModuleNotification`) by including the respective module's name & setup time in milliseconds.
[iOS] [Added] - Gain insights on the module setup times by observing `RCTDidSetupModuleNotification`. The `userInfo` dictionary will contain the module name and setup time in milliseconds. These values can be extracted via `RCTDidSetupModuleNotificationModuleNameKey ` and `RCTDidSetupModuleNotificationSetupTimeKey`.
Pull Request resolved: https://github.com/facebook/react-native/pull/23859
Differential Revision: D14579066
Pulled By: PeteTheHeat
fbshipit-source-id: 52645127c3fc6aa5bd73e3bd471fccd79cb05c14
Summary:
Fix for this issue I rasied: https://github.com/facebook/react-native/issues/24024
When I toggle `Show Perf Monitor` and reload JS `CMD+R` the Perf Monitor will be hidden, but settings in dev menu will persist. So to fix this state and need to `Hide Perf Monitor` and `Show Perf Monitor` again to see it.
[iOS] [Fixed] - Show Perf Monitor, after reloading JS
Pull Request resolved: https://github.com/facebook/react-native/pull/24073
Differential Revision: D14560025
Pulled By: cpojer
fbshipit-source-id: cd5602bd6ee041b8b3e61d163d10bd8bc47237b9
Summary:
Change the function called when an error is returned from the completionHandler of the `RCTJavaScriptLoader(loadBundleAtURL:onProgress:onComplete:)`.
* AS-IS
- `RCTLogError()`
- This error method does not post any notification for users. So users cannot receive notification when `RCTJavaScriptLoader(loadBundleAtURL:onProgress:onComplete:)` complete with error.(i.e. bundleURL is is invalid but not nil, `http://1`, `http://localho:8081/index.bundle`)
* TO-BE
- `handleError(error:)`
- Call this method will post notification for users when `completionHandler` fails with error.
[iOS] [Fixed] - Change the function called when an error is returned from the completionHandler of the `RCTJavaScriptLoader(loadBundleAtURL:onProgress:onComplete:)`.
Pull Request resolved: https://github.com/facebook/react-native/pull/23837
Differential Revision: D14538790
Pulled By: cpojer
fbshipit-source-id: da3306904c0d8113d204edfaa0f9e2a23793981a
Summary:
Put `prepareForRecycle` to last before enqueue to recycle pool, ensure only call it when count lower than RCTComponentViewRegistryRecyclePoolMaxSize.
cc. shergin .
[General] [Changed] - Reorder prepareForRecycle before adding recycle pool
Pull Request resolved: https://github.com/facebook/react-native/pull/24025
Differential Revision: D14536843
Pulled By: shergin
fbshipit-source-id: 82a816e2c0afb5a6bb72637d7d55d6a4fda708af
Summary:
We need to remove adding png file extension when path has not extension. Two reasons:
1. `imageWithContentsOfFile` or other `UIKit` methods can load png image correctly, even if path has not `png` file extension.
2. Sometimes, people may have file that actually not have file extension, it's the designated behavior for user. Like #23844 .
CC. sahrens cpojer .
[iOS] [Fixed] - Remove explicitly add png file extension when load local image
Pull Request resolved: https://github.com/facebook/react-native/pull/23864
Reviewed By: shergin
Differential Revision: D14425373
Pulled By: hramos
fbshipit-source-id: 3cc06c9a3d68cadf652c1de742f3cce26258c874
Summary:
I encountered invalid bridge error when I develop (reload high frequency) some times, and if it happened, I only can restart the app, because we would always show error invalid message if we trigger reload command. So I recommend Change invalid bridge log level from error to warn, I think it's enough.
cc. cpojer .
<img width="375" alt="892C97F4-E910-4E3B-935A-65C899916E6E" src="https://user-images.githubusercontent.com/5061845/54538344-0d32c600-49cf-11e9-818c-0e4cb5a0a518.png">
[iOS] [Fixed] - Change invalid bridge log level from error to warn
Pull Request resolved: https://github.com/facebook/react-native/pull/24005
Differential Revision: D14504820
Pulled By: cpojer
fbshipit-source-id: f24c4876fdb3e64183c09c5ddc598a8c87e405a7
Summary:
This is the couple of hacks I used after I finished #23802 in order to get fabric working on RNTester. This is inspired from prior work by kmagiera.
