Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42897
Changelog: [Internal] - Update nightly flow to use set-version
This change fixes `set-version` to update the `packages/react-native` native source and build files (as `set-rn-version` does) -- this was an oversight but not an issue as `set-version` isn't actually used anywhere right now.
Reviewed By: huntie
Differential Revision: D53463414
fbshipit-source-id: d0d9e4bbe246cccb8643a6ebf9794122bc343433
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42906
Makes way to relocate repo E2E testing + Verdaccio logic under `scripts/e2e/`. The contents of this script are minimal and are better located with `rn-tester-e2e`.
Changelog: [Internal]
Reviewed By: cipolleschi
Differential Revision: D53516117
fbshipit-source-id: e7e50af0383788f2219da190bf921ea93a6455eb
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42783
This change will fix the publishing of Nightlies for React Native with the right version
## Changelog:
[Internal] - Update the package.json of react native correctly.
Reviewed By: cortinico, huntie
Differential Revision: D53309082
fbshipit-source-id: 2fa4d4fdf4f984603c6b3d3690fa3c464ee6d030
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42774
Reorganise release scripts so that command entry points are grouped based on execution context, which also reflects dependencies between scripts.
Also:
- Document the current behaviours of these scripts.
- Relocate utils out of the root contents.
- Replace `exec` call to `set-rn-version` script with function import.
NOTE: `yarn trigger-react-native-release` (documented command in release process) is unchanged, since this is aliased from `package.json`.
```
├── releases
│ ├── templates/
│ ├── utils/
│ ├── remove-new-arch-flags.js
│ ├── set-rn-version.js
│ └── update-template-package.js
├── releases-ci
│ ├── prepare-package-for-release.js
│ └── publish-npm.js
└── releases-local
└── trigger-react-native-release.js
```
Changelog: [Internal]
Reviewed By: cipolleschi
Differential Revision: D53274341
fbshipit-source-id: eec2befc43e7a47fd821b2e2bcc818ddffbb6cf7
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42776
Changelog: [Internal] set all monorepo packages (including react-native) to one version and update all inter-dependencies (including the template)
Reviewed By: huntie
Differential Revision: D53251917
fbshipit-source-id: 95330ca66dcb7234a3f09752ecc3ed9087ced4bf
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42740
Follow-up to a CircleCI breakage introduced by D53001971.
This was missed by both the typechecker (no Flow in file) and CI (no PR-time jobs covering this script).
Changelog: [Internal]
Reviewed By: cortinico, GijsWeterings, cipolleschi
Differential Revision: D53228096
fbshipit-source-id: fbbe1538ee52b8452399d86489239434d3a068be
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42730
Reviewing and modernising this script as part of simplifying our release publish workflow.
- Drop unused `--dependency-versions` arg from CLI entry point
- Simplify templating approach
- Type as Flow
- Drop dependencies on `shelljs` and `yargs`
- Relocate under `scripts/releases/`
- Rewrite tests as snapshot tests
Changelog: [Internal]
Reviewed By: NickGerleman
Differential Revision: D53001971
fbshipit-source-id: e55a71a0bb37e3e18ba1e582a5c46ddd58823d81
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42681
CI failures in Windows JS tests recently (https://github.com/facebook/react-native/pull/41463) were caused by the triggering of Babel registration during tests, due to an import of `packages/dev-middleware` (index), breaking subsequent transformation of other tests.
## Root cause
Example of a problematic import:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/dev-middleware/src/__tests__/ServerUtils.js#L15
..which triggers a Babel registration:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/dev-middleware/src/index.js#L16-L18
That registration behaves differently on Windows due to the `ignore: [/\/node_modules\/\]`, which doesn't match against Windows path separators - Babel matches against system separators.
In particular, this changed whether `node_modules/flow-parser` was transformed when loading the RN Babel transformer. Transforming this file causes a `console.warn` from Babel due to its size:
> [BABEL] Note: The code generator has deoptimised the styling of /Users/robhogan/workspace/react-native/node_modules/flow-parser/flow_parser.js as it exceeds the max of 500KB.
This throws due to our setup:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/react-native/jest/local-setup.js#L27
This all manifests as the first test following a Babel registration (within the same Jest worker) that requires the RN Babel transformer throwing during script transformation.
