[pull] main from microsoft:main #31

pull · 2024-08-25T01:01:25Z

See Commits and Changes for more details.

Can you help keep this open source service alive? 💖 Please sponsor : )

Summary: Pull Request resolved: facebook#46072 Introducing extension fuctions in PixelUtils to replace Java static methods so Kotlin code can be more readable. Changelog: [Internal] Reviewed By: NickGerleman Differential Revision: D61372704 fbshipit-source-id: 489a74d60cd1463f0b1fad6f7c33925cec4000d2

…wing you to see both sides (facebook#46117) Summary: Pull Request resolved: facebook#46117 Currently in large cursive blocks like layout effect we can't tell what the slow leaf function is. With this fixed I'm able to root cause more complex issues in layout effects. Reviewed By: NickGerleman Differential Revision: D61486415 fbshipit-source-id: 5a4043b35eedcabcbea86953aac2173f66d7257b

Summary: Pull Request resolved: facebook#46118 Changelog: [Internal] Reviewed By: cyan33 Differential Revision: D61550740 fbshipit-source-id: 8c201b8f79416dec13ece9c671405af7b2de9c8c

Summary: Pull Request resolved: facebook#46076 This will add the shadows to iOS as well. let's see if anyone notices 🙂. I also removed dead styles, and removed some of the extra (excessive) padding specific to Android where the previous shadows would overlap. Changelog: [Internal] Reviewed By: rshest Differential Revision: D61421903 fbshipit-source-id: 887fa5aa96e3b0b4f81114ee814897c218db2b76

Summary: Pull Request resolved: facebook#46080 1. Force the examples to be alphabetized, where the hand-maintained list has some examples that are not 2. Remove reundant/not useful UI Changelog: [Internal] Reviewed By: cortinico Differential Revision: D61430910 fbshipit-source-id: 1f3e116fe81502faa7a72f2720912e26c9f04bb2

…in TextLayoutManager (facebook#46104) Summary: Pull Request resolved: facebook#46104 We want to use `PrecomputedText` to store glyph-level measurements on underlying Android Spannable. This means we need to consistently reuse the same Spannable, instead of recreating them on measurement. We have an opaque cache ID used by Android, for spannables originating from uncontrolled TextInput on UI-thread side. We also have `AttributedStringBox`, for a kind of similar purpose on iOS, which allows passing opaque pointer to the `TextLayoutManager`. This is only used for the `measure` function. This change makes us consistently use `AttributedStringBox` at the TextLayoutManager boundary, to let us migrate calls across TextLayoutManager to all pass opaque handle to underlying Spannable we will store, instead of passing the AttributedString each time. For now, every place previously passing an AttributedString value still passes one. There were also some egregious cases of accepting very large structures by value, causing unneeded copies. I changed the APIs to accept anything potentially larger than two pointers to pass by reference instead. This change is technically breaking, to any 3p code calling into TextLayoutManager (IIRC live-markdown exposed prefabs for this, but should be able to adapt fairly easily). Changelog: [General][Breaking] - Always use AttributedStringBox instead of AttributedString in TextLayoutManager Reviewed By: joevilches Differential Revision: D61484999 fbshipit-source-id: 07c5600cd917f2dab3d24559a25f27e0872ebddc

Summary: Pull Request resolved: facebook#46116 If your app raises an early js exception, and you cold start it, often you'll see this error: ``` SurfaceRegistryBinding::startSurface failed. Global was not installed. ``` {F1807125099} The reason why is because two different threads race to redbox: * The nativemodule thread: the early js error (reported [here](https://fburl.com/code/vcrqzsdp)) * The javascript thread: the SurfaceRegistryBinding error (a subsequent native -> js call) After this diff, the early js error will **not jump onto the nativemodule thread** to report this error. This ensures that we "always" (to the best of my knowledge) see the early js error first. Changelog: [Internal] Reviewed By: mdvacca Differential Revision: D61339213 fbshipit-source-id: f1b9ab30150b87377817c2fd93ca349c406db48b

