Summary:
I've realized that the gradle plugin is currently looking at `.js` files for task re-execution. This means that, while the *.ts would still be considered when the codegen is invoked, an edit on one of those file, won't retrigger the codegen on Android.
This change fixes it so that we consider both `*.ts` and `*.js` files.
## Changelog
[Android] [Fixed] - Make sure *.ts files are considered for task avoidance in the Gradle Plugin
Pull Request resolved: https://github.com/facebook/react-native/pull/34296
Test Plan: Tests are attached.
Reviewed By: cipolleschi
Differential Revision: D38246125
Pulled By: cortinico
fbshipit-source-id: 80efcc9ef747c598ca040d65b25d270593c8aed2
Summary:
VirtualizedList state is represented in terms of [first, last] ranges, where it is possible to express a zero-cell range by having [n, n-1]. This includes some awkward examples, like [0, -1] being valid.
CellRenderMask assumes `addCells` is called with at least one cell, with VirtualizedList previously guarding against adding the no cell case. This guard is present for adding main cell regions, but not for `_initialRenderRegion()`, which can be overridden to have a zero length as well.
This moves the `CellRenderMask` to be permissive of zero-length cell regions, and removes the caller guard.
Changelog:
[Internal][Fixed] - Allow empty cell ranges in CellRenderMask
Reviewed By: rshest
Differential Revision: D38184444
fbshipit-source-id: d2dadfdd9628b24f894126d63b1eb93f6387b877
Summary:
Fixes https://github.com/facebook/react-native/issues/34120
The new React Native architecture doesn't check `needsCustomLayoutForChildren` so it wrongly positions native views on Android. In https://github.com/facebook/react-native/issues/34120 there are videos comparing the positioning of a native action view in the old and the new architecture.
This PR passes the parent tag to the `updateLayout` method of the `SurfaceMountingManager`. The `SurfaceMountingManager` calls `needsCustomLayoutForChildren` on the parent view manager (copied the code from the `NativeViewHierarchyManager` in the old architecture).
**NOTE** - I wasn't sure where to get the parent shadow view from so I've put in my best guesses where I could and left it as `{}` otherwise.
## Changelog
[Android] [Fixed] - Migrate `needsCustomLayoutForChildren` check to the new architecture
Pull Request resolved: https://github.com/facebook/react-native/pull/34254
Test Plan:
I checked the fix in the repro from https://github.com/facebook/react-native/issues/34165. Here is a video of the action view closing using the native button that is now visible in the new architecture.
https://user-images.githubusercontent.com/1761227/180607896-35bf477f-4552-4b8a-8e09-9e8c49122c0c.mov
Reviewed By: cipolleschi
Differential Revision: D38153924
Pulled By: javache
fbshipit-source-id: e2c77fa70d725a33ce73fe4a615f6d884312580c
Summary:
The current `test-manual-e2e.sh` script is broken on Android + Hermes + New App Template.
This commit fixes it. Specifically:
- There is no need to generate Maven Artifacts for RN Tester, as RN Tester consumes them from source.
- There is instead a need to generate Maven Artifacts for New App Template, as they need to be included inside the NPM package.
- The `:ReactAndroid:hermes-engine:installArchives` task needs to invoked to also generate the Hermes-engine .aar for bundling.
## Changelog
[Internal] - Fix test-manual-e2e on Android/Hermes for New App Template
Pull Request resolved: https://github.com/facebook/react-native/pull/34294
Test Plan: I've tested this against the `0.70-stable` branch and I was able to run an App from the New Template properly.
Reviewed By: dmitryrykun
Differential Revision: D38239238
Pulled By: cortinico
fbshipit-source-id: b3d95bad21515b12a91e29147e70ba8323896660
Summary:
Existing code may pass negative values for `initialScrollIndex`, which the previous implemnentation handled gracefully. Handle the case gracefully in VirtualizedList_EXPERIMENTAL as well.
Changelog:
[Internal][Fixed] - Gracefully handle negative initialScrollIndex in VirtualizedList_EXPERIMENTAL
Reviewed By: rshest
Differential Revision: D38183625
fbshipit-source-id: 9aea91c4cf3ce2190d769f34ce728a109efc88d4
Summary:
If FlatList is passed non-array data it will return `undefined` as the result of `getItemCount()` to VirtualizedList. This change makes it return `0` instead, to signify there are no valid items to attempt to accces.