## This change
This is the minimally disruptive change that makes Babel registration behaviour consistent between Windows and other platforms. The more durable solution here would be *not* to rely on any Babel registration for Jest, which has its own `ScriptTransformer` mechanism for running code from source. Given the fragile way our internal+OSS Babel set up hangs together that's a higher-risk change, so I'll follow up separately.
Changelog: [Internal]
Reviewed By: huntie
Differential Revision: D53124578
fbshipit-source-id: 074a8e139e506a5dceec13f07d412599fb292d92
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42678
Changelog: [internal]
This is a re-application of https://github.com/facebook/react-native/pull/42430, which had to be reverted because of crashes in optimized builds on Android:
```
Abort Reason: terminating due to uncaught exception of type facebook::jni::JniException: java.lang.ClassNotFoundException: com.facebook.react.internal.featureflags.ReactNativeFeatureFlagsProvider
```
The root cause of that was that that class was removed because it wasn't statically referenced from Kotlin/Java, but it was dynamically referenced from C++ (in `ReactNativeFeatureFlagsProviderHolder.cpp`).
This applies the same changes + adds `DoNotStrip` annotations for the affected class and all its methods.
Reviewed By: huntie
Differential Revision: D53122992
fbshipit-source-id: efc4d5636a3f2d39b86e9c098bff408b6688b80b
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42430
This PR creates a new internal feature flags system for React Native. This is only meant to be used internally within the framework, but we might expose it externally in some form in the future to allow customizing specific feature flags in frameworks and applications.
Features:
* 2 types of flags:
* Common: can be overridden from native and are accessible from all layers of the stack (Objective-C/Swift, Java/Kotlin, C++ and JavaScript).
* JS-only: flags that can only be defined and accessed from JS (to allow things like hot reloading without a native build).
* 1 source of truth for each flag.
* Feature flags are application/process scoped (using C++ singletons).
See the `README.md` file in this PR for additional information.
This also adds modifies `run-ci-javascript-tests` to run a new check to make sure that the generate files are in sync with the JSON file that contains the definitions.
Changelog: [internal]
Reviewed By: huntie
Differential Revision: D52806730
fbshipit-source-id: 0ba95803f61ec2f05266ee535921321bf6d3dc6a
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42638
Enables these modules to be covered by `public-api-test`.
- Standardise as CommonJS modules, fixing compatibility with [`flow-api-translator`](https://www.npmjs.com/package/flow-api-translator).
- Use explicit object type in generated file template.
Changelog: [Internal]
Reviewed By: NickGerleman
Differential Revision: D52963967
fbshipit-source-id: c9f3e35f70859c1b99b7297228ee2498f91d9041
Summary:
Those scripts are all dead, and should not be used anymore.
I'm removing them.
## Changelog:
[INTERNAL] - Remove dead android scripts
Pull Request resolved: https://github.com/facebook/react-native/pull/42612
Test Plan: n/a
Reviewed By: cipolleschi
Differential Revision: D52997852
Pulled By: cortinico
fbshipit-source-id: cf57177eedb8bc0f40daf7c6c5fcd1d5ba89ba32
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42287
Changelog: [Internal]
`run-ci-e2e-tests.js` currently skips past package build errors and tries to keep going. At best, this fails somewhere downstream of a build error (without any clear diagnostics as to why). At worst, this can miss errors entirely.
Here we extend the script's existing error handling behaviour to cover errors during the build script. It's probably worth following up to make sure all unexpected failures bubble up and stop the script, as opposed to the current error-swallowing default.
Reviewed By: hoxyq
Differential Revision: D52785131
fbshipit-source-id: 08deedfdf5b3d3cb63e77c74b47eb75570a58fbb
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/42233
This diff removes the need for providing the `ios_folder` argument to `use_react_native`. We no longer do any manual path tranformations to get the iOS project root.
Instead we use `Pod::Config.instance.installation_root` which always points to the correct directory.
Changelog: [iOS][Breaking] - CocoaPods: remove the `ios_folder` argument from the `use_react_native` function.
