Summary:
Context: I'm trying to add support for sending packager progress events to the client that is downloading the bundle over HTTP multipart response.
The idea is for the client to send `Accept: multipart/mixed` header, and if present the server will stream progress events to the client. This will ensure the change is backwards-compatible - the clients who don't know about progress events won't receive them.
In the future we can use this approach to download RAM bundle modules in one request.
Reviewed By: davidaurelio
Differential Revision: D3926984
fbshipit-source-id: 39a6e38e40a79f7a2f2cf40765a0655fb13b7918
Summary:
Context: I'm trying to add support for sending packager progress events to the client that is downloading the bundle over HTTP multipart response.
In order to do that I need the server to know about these events. Currently the bundler doesn't expose any hooks for monitoring the progress, so this diff introduces `onProgress` option for that.
Reviewed By: davidaurelio
Differential Revision: D3926806
fbshipit-source-id: b7d9c649df4f94ddf5082791209844610650325e
Summary: Modified `node-haste` implementation to use the much faster `jest-haste-map` under the hood. The underlying `fastfs` now gets passed the entire file list from the `jest-haste-map` rather than crawl the filesystem.
Reviewed By: cpojer
Differential Revision: D3724387
fbshipit-source-id: 447d58ea0edf283662ec23d1e2deee992cf8d240
Summary: Changed “Finding dependencies” and “Building Module Graph” to “Transforming modules” and “Initializing Packager” for more clarity
Reviewed By: matryoshcow
Differential Revision: D3950811
fbshipit-source-id: 7a2e655ef7e1655244ce427e7adc8c1e5afa7329
Summary:
React native's reload javascript option doesn't always work on iOS devices since version 0.29, as described in https://github.com/facebook/react-native/issues/9617
It only doesn't work when you have a mac on a wireless connection, because react-native-xcode.sh can't find your IP address correctly in this case and will just fallback to use the pre-bundling option on your app.
This small change in react-native-xcode.sh fixed this issue for our project and should fix this issue for all mac users that use wireless connection and that will run a debug version of the app on a real iOS device.
Closes https://github.com/facebook/react-native/pull/9964
Differential Revision: D3923035
fbshipit-source-id: 436cfa2103e10221270034233552ce34720505d3
Summary: File platform identifiers (`fileName.<platform>.js`) are no longer part of the blacklist regular expression. This allows the upcoming `jest-haste-map` to include all files for all platforms, therefore enabling Packager to build bundles for different platforms using the same `HasteMap`(jest-haste-map) instance.
Reviewed By: davidaurelio
Differential Revision: D3907508
fbshipit-source-id: d7d7f3bd93287a634a1ef0590a736d021be2aaa5
Summary:
The current test checks whether the target platform is `'ios'`. When building with windows, errors will also be swallowed silently.
This adds a check whether the target platform is `'android'`, where we have to avoid throwing for FB-internal reasons.
Reviewed By: martinbigio
Differential Revision: D3907025
fbshipit-source-id: 5abeb06b9121265fe1aa3932ad1785f148f04ddf
Summary:
We could probably relieve a lot of pain (in [this issue](https://github.com/facebook/react-native/issues/4968) specifically) by fixing the error message for when a module cannot be resolved after trying every relevant `node_modules` directory.
Currently, you get a confusing error message that only gives you the last `node_modules` directory checked. By creating the error message where we can access the `searchQueue`, we're able to provide all of the attempted `node_modules` directories.
Here's an example error message:
```
Unable to resolve module leftpad from /Users/aleclarson/ReactProject/src/stuff/index.js:
Module does not exist in the module map or as these directories:
/Users/aleclarson/ReactProject/src/stuff/node_modules/leftpad
/Users/aleclarson/ReactProject/src/node_modules/leftpad
/Users/aleclarson/ReactProject/node_modules/leftpad
/Users/aleclarson/node_modules/leftpad
/Users/node_modules/leftpad
```
Closes https://github.com/facebook/react-native/pull/9832
Differential Revision: D3895408
Pulled By: davidaurelio
fbshipit-source-id: 872c9a3bb3633f751ec69b586a261616578ed511
Summary:
Adds a new URL option to the packager server called "assetPlugin". This can be a name of a Node module or multiple Node modules (`assetPlugin=module1&assetPlugin=module2`). Each plugin is loaded using `require()` and is expected to export a function. Each plugin function is invoked with an asset as the argument. The plugins may be async functions; the packager will properly wait for them to settle and will chain them.