There are a set of invariants on properties passed to FlatList, to curb incorrect types at runtime, but there is existing code which runs into the condition.
Changelog:
[Internal][Fixed] - Guard FlatList getItemCount() against non-array data
Reviewed By: javache
Differential Revision: D38198351
fbshipit-source-id: 9efd0df7eeeba17078e2c838d470c4b0d621b9a0
Summary:
VirtualizedList_EXPERIMENTAL has a different flow signature from VirtualizedList. This does not cause an error in unit tests, which are untyped, but prevents injecting VirtualizedList_EXPERIMENTAL via the current API from flowtype. This diff updates the injection interface to allow a more relaxed type, validating safety at runtime.
Changelog:
[Internal][Changed] - Allow injecting looser VirtualizedList
Reviewed By: javache
Differential Revision: D38143682
fbshipit-source-id: 054bbc5092823f4e4413d69d3d72a7ecdd71a6a2
Summary:
Implements most of systrace using androidx.tracing, this makes it usable using Android Studio profiler systrace.
## Changelog
[Android] [Added] - Improve OSS systrace
Pull Request resolved: https://github.com/facebook/react-native/pull/34252
Test Plan:
Run a systrace in Android Studio for RN Tester and make sure RN specific sections are there.
<img width="1263" alt="image" src="https://user-images.githubusercontent.com/2677334/180593493-fc087b4a-2253-43e1-b246-bed3e7bba7ac.png">
Reviewed By: NickGerleman
Differential Revision: D38116890
Pulled By: dmitryrykun
fbshipit-source-id: 744bedbf9ad4004488340a5b4e93d936d9a1e582
Summary:
Flow and TypeScript are two very similar programming languages. They are both recognizable as inputs for turbo module codegen. It is reasonable to require equivalent Flow and TypeScript turbo module definition generates the same output.
I add some test cases to ensure this. Glad that no functional inconsistency is found here.
## Changelog
[General] [Changed] - codegen: ensure equivalent Flow and TypeScript TM definition generates the same output
Pull Request resolved: https://github.com/facebook/react-native/pull/34251
Test Plan: `yarn jest` passed in `packages/react-native-codegen`
Reviewed By: dmitryrykun
Differential Revision: D38116756
Pulled By: cipolleschi
fbshipit-source-id: 476adbd171a35813923f2020c826d21b1fc2eeed
Summary:
cherry-pick changes from https://github.com/facebook/react-native/issues/34214 to main. because the `react_native_pods.rb` on main is quite different from 0.69, i have separated pr for the change.
## Changelog
[iOS] [Fixed] - Fix React-bridging headers import not found
Pull Request resolved: https://github.com/facebook/react-native/pull/34271
Test Plan: RNTester + pod install and verify pod targets to have `React-bridging` in header search paths.
Reviewed By: cipolleschi
Differential Revision: D38122074
Pulled By: dmitryrykun
fbshipit-source-id: 64569abbfa3a684f0d6b84c9e3222bfc9a171061
Summary:
I've noticed that `BackHandler.removeEventListener()` performs two same `indexOf()` calls on an array that is not changing. By removing extra `indexOf` we can slightly improve time complexity of `BackHandler.removeEventListener()` from O(2n) to O(n)
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[Android] [Fixed] - Remove extra indexOf call in BackHandler.removeEventListener
Pull Request resolved: https://github.com/facebook/react-native/pull/34281
Test Plan:
1. Add the following code to any function component
```javascript
BackHandler.addEventListener('hardwareBackPress', () => true).remove();
```
2. Press on hardware back button
Expected result: Application closes
Reviewed By: dmitryrykun
Differential Revision: D38198510
Pulled By: javache
fbshipit-source-id: eab6a57689a536623138a4b3ebddbf9ba87d281f
Summary:
At the moment we are having a bug that whenever we hit a "Done" button in an input field, a value is not submitted back, so the users cannot complete their work. Here is an example:
https://pxl.cl/28xFq
To fix it, we call a textInputShouldSubmitOnReturn method on Accessory button click.