…compared to native viewconfig (facebook#45859) Summary: Pull Request resolved: facebook#45859 "Fabric without SVCs" configuration is nearly gone, and so it doesn't make sense to need to add no-op methods, on normally Paper only code, etc to satisfy native viewconfig. These particular warnings are then more often noise, than things we need to action on. Checking for native code to be present can also break development where users are using distributed native app, slightly older than JS. This keeps the warning, only if static viewconfigs are missing native view config attributes (i.e. new prop would only be exposed to Paper, instead of only exposed to Fabric) Changelog: [Internal] Reviewed By: javache Differential Revision: D60575253 fbshipit-source-id: 1c118274b92eb7922c0dd92df060b24e44fceb3d

Summary: Pull Request resolved: facebook#46119 Straightforward migration of the `Appearance` module to use ESM named exports. Changelog: [Internal] Reviewed By: TheSavior Differential Revision: D61567882 fbshipit-source-id: a4a5a9771de001bd41cb0e23ba066f150267a55d

Summary: Pull Request resolved: facebook#46121 Updates `Appearance` on Android to supply the native module to `NativeEventEmitter` so that the native listener count can be managed like it is on iOS. This was previously required by macOS and iOS. Android and Windows also already implement: ``` interface NativeModule { addListener(eventType: string): void; removeListeners(count: number): void; } ``` So we should start passing `NativeAppearance` into the `NativeEventEmitter` constructor across all platforms. Changelog: [Internal] Reviewed By: TheSavior Differential Revision: D61567883 fbshipit-source-id: 1b3b76de9be3f35cacba1acbc43f6dcc0b41fde5

Summary: Pull Request resolved: facebook#46120 Currently, the implementation of `Appearance` duplicates the validation logic of string `colorScheme` values multiple times. This leads to more complicated code and also unnecessary work in certain edge cases (e.g. when `NativeAppearance` is not registered). This refactors `Appearance` to be simpler and to do less work. I've also configured `NativeAppearance.setColorScheme` to be non-nullable because it has existed since 2023. Changelog: [Internal] Reviewed By: TheSavior Differential Revision: D61567881 fbshipit-source-id: 61cb51709dc716ad97ae1397105414e74fe57a28

Summary: Pull Request resolved: facebook#46122 Implements a JavaScript cache for `colorScheme` in the `Appearance` module, so that we avoid potentially expensive and unnecessary native property accesses. Changelog: [General][Changed] - Improved `Appearance.getColorScheme` performance Reviewed By: rickhanlonii Differential Revision: D61567880 fbshipit-source-id: ca316946d68114b05239daa17105c85e637efe07

…6113) Summary: Pull Request resolved: facebook#46113 changelog: [internal] This showed promised in local tracing but that failed to translate to real perf improvement. Unshipping. Reviewed By: christophpurrer Differential Revision: D61537744 fbshipit-source-id: 03a2a69a6fed32a6b493bc17372e3783b9db2d1e

…cebook#46132) Summary: Pull Request resolved: facebook#46132 Temporaily disable the `nativeSourceCodeFetching` capability — which reverts this to the legacy handling in the Inspector Proxy. This is because we've noticed performance issues when loading large bundle source maps, particularly on Android, with a nontrivial path to optimising this ([raising the frontend `IO.read` size](facebookexperimental/rn-chrome-devtools-frontend#97) further is leading to WebSocket disconnections on Android 😐). Changelog: [Internal] Reviewed By: robhogan Differential Revision: D61543480 fbshipit-source-id: ee66b4cebd40f8cc6466270c5875df744d2b588a

Summary: Pull Request resolved: facebook#46131 React native 0.75.0, 0.75.1 and 0.75.2 has been published to NPM without the latest tag, despite the tag being on the commit. When debugging why that's happened, I realized that we were not downloading the tags when checking out the repo. This change fixes that. {F1816667285} ## Changelog: [Internal] - Publish React native as latest when the latest tag is specified on git Reviewed By: cortinico Differential Revision: D61593398 fbshipit-source-id: 96bf8346207f0bd0b01f60ee09879210d12d30af