Reviewed By: cipolleschi
Differential Revision: D52659429
fbshipit-source-id: 67c79cd9d74a0351ad2c242b74cbd48b6bd2dc94
Summary:
Currently our CI will auto-tag any `npm publish` as `latest` for the monorepo packages. This is because [we do not specify a tag](https://github.com/facebook/react-native/blob/main/scripts/monorepo/find-and-publish-all-bumped-packages.js#L104), so npm will [default to `latest`](https://docs.npmjs.com/cli/v10/commands/npm-dist-tag#description). We encountered a similar issue for `react-native` awhile ago and fixed that with [always specifying a tag](https://github.com/facebook/react-native/blob/main/scripts/npm-utils.js#L84), with the explicit opt-in for `latest`.
yarn and npm will resolve `*` dependencies using `latest`. This will be a problem for any React Native version that uses `*` deps. We have actively tried to remove these `*` versions but older patches may still contain them.
When we do a monorepo package bump, it may be for 0.71 and for a user who is initializing a 0.72 version project (that still has * deps), they will receive monorepo packages of version `0.71.x`, which is not compatible. (React Native monorepo packages do not faithfully follow semver)
This change allows us to specify what tags to use and suggest tags based on what branch you are on and asks for confirmation
```
> branch 0.73-stable
? Select suggested npm tags. (Press <space> to select, <a> to toggle all, <i> to invert selection)
❯◉ "0.73-stable"
◉ "latest"
? Confirm these tags for *ALL* packages being bumped: "0.73-stable","latest" (Y/n)
> branch 0.72-stable
? Select suggested npm tags. (Press <space> to select, <a> to toggle all, <i> to invert selection)
❯◉ "0.72-stable"
◯ "latest"
? Confirm these tags for *ALL* packages being bumped: "0.72-stable" (Y/n)
> branch main
? Select suggested npm tags. (Press <space> to select, <a> to toggle all, <i> to invert selection)
❯◉ "nightly"
? Confirm these tags for *ALL* packages being bumped: "nightly" (Y/n)
```
## Changelog:
[INTERNAL] [CHANGED] - Support dist-tags in publishing monorepo packages to avoid default "latest" tag.
Pull Request resolved: https://github.com/facebook/react-native/pull/42146
Test Plan: `yarn test scripts/`
Reviewed By: NickGerleman
Differential Revision: D52551769
Pulled By: lunaleaps
fbshipit-source-id: 52f923464387cffdc6ca22c6f0a45425965a3680
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41802
Those tests are not executing at all, they're just compiled.
Our internal infra is still depending on some bits of it though, so I'm moving them to `fbandroid/java/com/facebook/fbreact
Changelog:
[Internal] [Changed] - Move legacy tests from OSS to fbandroid/java/com/facebook/fbreact
Reviewed By: rshest
Differential Revision: D51805702
fbshipit-source-id: 2c5cec68efa9854184e981220202d8f356ff690a
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41832
I'm removing the `installArchives` task and all the setup to publish
the Maven Local inside the NPM package as we're not using this entirely
and we won't be able to use it anyway (as the Maven Local is too big to fit an NPM package).
Changelog:
[Internal] [Changed] - Remove the installArchives task
Reviewed By: GijsWeterings
Differential Revision: D51890224
fbshipit-source-id: 3ffdc67a9fe931118596f6f74a5a2df0313ca3f2
Summary:
Apple will require XCode 15 next year to ship to the app store, and it aligns with how we build and test React Native internally.
XCode 15 and 14.3 add support for a lot of [missing C++ 20 features](https://developer.apple.com/xcode/cpp/#c++20) from earlier versions as well.
Last I was aware, Riccardo was onboard with bumping min supported in 0.74 to XCode 15. This change does a slightly more conservative bump to min 14.3, and main of 15.0 (though we might want to move these before 0.74 comes out).
All of this will get migrated over to GHA soon enough as well, but... formalizing this is the only thing blocking usage of C++ 20 ranges today.
Changelog:
[ios][breaking] - Require XCode >= 14.3
Pull Request resolved: https://github.com/facebook/react-native/pull/41798
Test Plan:
1. CircleCI Passes
2. Can still boot RNTester from XCode with code signing related changes.
Reviewed By: cortinico
Differential Revision: D51840617
Pulled By: NickGerleman
fbshipit-source-id: 58f8951a436eb7c892a00432a8aad0ddd0a49da1
Summary:
iOS?ios?android?Android?