Changelog
[IOS][Fixed] - Fix keyboard accessory button not triggering onSubmitEditing
Reviewed By: dmitryrykun
Differential Revision: D38152974
fbshipit-source-id: fc026dffd641966cd3d342d95a56c43c71008036
Summary:
The removeEventListener method has been removed from AccessibilityInfo and Linking in this PR: https://github.com/facebook/react-native/issues/33580, so I believe we can remove it from the mocks.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[Internal] [Removed] - Remove removeEventListener from AccessibilityInfo and Linking mocks
Pull Request resolved: https://github.com/facebook/react-native/pull/34260
Test Plan: I executed jest and everything is still green.
Reviewed By: dmitryrykun
Differential Revision: D38198653
Pulled By: GijsWeterings
fbshipit-source-id: 72d10ca54cd505d7c76e7531c9df718b1a9b9ed1
Summary:
# This Change
https://github.com/react-native-community/discussions-and-proposals/pull/335 discussed a set of problems with VirtualizedList and focus. These were seen as severe externally for a11y on desktop. The issues center on users of keyboard and accessibility tools, where users expect to be able to move focus in an uninterrupted loop.
The design and implementation evolved to be a bit more general, and without any API-surface behavior changes. It was implemented and rolled out externally as a series of changes. The remaining changes that were not upstreamed into RN are rolled into https://github.com/facebook/react-native/pull/32646
This diff brings this change into the repo, as a separate copy of VirtualizedList, to measure its impact to guardrail metrics, without yet making it the default implementation. The intention is for this to be temporary, until there is confidence the implementation is correct.
## List Implementation (more on GitHub)
This change makes it possible to synchronously move native focus to arbitrary items in a VirtualizedList. This is implemented by switching component state to a sparse bitset. This was previously implemented and upstreamed as `CellRenderMask`.
A usage of this is added, to keep the last focused item rendered. This allows the focus loop to remain unbroken, when scrolling away, or tab loops which leave/re-enter the list.
VirtualizedList tracks the last focused cell through the capture phase of `onFocus`. It will keep the cell, and a viewport above and below the last focused cell rendered, to allow movement to it without blanking (without using too much memory).
## Experimentation Implementation
A mechanism is added to gate the change via VirtualizedListInjection, mirroring the approach taken for Switch with D27381306 (https://github.com/facebook/react-native/commit/683b825b327e7fa71e87e3d613cb9acd37c24288).
It allows VirtualizedList to delegate to a global override. It has a slight penalty to needing to import both modules, but means code which imports VirtualizedList directly is affected the changes.
Changelog:
[Internal][Added] - Add VirtualizedList_EXPERIMENTAL (CellRenderMask Usage)
Reviewed By: lunaleaps
Differential Revision: D38020408
fbshipit-source-id: ad0aaa6791f3f4455e3068502a2841f3ffb40b41
Summary:
This change is in preparation of adding a separate `VirtualizedList_EXPERIMENTAL` component.
Both the original, and experimental lists use `VirtualizedListContext`, which itself references back to the VirtualizedList class type. VirtualizedList private methods are currently included in the type system, and are called in other VirtualizedList code (see https://github.com/facebook/react-native/commit/b2f871a6fa9c92dd0712055384b9eca6d828e37d). This prevents Flow from seeing the two classes are compatible if "private" methods change.
My first attempt was to parameterize the context, to allow both `VirtualizedList`, and `VirtualizedList_EXPERIMENTAL` to use the same code without sacrificing type safety or adding further duplication. This added more complexity than it is worth, so I am instead loosening the type on VirtualizedListContext to pass around a more generic handle.
Changelog:
[Internal][Changed] - Allow VirtualizedListContext to support different component type
Reviewed By: javache
Differential Revision: D38017086
fbshipit-source-id: 91e8f6ab2591d3ae9b7f9263711b4a39b78f68e0
Summary:
Bumping main to latest version of CLI available; in particular, we want to ensure to propagate this fix https://github.com/react-native-community/cli/pull/1655.