A plugin may be used to add extra metadata to an asset. For example it may add an array of hashes for all of the files belonging to an asset, or it may add the duration of a sound clip asset.
Closes https://github.com/facebook/react-native/pull/9993
Differential Revision: D3895384
Pulled By: davidaurelio
fbshipit-source-id: 0afe24012fc54b6d18d9b2df5f5675d27ea58320
Summary:
We can keep the `searchQueue` cleaner by using `this._fastfs.dirExists` on the potential `node_modules` directory. In addition to avoiding useless lookups, this also makes the error message provided by #9832 much more understandable.
/cc davidaurelio
Closes https://github.com/facebook/react-native/pull/9973
Differential Revision: D3887053
Pulled By: davidaurelio
fbshipit-source-id: 6a5a488bb4217dd0f9db1c9c9f988b498c746ca9
Summary:
Hi,
This PR fixes the problem described by chrisnojima in https://github.com/facebook/react-native/issues/9581#issuecomment-243766310
**Test plan**
In development mode,
- Run an app with an image: `<Image source={ require('./logo.png') }/>`
- Notice that you see the following in packager console:
```txt
6:46:42 PM] <START> processing asset request logo.png
[6:46:42 PM] <END> processing asset request logo.png (1ms)
```
- Reload the app, or navigate to another page of the app with the same image
- Notice that you see again:
```txt
6:47:23 PM] <START> processing asset request logo.png
[6:47:23 PM] <END> processing asset request logo.png (1ms)
```
Now wih the fix applied,
notice that you only see `logo.png` fetched once, even if you reload or show the same image in a different part of the app.
Let me know what you think.
Closes https://github.com/facebook/react-native/pull/9795
Differential Revision: D3876945
Pulled By: davidaurelio
fbshipit-source-id: f41f4719e87644692a690123fd6e54eead9cc87d
Summary:
The introduction of groups for random access bundles broke hot module reloading, because HMR uses a different Bundle class. And that didn’t have the `setRamGroups()` method.
This just adds an empty method on the base class.
**Test plan (required)**
I ran `node scripts/run-ci-e2e-tests.js --android`
Closes https://github.com/facebook/react-native/pull/9982
Differential Revision: D3886096
Pulled By: bestander
fbshipit-source-id: 46e6183607b0275d3cfe0d892360248b7b2154e9
Summary:
Adds the possibility to specify an array of files (group roots) that are used to bundle modules outside of the “startup section” into bigger groups by colocating their code.
A require call for any grouped module will load all modules in that group.
Files contained by multiple groups are deoptimized (i.e. bundled as individual script)
Reviewed By: martinbigio
Differential Revision: D3841780
fbshipit-source-id: 8d37782792efd66b5f557c7567489f68c9b229d8
Summary:
This adds support for `false` values in `package.json` `"browser"` and `"react-native"` mappings.
All `false` values are not longer silently ignored, but redirected to an empty file.
Fixes#9854#9518
Reviewed By: bestander
Differential Revision: D3876521
fbshipit-source-id: 96d1ba03518812bc88c51672c374956eabd40c9b
Summary:
For example, I could have a root named `random` and another root named `randomColor`.
A child of `randomColor` would **incorrectly** pass as a descendant of `random`.
So this commit changes `isDescendant` to do 2 things:
- check for exact matches (eg: the path is a root)
- call `startsWith` like before, but append `path.sep` to avoid false positives
Closes https://github.com/facebook/react-native/pull/9831
Differential Revision: D3864968
fbshipit-source-id: 7fe04913579aa0741840fc925216283304ae3433
Summary:
This piece of code can assemble all transitive dependencies of a set of entry modules. It is supposed to replace `ResolutionRequest.getOrderedDependencies`.