The goal of this PR is to show others what I’m struggling with, and to eventually merge it sans hacks.
- Yarn Install
- Uncomment the commented out pods in RNTester's pod file
- Open RNTesterPods workspace
- Run App
- this is only for pods, the non-pod RNTester will no longer work until updated with fabric too.
- `SurfaceHostingView` & `SurfaceHostingProxyRootView` both try to start the surface immediately, this leads to a race condition due to the javascript not having loaded yet, the hack here is:
1. Swizzle the `start` method on `RCTFabricSurface` to no-op when called.
2. Add observer for `RCTJavaScriptDidLoadNotification`
3. Call private method `_startAllSurfaces` on `_surfacePresenter` in AppDelegate when we receive `RCTJavaScriptDidLoadNotification`.
[General] [Added] - Use Fabric in RNTester
Pull Request resolved: https://github.com/facebook/react-native/pull/23803
Reviewed By: shergin, mdvacca
Differential Revision: D14450726
Pulled By: fkgozali
fbshipit-source-id: 8ae2d48634fecb60db539aaf0a2c89ba1f572c27
Summary: Several things need to ironed out: 1) LocalState in Fabric C++, 2) setting dimensions of BottomSheet component to 0,0 for parent.
Reviewed By: shergin
Differential Revision: D14426167
fbshipit-source-id: 45a90a7971c87672872108a9e360926b4a6095f0
Summary:
From https://github.com/facebook/react-native/pull/23861#issue-260337314
> changing "interval" example from "time-only" to "datetime" because there's a known bug that prevented the previous example from working
We need to ensure set minuteInterval after set datePickerMode, otherwise minuteInterval invalid in time mode.
cc. grabbou cpojer .
[iOS] [Fixed] - Fixed minuteInterval invalid in time mode
Pull Request resolved: https://github.com/facebook/react-native/pull/23923
Differential Revision: D14477549
Pulled By: cpojer
fbshipit-source-id: 2c612d488b6d592b1907e150df5e07fe83132829
Summary:
As a follow-up to this other PR #23839, it adds support for other, iOS only, flags into `AccessibilityInfo`.
It adds these other 4 methods:
* `isBoldTextEnabled()`
* `isGrayscaleEnabled()`
* `isInvertColorsEnabled()`
* `isReduceTransparencyEnabled()`
P.S: Android implementation for those methods just return `false` (with `Promise.resolve(false)`)
And the corresponding event listeners:
* `boldTextChanged`
* `grayscaleChanged`,
* `invertColorsChanged`,
* `reduceTransparencyChanged`
Pull Request resolved: https://github.com/facebook/react-native/pull/23913
Differential Revision: D14482214
Pulled By: cpojer
fbshipit-source-id: b97725fd12706957d4dad880a97e6b0993738272
Summary:
D14323747 broke SSTs. This PR optimized implementation of nativeID. Instead of searching entire tree to find a view, it uses a map to cache views which have nativeID set. The map uses keys with format <nativeID-rootViewTag>. Finding the rootViewTag causes a race condition, due to [this syncing code](https://fburl.com/93vqzlbg). The SST runner attempts to read the map to find a [SST view](https://fburl.com/8zadz024) before that view is added to the map.
My fix is to change the key format to simply <nativeID>. Product code should not duplicate nativeID since it is used to uniquely identify react managed views from native. I'll comment on PR with this fix.
Reviewed By: shergin, mmmulani
Differential Revision: D14416262
fbshipit-source-id: 3b707f2ff4049ac83ac8861e3cda435224d973d8
Summary:
Part of: #23313.
This moves the `RCTTest` lib from `Libraries/RCTTest` to `RNTester/RCTTest`. This also removes `takeSnapshot` from React Native, and implements it as a standalone module in RNTester called `ScreenshotManager`.
[General] [Removed] - RCTTest & ReactNative.takeSnapshot
Pull Request resolved: https://github.com/facebook/react-native/pull/23721
Differential Revision: D14434796
Pulled By: PeteTheHeat
fbshipit-source-id: d6e103a0ea0b6702701cdb5ce8449163ca4628ce
Summary:
We need to add SurfacePresenterStub header to project header for tvOS.