After merging, we'll cherry-pick in the `0.70-stable` branch.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[General] [Changed] - bump CLI to v9.0.0-alpha.5
Pull Request resolved: https://github.com/facebook/react-native/pull/34275
Test Plan: CI
Reviewed By: NickGerleman
Differential Revision: D38151383
Pulled By: dmitryrykun
fbshipit-source-id: afdafb496a159ad766dd322a4aab4bda6146edf0
Summary:
The `NSMutableAttributedString` was initialized with the same backing store as the cached attributed string on the shadow queue (background thread), but then passed to the main thread and ultimately the JS thread.
This explicitly copies the mutable attributed string into an immutable one on the shadow thread before passed off to other threads, which hopefully will address the `convertIdToFollyDynamic` crash that always includes `RCTBaseTextInputShadowView` touching an attributed string on the shadow queue in the trace.
Changelog:
[iOS][Fixed] - Possible fix for convertIdToFollyDynamic crash in RCTBaseTextInputView and RCTEventDispatcher
Reviewed By: sammy-SC
Differential Revision: D38133150
fbshipit-source-id: f371da5d17a32c3341287cd3e9730b31a98495f9
Summary:
Changelog: [iOS][Internal] - Add isPrimary property implementation to the PointerEvent object
This diff adds the `isPrimary` property to the PointerEvent object iOS implementation. In addition this adds a related change where we "reserve" the 0 touch identifier for mouse events and the 1 identifier for apple pencil events. This is an easy way to ensure that these pointers are always consistent no matter what happens. Since mouse & pencil pointers should always be considered the primary pointer, that allows us to focus the more advanced primary pointer differentiation purely on touch events.
The logic for this touch event primary pointer differentiation is essentially setting the first touch it recieves as a primary pointer, setting it on touch registration, and sets all subsequent touchs (while the first touch is down) as not the primary pointers. When that primary pointer is lifted, the class property keeping track of the primary pointer is reset and then the **next** pointer (secondary pointers which had already started before the previous primary pointer was lifted are not "upgraded" to primary) is marked as primary. A new platform test is also included in this diff in order to verify the aforementioned behavior.
Reviewed By: lunaleaps
Differential Revision: D37961707
fbshipit-source-id: ae8b78c5bfea6902fb73094fca1552e4e648ea44
Summary:
In `0.70-stable` we started seeing `yarn run lint` fail; turns out, this is happening because the latest release of `react-native/eslint-plugin-specs` 0.0.4 (from https://github.com/facebook/react-native/commit/ea8d8e2f49ea3ce15faeab500b661a1cacacf8a8) is faulty: the underlying reason is that since 0.0.2 there was this local workaround in place https://github.com/facebook/react-native/commit/2d06e6a4c9261bb7790cf217b66145415301bc54 (that changes a flag from false to true) but in NPM 8 the lifecycle hook `prepublish` has [been deprecated](https://docs.npmjs.com/cli/v8/using-npm/scripts#life-cycle-scripts) so this pre publishing script to change the flag was not run (you can easily verify by checking the node_module and see `react-native-modules.js` the flag on L17 is set to false).
This PR addresses it by moving to the new lifecycle hook `prepack` (and modifies the other files accordingly).
After this PR is merged, we'll need to cherry pick it into 0.70 and do both a new release from the 0.70 branch and one from the main branch to have new versions of this module with the flag set correctly.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[General] [Fixed] - Fix eslint-plugin-specs prepack npm lifecycle hook now that we use npm 8
Pull Request resolved: https://github.com/facebook/react-native/pull/34273
Test Plan: Go in the `eslint-plugin-specs` folder, run `npm pack`, unzip the generated tar file, go into `package`, verify that L17 of `react-native-modules.js` is correctly changed to `const PACKAGE_USAGE = true;` (instead of false - which is what happen without this fix).
Reviewed By: GijsWeterings
Differential Revision: D38151433
Pulled By: dmitryrykun
fbshipit-source-id: 7c4f13dae70eb731d57cbafa90f7be05c9bb8576
Summary:
Add capture-phase focus events to the type system, for use in the refactored VirtualizedList https://github.com/facebook/react-native/pull/32646/files
Tracking the last focused child is done via focus events. Focus events are bubbling (vs direct events like onLayout), and are given both a "capture" phase, and "bubbling phase", like DOM events on the web. https://stackoverflow.com/questions/4616694/what-is-event-bubbling-and-capturing
The VirtualizedList change wants to know if a child will receive focus. This is not possible to reliably capture in the bubbling phase, since a child may stop propagation.