Summary: In the recent 0.75 release I've noticed new `CONFIG_CMD` option in `react-native-xcode.sh`. But this option was not used. Insted when set `CONFIG_APP` was used. This seems like a bug. As the usage before this PR would be as follow: ```bash export CONFIG_CMD=true export CONFIG_APP="/path/to/node /path/to/node_modules/react-native/cli.js config" ``` After this PR ``` export CONFIG_CMD="/path/to/node /path/to/node_modules/react-native/cli.js config" ``` This PR also removed unused explicite `--config-cmd "$CONFIG"` flag, as this is always overwriten by the code above, by default to `--config-cmd" "$NODE_BINARY $NODE_ARGS $REACT_NATIVE_DIR/cli.js config`. ## Changelog:  [iOS][Fixed] - Use CONFIG_CMD if set Pull Request resolved: facebook#46112 Test Plan: I've set `CONFIG_CMD` and run Xcode Release build to check that the set command is executed. Reviewed By: christophpurrer Differential Revision: D61545010 Pulled By: blakef fbshipit-source-id: ebbf8ebc08404bc6816277518a3b86c6f7e41e6e

Summary: Pull Request resolved: facebook#46136 Changelog: [iOS] - Update RNTester and HelloWorld template app to `min iOS 15.1` Reviewed By: philIip, blakef Differential Revision: D61604036 fbshipit-source-id: 9260b9b51a9c976792fd9d705b9b6a35a88a7f86

Summary: Pull Request resolved: facebook#46137 Changelog: [iOS] - Bump OSS Build infra to min iOS 15.1 Reference: react-native-community/discussions-and-proposals#812 Reviewed By: christophpurrer Differential Revision: D61577939 fbshipit-source-id: 8f2929af266f61aedd55358b13d08544fd79e434

Summary: Pull Request resolved: facebook#46135 Changelog: [iOS] - Bump iOS min version for CI Reviewed By: cipolleschi Differential Revision: D61604034 fbshipit-source-id: 5d2b0001b6437e62c64d821e36a4183096e8ef74

…6125) Summary: Clean up some dead code after some refactoring of RCTParagraphComponentView. cc cipolleschi ## Changelog: [IOS] [FIXED] - Clean up RCTParagraphComponentView & RCTParagraphTextView Pull Request resolved: facebook#46125 Test Plan: CI green. Reviewed By: christophpurrer Differential Revision: D61603193 Pulled By: cipolleschi fbshipit-source-id: a357e8c5355707b2296462de513010acda4ee6ea

…irtualized list (facebook#46103) Summary: Pull Request resolved: facebook#46103 Reducing the boundary of rerender of virtual lists. Previously with prop: "strictMode={true}" the VirtualizedList still re rendered each CellRenderer component. Because method getDerivedStateFromProps generated every time a new uniq state and the cells didn’t have a PureComponent. It helps to improve react performance for lists which have 5+ elements. I reused recomended approach from react doc https://legacy.reactjs.org/blog/2018/06/07/you-probably-dont-need-derived-state.html#what-about-memoization changelog: [internal] Optimizing CellRenderer of VirtualizedList Reviewed By: NickGerleman, sammy-SC Differential Revision: D61493434 fbshipit-source-id: 917a33e48bd2f18e8ac150e5701d2e7c45dbe879

Summary: Pull Request resolved: facebook#46140 Add default value for newArchitectureEnabled param changelog: [internal] internal Reviewed By: shwanton Differential Revision: D61621224 fbshipit-source-id: 728c91f9ce10ad5872f86cd223ea30c8f9bde413

Summary: Pull Request resolved: facebook#46139 Potential fix for IndexOutOfBounds crashes in clipping rect logic. Changelog: [Internal] Reviewed By: mdvacca Differential Revision: D61615009 fbshipit-source-id: 1208e11286ad2a84c22262462832a66396a1b35b

Summary: Pull Request resolved: facebook#46148 Try fixing the import path (not sure if access from external modules is allowed though). Changelog: [Internal] Reviewed By: realsoelynn Differential Revision: D61643115 fbshipit-source-id: a840ee375430b0e20050a36b695a5a4c93c29258

Summary: Pull Request resolved: facebook#46146 Changelog: [Internal] `unit` is of type `UnitType`, so there's no reason to have a default case here. i found this because my build failed when pulling in this dependency, there was a compiler flag that enforced that all cases must be enumerated. this seems like the right practice anyways. Reviewed By: NickGerleman Differential Revision: D61635463 fbshipit-source-id: b84b5518f2a17e792309f85ae91514a17abad295