Related 544d9fb10b
CC. sahrens .
[iOS] [Fixed] - Add SurfacePresenterStub header to project header for tvOS
Pull Request resolved: https://github.com/facebook/react-native/pull/23855
Differential Revision: D14436185
Pulled By: cpojer
fbshipit-source-id: b59369541e78967346fc9e9fa6a9685725834f39
Summary:
This fixes a few memory leaks in fabrics handling of colors for borders, when using CGColorRef's we must be diligent about releasing the memory back.
[iOS] [Fixed] - Border style memory leaks
Pull Request resolved: https://github.com/facebook/react-native/pull/23815
Differential Revision: D14431250
Pulled By: shergin
fbshipit-source-id: dc663c633ae24809cb4841800d31a6ac6eeb8aa5
Summary:
<!-- Explain the **motivation** for making this change. What existing problem does the pull request solve? -->
I have used RN for a long time, and for all this time, crash reporting has been less great than native development crash reporting. At some point, companies like sentry, bugsnag and a bunch of others started supporting sourcemaps for js crashes in RN, which helped a lot.
But native crashes were (and still are) much harder to diagnose.
..Until now :D
I have make a repo of a sample RN app, included this PR in it, and some code and screenshots to help.
The repo is [here](https://github.com/pvinis/react-native-project-with-crash-heaven-pr).
I was trying to get good crash reports from native crashes in iOS for a looong time. I spoke with people in sentry, in bugsnag and more, and I could not get this solved. There was no clear way to get the **native** crashed to display correctly.
I made two repos here, one for [sentry](https://github.com/pvinis/SentryBadStack) and one for [bugsnag](https://github.com/pvinis/BugsnagBadStack), demonstrating the correct js handling and the bad native handling.
After all this, and talks with their support, twitter etc, I investigated further, on **why** this was happening. I thought there must be some reason that native crashes look bad in all the tools, and in the same way. Maybe it's not their fault, or up to them to fix it, or maybe they didn't have the experience to fix it.
In a test project I created, I checked what's up with the `RCTFatalException`, and I found out that the React Native code is catching the `NSException`s that come from any native modules of a RN app and converting it into an string and sending it to `RCTFatal` that created an `NSError` out of that string. Then it checks if the app has set a fatal error handler and if not, goes ahead and throws that `NSError`.
The problem here is that `NSException` has a bunch more info that the resulting `NSError` is missing or is altering. Turning the callstack into a string renders crash reporting tools useless as they are missing the original place the exception was thrown, symbols, return addresses etc. In both repos above it can be seen that both tools were thinking that the error happened somewhere in the `RCTFatal` function, and it did, since we create it there, losing all the previous useful info of the original exception. That leaves us with just a very long name including a callstack, but very hard to actually map this to the code and dsym.
I added a fatal exception handler, that mirrors the fatal error handler, as the error handler is used around React Native internal code.
Then I stopped making a string out of the original `NSException` and calling `RCTFatal`, and I simply throw the exception. This way no info is lost!
Finally, I added some code examples of native and js crashes and added a part in the `RNTester` app, so people can see how a js and a native error look like while debugging, as well as try to compile the app in release mode and see how the crash report would look like if they connect it to bugsnag or sentry or their tool of choice.
I have attached some images at the bottom of this PR, and you can find some in the 3 repos I linked above.
[iOS] [Fixed] - Changed the way iOS native module exceptions get handled. Instead of making them into an `NSError` and lose the context and callstack, we keep them as `NSException`s and propagate them.
[General] [Added] - Example code for native crashes in iOS and Android, with buttons on RNTester, so developers can see how these look when debugging, as well as the crash reports in release mode.
Pull Request resolved: https://github.com/facebook/react-native/pull/23691
Reviewed By: fkgozali
Differential Revision: D14276366
Pulled By: cpojer
fbshipit-source-id: b308d5608e1432d7676447347ae77c0721094e62
Summary:
This change fixes https://github.com/facebook/react-native/issues/23645.
The issue is that the `YGMarker.cpp`, `YGValue.cpp`, `YGConfig.cpp` and `log.cpp` files are already included in the Yoga compilation unit, so including these files in React's compile sources too results in "duplicate symbols" errors when loading React Native with `-force_load` (which behaves slightly differently to `-ObjC` - according to a colleague, "ObjC will scan through each object file in each library and force linking of any object file that contains Objective C code, while force_load will link every object file in a particular staticlib (regardless of whether or not the object file contains Objective C code)").