See https://github.com/react-native-community/discussions-and-proposals/pull/335#discussion_r584851337 for some discussion with Scott Kyle about this issue back in the day
This is done by convention in React by adding a "capture" variant of the `onXXX` method. For all platforms I've seen with focus events, these map the `topFocus` native event to `onFocus` for bubbling phase, and `onFocusCapture` for capture phase. See https://reactjs.org/docs/events.html#supported-events
Changelog:
[General][Added] - Add types for onFocusCapture/onBlurCapture
Reviewed By: javache
Differential Revision: D38013861
fbshipit-source-id: 7bda22e1a4d5e36ac5e34e804abf6fb318a41baf
Summary:
We are already returning the `remove` function, and the `removeEventListener` method has been removed in this PR: https://github.com/facebook/react-native/issues/33580. So I believe this comment is now outdated and no longer relevant.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[Internal] [Removed] - Remove outdated TODO in AppState
Pull Request resolved: https://github.com/facebook/react-native/pull/34264
Reviewed By: dmitryrykun
Differential Revision: D38120344
Pulled By: GijsWeterings
fbshipit-source-id: c6c4d42f5b631a67bb6ad49c14df1a772f1cc71b
Summary:
A crash encountered in react-native-macOS is very similar to one fixed by https://github.com/microsoft/react-native-macos/pull/489#discussion_r451789471 (see discussion), and it's possible this `replacement` string also suffers from sharing the same backing store as the attributed string (maybe only when the range encompasses the entire string?) and therefore should be copied as well.
Changelog:
[iOS][Fixed] - Possible fix for convertIdToFollyDynamic crash in RCTBaseTextInputView and RCTEventDispatcher
Reviewed By: sammy-SC
Differential Revision: D38064551
fbshipit-source-id: 9c15f2a980155ab3cbb3fde79fcb93b24ee2091a
Summary:
The purpose of this diff is to un-deprecate the method DisplayMetrics.getWindowDisplayMetrics().
As React Native is used in devices with multiple screens and displays, we need to provide a non-deprecated way to access display metrics (e.g. screen density) for the window that's used to render react native views.
This diff doesn't make any change in behavior, but it un-deprecates the API
changelog: [Android][Added] Un-deprecate DisplayMetrics.getWindowDisplayMetrics() method
Reviewed By: javache
Differential Revision: D37871954
fbshipit-source-id: d8eb97cfae096f2f62ed1389a6de17a892a46b43
Summary:
This method assumes a semicolon existed before the closing bracket (`>`), but only does on iOS. This instead puts the content before the closing bracket, which is always there on both platforms.
Changelog:
[macOS][Fixed] - Fix exception thrown by [RCTTextView description] on macOS
Reviewed By: sammy-SC
Differential Revision: D38074642
fbshipit-source-id: f46d15c2bf2d966d1c1430568f083e4d501d4b40
Summary:
Adds changelog for new patch.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[Internal] [Changed] - add changelog entry for 0.69.3
Pull Request resolved: https://github.com/facebook/react-native/pull/34270
Test Plan: N/A
Reviewed By: lunaleaps
Differential Revision: D38116297
Pulled By: dmitryrykun
fbshipit-source-id: 962e34b4cfd58947cf6d95ae05affda7b9271716
Summary:
In D37873933, we update the buildifier binaries, which apply some minor changes to existing lint. Specifically, entries are now properly sorted.
Update the files in xplat such that updating buildifier does not cause lint changes in users diffs.
drop-conflicts
allow_many_files
#nocancel
#forcetdhashing
(Note: this ignores all push blocking failures!)
Reviewed By: d16r
Differential Revision: D37873936
fbshipit-source-id: f31d9c50d64ae99f99298977b471bf13e7ed5262
Summary:
Currently, with the Alert API on iOS, the only way to bold one of the buttons is by setting the style to "cancel". This has the side-effect of moving it to the left. The underlying UIKit API has a way of setting a "preferred" button, which does not have this negative side-effect, so this PR wires this up.