…46172) Summary: In facebook#45102 I've implemented a baseline alignment function for the new architecture. I've noticed one thing I've missed previously - `locationForGlyphAtIndex` is [relative to the line fragment](https://developer.apple.com/documentation/appkit/nslayoutmanager/1403239-locationforglyphatindex), not the container. This means that the attachments would be put in the wrong place in multiline text. This PR fixes that by adding the position of the entire line to the attachment position. ## Changelog: [IOS] [FIXED] - Fixed baseline attachment position in multiline text Pull Request resolved: facebook#46172 Test Plan: Checked on relevant example of RNTester. |Before|After| |-|-| |<img width="546" alt="Screenshot 2024-08-22 at 15 53 14" src="https://github.com/user-attachments/assets/c1861655-9253-44fd-9f2f-796aff83df1e">|<img width="546" alt="Screenshot 2024-08-22 at 15 52 37" src="https://github.com/user-attachments/assets/27f07bc7-a8f6-4696-a414-815e7ece642c">| Reviewed By: andrewdacenko Differential Revision: D61662006 Pulled By: cipolleschi fbshipit-source-id: 5eafdae1800c06d9fc61bfac99584e6e25a05c24

… layer (facebook#46138) Summary: Pull Request resolved: facebook#46138 jorge-cab noticed that filters on iOS do not fit the shape of the layer if we have rounded corners. Fix is pretty straight forward. Changelog: [Internal] Reviewed By: NickGerleman Differential Revision: D61612655 fbshipit-source-id: 91785ed10a039e031c5440bde131c1583ba3992a

…46157) Summary: Pull Request resolved: facebook#46157 Changelog: [internal] This improves the handling of disconnected nodes in `MutationObserver`. Specifically: * When observing a node, if the node is disconnected (unmounted) this is just a no-op (without logging errors). We can't observe an unmounted node. * When disconnecting the observer, if the observed nodes are disconnected, we get the target shadow node from an internal map, which we always have access to if we successfully started observing the node. If this logs an error now, it's something to look into but it won't generally log it if the target is just disconnected. That will work correctly. Reviewed By: bvanderhoof Differential Revision: D61655856 fbshipit-source-id: d18a885350ef000fc563c85f6775ba864d184ad1

…ook#46156) Summary: Pull Request resolved: facebook#46156 Changelog: [internal] This improves the handling of disconnected nodes in `IntersectionObserver`. Specifically: * When observing a node, if the node is disconnected (unmounted) this is just a no-op (without logging errors). We can't observe an unmounted node. * When disconnecting the observer, if the observed nodes are disconnected, we get the target shadow node from an internal map, which we always have access to if we successfully started observing the node. If this logs an error now, it's something to look into but it won't generally log it if the target is just disconnected. That will work correctly. Reviewed By: bvanderhoof Differential Revision: D61656597 fbshipit-source-id: 6a39c878acc976ddc0789260106da104a3f2a57f

Summary: Pull Request resolved: facebook#45966 JSI performs the check itself, no need to do it here. Plus, bytecode bundles must not be zero terminated. ## Changelog: [IOS] [FIXED] - Fixes NSDataBigString length calculation Reviewed By: realsoelynn Differential Revision: D61058869 fbshipit-source-id: 15b99ef13f9aebd11ff410d02c21db8e46cc6ac3

Bumps [cross-spawn](https://github.com/moxystudio/node-cross-spawn) from 7.0.3 to 7.0.6. - [Changelog](https://github.com/moxystudio/node-cross-spawn/blob/master/CHANGELOG.md) - [Commits](moxystudio/node-cross-spawn@v7.0.3...v7.0.6) --- updated-dependencies: - dependency-name: cross-spawn dependency-type: indirect ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Summary: Fix AppKit exception throws when focusing text inputs by calling becomeFirstResponder directly on the backing text input view. Making a view first responder has to happen through the window using makeFirstResponder. Test Plan: - Run Zeratul with Fabric - Focus the search text input above the message threads - Click inside the active message thread to trigger the auto-focus of the composer - The composer gets focus without AppKit throwing an exception. https://pxl.cl/3dVMx Reviewers: shawndempsey, #rn-desktop Reviewed By: shawndempsey Differential Revision: https://phabricator.intern.facebook.com/D48696690