It has the following advantages:
- has minimal API surface to other components of the system (uses two functions, exposes one function)
- allows to separate concerns into loading + transforming files, resolving dependencies, gathering all modules belonging to a bundle (this code), and bundling
- allows to specify multiple entry points
- allows to use any kind of dependency ID as entry point (haste IDs, node module IDs, relative paths, absolute paths – depends on the resolver)
- allows to skip files, which allows callers to incrementally update previously retrieved collections of modules
Reviewed By: cpojer
Differential Revision: D3627346
fbshipit-source-id: 84b7aa693ca6e89ba3c1ab2af9a004e2e0aaed3d
Summary: This makes the `hash` property of asset data depend on asset file contents rather than modification time of files. That means that bundles will be consistent across different checkouts.
Reviewed By: martinbigio
Differential Revision: D3856815
fbshipit-source-id: 8bfea4e0a714f48fc6a4ae5ed2a1426dc8d5868e
Summary:
I'm not sure if this was actually breaking anything, but I don't think `new` before `Promise.race()` is conventional.
Closes https://github.com/facebook/react-native/pull/9817
Differential Revision: D3836814
Pulled By: javache
fbshipit-source-id: 3bde46dca3760c3f7f3427f6346dd5f16f8ac011
Summary:
Explain the **motivation** for making this change. What existing problem does the pull request solve?
Fix issue with browser and react-native module mappings in node-haste when running the packager from Windows.
**Test plan (required)**
Tested on the NPM package uuid, which has a browser mapping, and it worked.
Fixes#9570
Closes https://github.com/facebook/react-native/pull/9572
Differential Revision: D3821761
Pulled By: davidaurelio
fbshipit-source-id: ebbd91f9e3215385afe3b81a97a962b45a6286f9
Summary:
When I use local static files as Image resources, it will occur a 404 error if the image file's name contains some non-latin letters.
The reason is that the request's url will be encoded if it contains some non-latin letters.
Closes https://github.com/facebook/react-native/pull/9604
Differential Revision: D3821328
Pulled By: javache
fbshipit-source-id: bfdc7f71517b5d4ba9e0a013979e5dcf6c31a237
Summary: Activity events now have a `silent` boolean option that specifies whether or not the event phases are to be logged to the console. Defaults to `false`, preserving current behaviour for events that do not explicitly have this option set.
Reviewed By: davidaurelio
Differential Revision: D3810802
fbshipit-source-id: 38d14b9e6c6502fbc73eece9466f20b60d19965e
Summary:
When bringing back `node-haste` to React Native, I left an `fdescribe` in a test that led to ~70 tests being skipped.
This re-enables these tests, and fixes test failures
Reviewed By: cpojer
Differential Revision: D3811225
fbshipit-source-id: 67a16f385759bb829f1f3f559862eab7e78f2097
Summary: Updating an existing bundle is now logged as a telemetric event.
Reviewed By: davidaurelio
Differential Revision: D3804754
fbshipit-source-id: adab3b054a161bd9535ad01d52573fb7bb177d43
Summary:
Revised the Activity implementation
- added flow annotations
- fixed some lint warnings
- added event `options`, specifically a `telemetric` option which indicates that events tagged in this manner are relevant for telemetry. The duration of these events is now highlighted in the Activity log
Reviewed By: bestander, kentaromiura
Differential Revision: D3770753
fbshipit-source-id: 6976535fd3bf5269c6263d825d657ab0805ecaad
Summary:
Motivation: While testing the RC, the bundle step failed (because the reset-cache true flag was still present) for an Archive build causing an app to be distributed without the JS bundle. This was obviously not great because the app just crashed on startup.
This fix just checks that the bundle file is present and if not fails the build, as you would expect it to.
Closes https://github.com/facebook/react-native/pull/9284
Differential Revision: D3749572
fbshipit-source-id: f1ccb8b6cd0b78f6d8267e7b4b0f10b4a3cec5a1
Summary: The logic of assigning module IDs to the initial require calls of a bundle was faulty, counting up from -1 instead of counting down. This change ensures that IDs are -1, -2, ...
Reviewed By: matryoshcow
Differential Revision: D3735560
fbshipit-source-id: 89efa3e73b39c2f8bfed8a6a30487733d1a8b145
Summary: If a bundle failed to build, a file change would trigger the *update bundle* path, without ever being able to resolve the bundle. If a bundle can't be updated, we evict it from the cache.