See preferredAction on UIAlertController https://developer.apple.com/documentation/uikit/uialertcontroller/
Docs PR: https://github.com/facebook/react-native-website/pull/2839
## Changelog
[iOS] [Added] - Support setting an Alert button as "preferred", to emphasize it without needing to set it as a "cancel" button.
Pull Request resolved: https://github.com/facebook/react-native/pull/32538
Test Plan:
I ran the RNTesterPods app and added an example. It has a button styled with "preferred" and another with "cancel", to demonstrate that the "preferred" button takes emphasis over the "cancel" button.
Luna:
* Also tested this on Catalyst
{F754959632}
Reviewed By: sammy-SC
Differential Revision: D34357811
Pulled By: lunaleaps
fbshipit-source-id: 3d860702c49cb219f950904ae0b9fabef03b5588
Summary:
When building `igios-no-extensions` with buck2 on linux, we the following analysis error:
```
When running analysis for `fbsource//fbobjc/Apps/Instagram/Instagram:InstagramNoExtensions (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
Caused by:
0: When running analysis for `fbsource//fbobjc/Apps/Instagram/Instagram:InstagramARVRFrameworkDylibFramework (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
1: When running analysis for `fbsource//fbobjc/Apps/Instagram/Instagram:InstagramDependencyPoolForLinkGroupsFramework (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
2: When running analysis for `fbsource//fbobjc/Libraries/FBReactKit:RCTNetInfo (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
3: When running analysis for `fbsource//xplat/js/RKJSModules/Libraries/NetInfo:FBReactNativeNetInfoSpecApple (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
4: When running analysis for `fbsource//xplat/js/react-native-github:RCTTypeSafety (fbsource//tools/build_defs/config/platform/apple:Instagram-125ae32099392ce8)`
5: Traceback (most recent call last):
* fbcode/buck2/prelude/apple/apple_library.bzl:48, in <module>
constructor_params, swift_providers = apple_library_rule_constructor_params_a...
* fbcode/buck2/prelude/apple/apple_library.bzl:111, in apple_library_rule_constructor_params_and_swift_providers
args = [cmd_args(get_framework_search_path_flags(ctx))],
* fbcode/buck2/prelude/apple/apple_frameworks.bzl:46, in get_framework_search_path_flags
return _get_framework_search_path_flags(_get_non_sdk_framework_directories(ct...
* fbcode/buck2/prelude/apple/apple_frameworks.bzl:59, in _get_non_sdk_framework_directories
return dedupe(filter(None, [_non_sdk_framework_directory(ctx, x) for x in fra...
* fbcode/buck2/prelude/apple/apple_frameworks.bzl:95, in _non_sdk_framework_directory
expanded_framework_path = _expand_sdk_framework_path(ctx, framework_path)
* fbcode/buck2/prelude/apple/apple_frameworks.bzl:81, in _expand_sdk_framework_path
expanded_path = expanded_path.replace(path_variable, path_value)
error: Type of parameter `new` doesn't match, expected `str`, actual `artifact`
--> fbcode/buck2/prelude/apple/apple_frameworks.bzl:81:25
|
81 | expanded_path = expanded_path.replace(path_variable, path_value)
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
```
This is related to running on Linux with buck2 and how we get the framework paths for linux/pika toolchains for non SDK frameworks. However, Foundation is an SDK framework, and updating it to be as such allows us to build, both with buck1 and buck2.
Differential Revision: D38080376
fbshipit-source-id: b226c3511e3312d0bf0ddb792dbe4ca99768cad1
Summary:
How the Hermes cache worked had a concurrency issue. It used to work by asking the Hermes repository for the latest commit and using that commit as cache key to try and retrieve a built version of Hermes to avoid to compile it more than once.
The problem happened when the `build_hermes_macos` was building Hermes using a commit A.
While building, another PR could be merged into `hermes:main`, creating commit B.
When this happened, the tasks `test_ios` and `test_ios_rntester`would try to retrieve a cached version of Hermes using commit B. That version did not exist because Hermes was created using commit A, and the job would either fail or trying to build Hermes from source, ending up taking a lot of time.