Always making typos when using the local testing script with the platform argument... No more!
## Changelog:
[INTERNAL][ADDED] - Improved E2E local testing script to be more flexible
Pull Request resolved: https://github.com/facebook/react-native/pull/41751
Reviewed By: cortinico
Differential Revision: D51758529
Pulled By: huntie
fbshipit-source-id: d9e633567a59fcfac1057cf1f21714ccef27ebb2
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41526
CI failures in Windows JS tests recently (https://github.com/facebook/react-native/pull/41463) were caused by the triggering of Babel registration during tests, due to an import of `packages/dev-middleware` (index), breaking subsequent transformation of other tests.
## Root cause
Example of a problematic import:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/dev-middleware/src/__tests__/ServerUtils.js#L15
..which triggers a Babel registration:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/dev-middleware/src/index.js#L16-L18
That registration behaves differently on Windows due to the `ignore: [/\/node_modules\/\]`, which doesn't match against Windows path separators - Babel matches against system separators.
In particular, this changed whether `node_modules/flow-parser` was transformed when loading the RN Babel transformer. Transforming this file causes a `console.warn` from Babel due to its size:
> [BABEL] Note: The code generator has deoptimised the styling of /Users/robhogan/workspace/react-native/node_modules/flow-parser/flow_parser.js as it exceeds the max of 500KB.
This throws due to our setup:
https://github.com/facebook/react-native/blob/a5d8ea4579c630af1e4e0fe1d99ad9dc0915df86/packages/react-native/jest/local-setup.js#L27
This all manifests as the first test following a Babel registration (within the same Jest worker) that requires the RN Babel transformer throwing during script transformation.
## This change
This is the minimally disruptive change that makes Babel registration behaviour consistent between Windows and other platforms. The more durable solution here would be *not* to rely on any Babel registration for Jest, which has its own `ScriptTransformer` mechanism for running code from source. Given the fragile way our internal+OSS Babel set up hangs together that's a higher-risk change, so I'll follow up separately.
Changelog: [Internal]
Reviewed By: huntie
Differential Revision: D51424802
fbshipit-source-id: 8b733c0c159ee84690aef04abced682d126c6d27
Summary:
This change allow our CI to run E2E tests using a specific tag in the commit message
## Changelog:
[Internal] - Allow to run e2e test using a specific tag in the last commit message
Pull Request resolved: https://github.com/facebook/react-native/pull/41308
Test Plan:
CircleCI is green
Tested interacting with the PR/branch
Reviewed By: NickGerleman
Differential Revision: D50975588
Pulled By: cipolleschi
fbshipit-source-id: 6318800d7e86e1cab394af2b320e280304189dd2
Summary:
Last week, I modified the e2e script to make sure it was working properly with 0.73.
This change backport those changes in main
## Changelog:
[Internal] - Backport e2e script changes
Pull Request resolved: https://github.com/facebook/react-native/pull/41332
Test Plan: Tested locally
Reviewed By: dmytrorykun
Differential Revision: D51025796
Pulled By: cipolleschi
fbshipit-source-id: 89ecd3701eaac4ba4bdde2c640df45a158329158
Summary:
While releasing RN 0.73.0-RC3, we relaized that the e2e test script was bugged for Android when used to test RNTestProject with the `-c` option.
There were 2 problems:
- The downloaded maven-local was not actually used because it doesn't work with a zip. (We were always downloading a version from Maven)
- The versions of React Native between maven-local and the locally packaged React Native were different.
This change fixes the script by:
- Downloading maven-local
- Unzipping maven-local and passing the new folder to the Android app
- Downloading the React Native version that has been packaged in CI
By unzipping maven-local and using the unzipped folder, we make sure that Android is actually using the local repository.
By downloading both the packaged react native and the maven-local from the same CI workflow, we ensure that the versions are aligned.
This also speeds-up further the Android testing.
While running this change, we also moved the `pod install` step inside the `if (iOS)` branch, so we do not install Cocoapods if we need to test
Android.