These changes seemed to be introduced by a few commits:
- D13819111 -> 43601f1a17
- D13439602 -> b5c66a3fbe
- D14123390-> e8f95dc7a1
- D7530369 -> 95f625e151
Perhaps we need to check with the original authors/any C++ experts to confirm if this fix is correct - it compiles for me but I'm not sure what the original intention of these changes was.
[iOS] [Fixed] - Remove duplicated Yoga compile sources to prevent "duplicate symbols" errors when linking using -force_load
Pull Request resolved: https://github.com/facebook/react-native/pull/23823
Reviewed By: davidaurelio
Differential Revision: D14387657
Pulled By: hramos
fbshipit-source-id: d85221b6dc1a0377662624f4201b27222aed8219
Summary:
I measured the performance of UIMananger methods and found that `create` method spends a significant amount of time on preallocating views (around 10 ms on my device). Interestingly, this time was actually spent on dispatching operations on the main thread, not on performing those operations. That makes me think about think about the preallocation problem one more time.
(Here and later I discuss preemptive optimistic preallocation views on iOS, but the issues are more-less applicable for all platforms.)
BTW, what's view preallocation? – In React Native we believe that we can get significant TTI wins instantiating platform views ahead of time while JavaScript thread is busy doing reconciliation and the main one is idle.
So, seems the current approach has those downsides:
* We spend too much time on dispatching only (jumping threads, creation Objective-C blocks, managing them in GCD, incrementing atomic refcountes inside `ShadowView` and so on). That's wasteful.
* We don't have much information during preallocations that can help prepare something to save time in the future. We don't know the future size of the component, so we cannot e.g. pre-draw a border or even allocate memory for a future drawing.
* We pre-allocate to much. At the moment of component creation, we don't know will this component be filtered out by view-flattening infra or not, so we always allocate a view for it. That's ~30% more views that we actually need.
* We don't stop allocating (or dispatching instructions for that) after the recycle pool is already filled in. That's also wasteful.
Why is this so bad and how can we fix it properly?..
I thought about this problem for months. And I think the answer is trivial: We don't know the exact shape of the interface until we finish creating it, and there is no way to overcome this problem on the client side. In the ideal world, the server size (or hardcoded manifest somewhere) tells the renderer (at the moment where we started navigating to it) how many views of which type will some surface use.
And until our world is not so perfect, I think we can get a significant performance improvement doing simple preallocation of 100 regular views, 30 text views, and 30 image views during rendering the first React Native surface. So I hardcoded that.
Reviewed By: JoshuaGross
Differential Revision: D14333606
fbshipit-source-id: 96beeb58b546258de1b8fd58550e0ae412b78aa9
Summary:
Right now we rely on the Paper UIManager to update animated node graphs - this hooks us into `RCTSurfacePresenter` in the same way so we are no longer reliant on Paper. Should also help with complex ordering corner cases with pre vs. post operations and restoring defaults when nodes are removed. More info:
https://github.com/facebook/react-native/pull/11819/files
Note that we don't have a way to differentiate animation nodes related to fabric views vs. paper views, so if paper and fabric are both rendering updates simultaneously it's possible they could get processed by the wrong callback. That should be very rare, rarely cause problems even if it does happen, and won't be a problem at all in a post-Paper world.
Reviewed By: shergin
Differential Revision: D14336760
fbshipit-source-id: 1c6a72fa67d5fedbaefb21cd4d7e5d75484f4fae
Summary:
We currently rely on the Paper UIManager calling `uiManagerWillPerformMounting` to flush the animated operations queue, which includes starting and stopping animations. This mostly works right now because Fabric always starts after Paper, but sometimes Paper doesn't fire `uiManagerWillPerformMounting` for a while, which can delay an animation starting.
To fix this, I force a flush of the queues on the UIThread whenever start or stop is called. This should be safe because the order of animation operations is still preserved, and start/stop are (almost?) always called in dedicated event handler loops, so any other updates like changing the way nodes are attached should already have been processed from a previous JS execution loop.
Reviewed By: JoshuaGross
Differential Revision: D14313502
fbshipit-source-id: 2a2b0c614fd1a591bd04b6b3fafcc09ff6c9d6e7
Summary: Use the codegen for the Slider component with the new `inferfaceOnly` option
Reviewed By: TheSavior
Differential Revision: D14295981
fbshipit-source-id: 0482572892fbcffada43c7c6fbf17e70546300b8