…stance Summary: Fix AppKit exception throws when blurring text inputs by calling `resignFirstResponder` directly on the backing text input view. Resigning the first responder state has to happen through the window by calling `[window makeFirstResponder:nil]` which will: - call `resignFirstResponder` on the current first responder - if successful, the window will become the first responder Test Plan: - Run Zeratul with Fabric - Focus the search text input above the message threads - Click inside the active message thread to trigger the auto-focus of the composer and the blur on the search field - The focused field resigns the first responder status without throwing an exception https://pxl.cl/3GvZD Reviewers: shawndempsey, #rn-desktop Reviewed By: shawndempsey Differential Revision: https://phabricator.intern.facebook.com/D50700782

…ocus / blur

Summary: On macOS, Core Animation layers have their anchor point set to (0,0) which is the top-left. On iOS the anchor point is set to (0.5, 0.5) which matches the center. Transforms assigned to views are built based on having the anchor point at the center. This diff updates the transform matrix to apply to transform from the center. Test Plan: Check zoom/rotation animations in Zeratul. | Before | After | |--| | https://pxl.cl/3G8HG | https://pxl.cl/3G8HQ | Reviewers: shawndempsey, #rn-desktop Reviewed By: shawndempsey Differential Revision: https://phabricator.intern.facebook.com/D50628807 Tasks: T161413049

...ages/debugger-frontend/dist/third-party/front_end/third_party/csp_evaluator/csp_evaluator.js

+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+function T(e){return e=(e=e.replace(/^\w[+\w.-]*:\/\//i,"")).replace(/^\/\//,"")}function w(e){const s=new URL("https://"+T(e).replace(":*","").replace("*","wildcard_placeholder")).hostname.replace("wildcard_placeholder","*"),t=/^\[[\d:]+\]/;return T(e).match(t)&&!s.match(t)?"["+s+"]":s}function N(e){return e.startsWith("//")?e.replace("//","https://"):e}function v(e,s){const t=new URL(N(e.replace(":*","").replace("*","wildcard_placeholder"))),o=s.map((e=>new URL(N(e)))),i=t.hostname.toLowerCase(),n=i.startsWith("wildcard_placeholder."),a=i.replace(/^\wildcard_placeholder/i,""),c=t.pathname,r="/"!==c;for(const e of o){const s=e.hostname;if(s.endsWith(a)&&(n||i===s)){if(r)if(c.endsWith("/")){if(!e.pathname.startsWith(c))continue}else if(e.pathname!==c)continue;return e}}return null}function A(e,s){const t=Object.keys(e.directives);for(const o of t){const t=e.directives[o];t&&s(o,t)}}


...ages/debugger-frontend/dist/third-party/front_end/third_party/csp_evaluator/csp_evaluator.js

+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+function T(e){return e=(e=e.replace(/^\w[+\w.-]*:\/\//i,"")).replace(/^\/\//,"")}function w(e){const s=new URL("https://"+T(e).replace(":*","").replace("*","wildcard_placeholder")).hostname.replace("wildcard_placeholder","*"),t=/^\[[\d:]+\]/;return T(e).match(t)&&!s.match(t)?"["+s+"]":s}function N(e){return e.startsWith("//")?e.replace("//","https://"):e}function v(e,s){const t=new URL(N(e.replace(":*","").replace("*","wildcard_placeholder"))),o=s.map((e=>new URL(N(e)))),i=t.hostname.toLowerCase(),n=i.startsWith("wildcard_placeholder."),a=i.replace(/^\wildcard_placeholder/i,""),c=t.pathname,r="/"!==c;for(const e of o){const s=e.hostname;if(s.endsWith(a)&&(n||i===s)){if(r)if(c.endsWith("/")){if(!e.pathname.startsWith(c))continue}else if(e.pathname!==c)continue;return e}}return null}function A(e,s){const t=Object.keys(e.directives);for(const o of t){const t=e.directives[o];t&&s(o,t)}}