## Changelog:
[Internal] - Fix Android E2E test script when downloading artefacts from CI
Pull Request resolved: https://github.com/facebook/react-native/pull/41172
Test Plan: Tested locally on both main and 0.73-stable, on both Android and iOS
Reviewed By: cortinico
Differential Revision: D50651448
Pulled By: cipolleschi
fbshipit-source-id: 70a9ed19072119d19c5388e8a4309d7333a08e13
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/39540
This simplifies the use of Codegen when creating dev builds of `rn-tester` in the monorepo. It now runs from source for this internal scenario, and this package is now built using the shared monorepo build setup.
Changes:
- Migrate `packages/react-native-codegen` to the shared `yarn build` setup.
- Update package to use `"exports"` field and wrap entry point modules with `babel-register` (NOTE: This is only required for each entry point internally used in the monorepo).
- Fixup small Flow syntax quirks that fail under `hermes-parser`.
- Remove `BuildCodegenCLITask` task from Android build.
- Remove Codegen `build.sh` call from iOS build, use `require.resolve` for `combine-js-to-schema-cli.js` entry point.
Externally significant FYIs:
- `react-native/codegen` is converted to use the `"exports"` field — it should export all `.js` files, as before.
- `codegenPath` is now ignored and marked as deprecated on `ReactExtensions.kt`.
NOTE: TypeScript auto-generation is not yet enabled on this package, since it uses CommonJS `module.exports` syntax (unsupported by `flow-api-translator`).
Changelog: [Internal]
Reviewed By: cipolleschi
Differential Revision: D49370200
fbshipit-source-id: 992913155169912ea1a3cb24cb26efbd3f783058
Summary:
E2E tests in OSS are expensive and flaky.
They already prevented some broken changes to land on main, but as of today:
- they are always green, so they are not bloking
- nobody is looking at the reporting job
- the reporting job takes a lot of time to run and prevent other useful signals to be available soon
- it is expensive
So we decide to disable them for the time being, while we iterate on those with Callstack and MSFT.
## Changelog:
[Internal] - Disable E2E tests
Pull Request resolved: https://github.com/facebook/react-native/pull/41153
Test Plan: CircleCI stays green
Reviewed By: cortinico
Differential Revision: D50552818
Pulled By: cipolleschi
fbshipit-source-id: 7160a8074492c3c9a55485d8a17a6883eb4b35b5
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41118
Updates the release process and guidance for `debugger-frontend`, now that the source [facebookexperimental/rn-chrome-devtools-frontend](https://github.com/facebookexperimental/rn-chrome-devtools-frontend) repo is published.
The `sync-and-build` script now requires a `--branch` argument, allowing us to match release branches across repos for hotfixes (e.g. `0.73-stable`).
Changelog: [Internal]
Reviewed By: motiz88
Differential Revision: D50496327
fbshipit-source-id: 671fd1581e23032eec0a419a6e50dac6c76feeb0
Summary:
Pull Request resolved: https://github.com/facebook/react-native/pull/41051
Strictifies flow to flow strict-local in files where doing that doesn't cause new flow errors.
Changelog: Internal
Reviewed By: yungsters
Differential Revision: D50369011
fbshipit-source-id: b4a5a26b839b7327a3178e6f5b35246dea365a38
Summary:
CircleCI is red because verdaccio fails to publish `normalize-colors`.
For some old dependencies, normalize-colors has been published on the official npmjs with the version we needs.
In order to mitigate the RN red ci, we can consume them directly from NPMJS.
As a followup, we created a task to investigate it next week.
## Changelog:
[Internal] - Skip publishing or normalize colors
Pull Request resolved: https://github.com/facebook/react-native/pull/40977
Test Plan:
Tested locally, running verdaccio and simulating CI. It worked.
CircleCI is green
Reviewed By: robhogan
Differential Revision: D50300449
Pulled By: cipolleschi
fbshipit-source-id: 2259b450deff15a117d1de4690bcfe8a9ba7d115
Alex Hunt
Luna Wei
Riccardo Cipolleschi
Rob Hogan
Rubén Norte
David Vacca
szymonrybczak
Nicola Corti
Luna Wei
Moti Zilberman
Dmitry Rykun
Oskar Kwaśniewski
Nick Gerleman
Thibault Malbranche
Gijs Weterings
Riccardo Cipolleschi
Gijs Weterings