This PR attaches the `.hermes-cache-key-file` to the workspace and restores it in the other jobs, making sure that the same cache key is used in all three jobs.
## Changelog
<!-- Help reviewers and the release process by writing your own changelog entry. For an example, see:
https://reactnative.dev/contributing/changelogs-in-pull-requests
-->
[General] [Changed] - Attach the `.hermes-cache-key-file` to the workspace to avoid race conditions for new PR landing on Hermes and changing the head commit between the time Hermes is built and the time it has to be consumed.
Pull Request resolved: https://github.com/facebook/react-native/pull/34235
Test Plan:
Tested manually, looking at the messages in CI.
**build_hermes_macos**
<img width="881" alt="Screenshot 2022-07-21 at 15 40 02" src="https://user-images.githubusercontent.com/11162307/180241834-776f2291-63bb-4bb2-8837-14434b50fe61.png">
**test_ios_rntester**: notice that the `echo` is not executed, meaning that the `if` does not evaluate to true and, therefore, the file has been correctly attached.
<img width="956" alt="Screenshot 2022-07-21 at 15 40 52" src="https://user-images.githubusercontent.com/11162307/180242004-d9db0336-18d3-4321-a997-b538baa6beee.png">
**test_ios**: notice that the `echo` is not executed, meaning that the `if` does not evaluate to true and, therefore, the file has been correctly attached.
<img width="900" alt="Screenshot 2022-07-21 at 15 46 33" src="https://user-images.githubusercontent.com/11162307/180243359-79de5c7a-d2f0-4331-90c6-5bd2c0b5e1ac.png">
Reviewed By: cortinico
Differential Revision: D38037386
Pulled By: cipolleschi
fbshipit-source-id: 4db4f7c478e1afb2e4a18ba3d3f70896ed41d235
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/34232
This diff simplify the logic to decide whether we want to build hermes from source or not.
The requirement we have is that we don't want our users to build hermes.
So, we don't want to build hermes when there is a precompiled tarball available, while we want to build hermes in CI.
## Changelog
[General][Changed] - Build hermes when in CI and not when there is a tarball
Reviewed By: cortinico
Differential Revision: D37999748
fbshipit-source-id: 7d9cab51c37fb47db216055c057a22081e10df07
Summary:
Minor improvements to NativeState implementation and test. Does not change the API.
Changelog:
[Internal][Changed] - Minor improvements to jsi::NativeState
Reviewed By: neildhar
Differential Revision: D37961515
fbshipit-source-id: b87dcf2458ace883c784f3127a13f5a6c296d4a2
Summary:
Turbo module codegen for component parse `readonly T[]` and `readonly Object[][]` incorrectly. In this change, it is fixed, with necessary test cases added.
## Changelog
[General] [Added] - Support TypeScript array types for turbo module (component only)
Pull Request resolved: https://github.com/facebook/react-native/pull/34216
Test Plan: `yarn jest` passed in `packages/react-native-codegen`
Reviewed By: sammy-SC
Differential Revision: D37953004
Pulled By: cipolleschi
fbshipit-source-id: b71ea35eb9c4bd5dc41130f1dcf9201704ec298e
Summary:
changelog: [internal]
This was used as a kill switch, it has been in place for over 4 months, let's get rid of it.
Reviewed By: mdvacca
Differential Revision: D37921377
fbshipit-source-id: 594586033694766e13d0f2ab2bcd73b28ba180a9
Marshall Roch
Nicola Corti
Nick Gerleman
Matt Blagden
Graham Mendick
Janic Duplessis
Zihan Chen (MSFT)
Dmitry Rykun
Kudo Chien
fabriziobertoglio1987
Vitalii T
Oleh Malanchuk
Antoine Doubovetzky
Lorenzo Sciandra
Christoph Purrer
Vincent Riemer
Gijs Weterings
David Vacca
Harold Pratt
Danilo Bürger
Chatura Atapattu
Michael Sokolnicki
Robbie Coomber
Tony Du
Riccardo Cipolleschi
LeoTM
Sam Zhou
Sim Sun
Pieter Vanderwerff
Daniel Andersson
Dark Knight
Samuel Susla
Eric Edouard