Reviewed By: cpojer
Differential Revision: D3726567
fbshipit-source-id: a342f00c5a41364551194c33082718e5483fd7a4
Summary: Rejected promises in the cache would prevent the cache from persisting. This removes rejected promises from the cache, so that subsequent successful cache updates can be persisted.
Reviewed By: cpojer
Differential Revision: D3726691
fbshipit-source-id: ddec03676a7a89264fe64b4af4b183cbead638fb
Summary:
This clears the packager server cache for potential changes of module resolutions.
We will make this a lot better in the future, but for now this crude mechanism will have to do.
Reviewed By: cpojer
Differential Revision: D3713143
fbshipit-source-id: 7c81f40e8ec71404c3369211b29f63928d6634b9
Summary:
This is a very hacky solution to make reloads from packager faster for simple file changes.
Simple means that no dependencies have changed, otherwise packager will abort the attempt to update and fall back to the usual rebuilding method.
In principle, this change avoids re-walking and analyzing the whole dependency tree, and just updates modules in existing bundles.
Reviewed By: bestander
Differential Revision: D3713704
fbshipit-source-id: ba182325c4f4003c0a7402ea87444a94c75ebaf8
Summary:
This is a very hacky solution to make reloads from packager faster for simple file changes.
Simple means that no dependencies have changed, otherwise packager will abort the attempt to update and fall back to the usual rebuilding method.
In principle, this change avoids re-walking and analyzing the whole dependency tree, and just updates modules in existing bundles.
Reviewed By: bestander
Differential Revision: D3703896
fbshipit-source-id: abc2a41144536baf969d346522a17044c1c9558b
Summary:
This is a very hacky solution to make reloads from packager faster for simple file changes.
Simple means that no dependencies have changed, otherwise packager will abort the attempt to update and fall back to the usual rebuilding method.
In principle, this change avoids re-walking and analyzing the whole dependency tree, and just updates modules in existing bundles.
Reviewed By: bestander
Differential Revision: D3703896
fbshipit-source-id: 671206618dc093965822aed7161e3a99db69a529
Summary:
We have a weak build machine for React Native app: a poor Mac Book Air with HDD, which is very slow.
So, fs operations sometimes fail because of timeout.
We likely don't care about build time, but it's pretty annoying to restart builds sometimes.
In this PR I make timeout injectable via environment variable.
__USAGE__:
export this variable into environment. It's measured in miliseconds.
Example, for 1 minute timeout:
`export REACT_NATIVE_FSOP_TIMEOUT=60000`
If you don't specify it, it will remain default value: `15000`
This should not break anything, but repair slow builds.
Related to : #9373 , #8794 (I think so)
This PR should handle case of issue and close#9373
Closes https://github.com/facebook/react-native/pull/9374
Differential Revision: D3709325
Pulled By: bestander
fbshipit-source-id: b00c89e10d05362314546faea7a4524f3d327c97
Summary: This removes asset data that is not used at runtime from the bundle.
Reviewed By: javache
Differential Revision: D3628486
fbshipit-source-id: 33cd579c904e0b0e29502df39a4ff92cad43367c
Summary:
https://github.com/facebook/react-native/pull/8219 adds range requests to the asset server, but there was an off-by-one-error that made responses end prematurely. This made (for example) react-native-video not work for video assets. This change fixes the off-by-one error and react-native-video works with assets.
**Test plan (required)**
Try the test in the original pull request for range requests: https://github.com/facebook/react-native/pull/8219
Closes https://github.com/facebook/react-native/pull/9254
Differential Revision: D3680070
fbshipit-source-id: 3f2a18ba9f35b45b340f4a1046bc099b8444eb7d
Summary: Looks like react-addons-perf is sometimes missing entries. This prevents it from crashing.
Reviewed By: davidaurelio
Differential Revision: D3669007
fbshipit-source-id: 18a1102b5ad8dcfd9b80d39e1172ba85ad0e9dd2
Alex Kotliarskyi
Ovidiu Viorel Iepure
David Aurelio
Konstantin Raev
Guilherme Bruzzi
T. Brandon Ashley
aleclarson
James Ide
Cristian Carlesso
Jean Regisser
Charles Dick
Jing Chen
Eric Rozell
wusuopu
Alexander Jarvis
ColCh
Nikhilesh Sigatapu
Spencer Ahrens