Summary:
This is a follow up of the work by brentvatne in #5052. This addresses the feedback by astreet.
- Uses ReactChoreographer with a new callback type
- Callback dispatch logic moved to JS
- Only calls into JS when needed, when there are pending callbacks, it even removes the Choreographer listener when no JS context listen for idle events.
** Test plan **
Tested by running a background task that burns all remaining idle time (see new UIExplorer example) and made sure that UI and JS fps stayed near 60 on a real device (Nexus 6) with dev mode disabled. Also tried adding a JS driven animation and it stayed smooth.
Tested that native only calls into JS when there are pending idle callbacks.
Also tested that timers are executed before idle callback.
```
requestIdleCallback(() => console.log(1));
setTimeout(() => console.log(2), 100);
burnCPU(1000);
// 2
// 1
```
I did *not* test with webworkers but it should work as I'm using executor tokens.
Closes https://github.com/facebook/react-native/pull/8569
Differential Revision: D3558869
Pulled By: astreet
fbshipit-source-id: 61fa82eb26001d2b8c2ea69c35bf3eb5ce5454ba
Summary:
In preparation for Blob support (wherein binary XHR and WebSocket responses can be retained as native data blobs on the native side and JS receives a web-like opaque Blob object), this change makes RCTNetworking aware of the responseType that JS requests. A `xhr.responseType` of `''` or `'text'` translates to a native response type of `'text'`. A `xhr.responseType` of `arraybuffer` translates to a native response type of `base64`, as we currently lack an API to transmit TypedArrays directly to JS. This is analogous to how the WebSocket module already works, and it's a lot more versatile and much less brittle than converting a JS *string* back to a TypedArray, which is what's currently going on.
Now that we don't always send text down to JS, JS consumers might still want to get progress updates about a binary download. This is what the `'progress'` event is designed for, so this change also implements that. This change also follows the XHR spec with regards to `xhr.response` and `xhr.responseText`:
- if the response type is `'text'`, `xhr.responseText` can be peeked at by the JS consumer. It will be updated periodically as the download progresses, so long as there's either an `onreadystatechange` or `onprogress` handler on the XHR.
- if the response type is not `'text'`, `xhr.responseText` can't be accessed and `xhr.response` remains `null` until the response is fully received. `'progress'` events containing response details (total bytes, downloaded so far) are dispatched if there's an `onprogress` handler.
Once Blobs are landed, `xhr.responseType` of `'blob'` will correspond to the same native response type, which will cause RCTNetworking to only send a blob ID down to JS, which can then create a `Blob` object from that for consumers.
Closes https://github.com/facebook/react-native/pull/8324
Reviewed By: javache
Differential Revision: D3508822
Pulled By: davidaurelio
fbshipit-source-id: 441b2d4d40265b6036559c3ccb9fa962999fa5df
Summary: Changes the time drift error into a warning that will only get logged once per debugging session.
Reviewed By: jingc
Differential Revision: D3539067
fbshipit-source-id: 357db15750d867a91c39b5fc5fd6ed4ae2852bc7
Summary: This is caused by receiving \u2028/2029 in callbacks/function calls. The correct solution is to not evaluate these strings as scripts but instead parse them as json and pass them through the JSC API.
Reviewed By: lexs
Differential Revision: D3543098
fbshipit-source-id: 4d8acce1d510bb17361d32103d4738fc0208b0a8
Summary: Add a new interface to JSC that allows loading a file lazily from disk, i.e. using mmap, instead of loading the whole file upfront and copying into the VM.
Reviewed By: michalgr
Differential Revision: D3534042
fbshipit-source-id: 98b193cc7b7e33248073e2556ea94ce3391507c7
Summary:
https://github.com/facebook/css-layout/pull/199
- Nodes were measured with the assumption of being text nodes (height depends on width) when determining flex basis. This is not always true. Even when we are just interested in the main axis (flex basis) we need to correctly constrain the cross axis.
- Some tests were wrong. Measuring texts.big and expecting it to have textSizes.smallHeight which doesn't make a lot of sense.
Reviewed By: vjeux
Differential Revision: D3510163
fbshipit-source-id: ee53b548dd078005fdd153d279e4c7fef3dd02d0
Summary:
https://github.com/facebook/css-layout/pull/200
Make use of max dimension styles to allow root to be measured with AT_MOST measure mode
Reviewed By: IanChilds
Differential Revision: D3513505
fbshipit-source-id: 169f49717e896eb6270b52fb7115ce005aa0e3a8
Summary: If stacktrace-parser can't parse a stack trace, it'll return null. This can cause us to accidentally enter a crash loop where whenever you start your app, you load the last JS bundle you had, get a crash, and then hard crash trying to print the stack trace.
Reviewed By: frantic
Differential Revision: D3528141
fbshipit-source-id: 1146f43bc40492bfa79b6a1c0f81092383896164
Summary:
This adds proper support for tracking a TextInput content size as discussed in #6552 by adding a new callback that is called every time the content size changes including when first rendering the view.
Some points that are up for discussion are what do we want to do with the onChange callback as I don't see any use left for it now that we can track text change in onChangeText and size changes in onContentSizeChange. Also a bit off topic but should we consider renaming onChangeText to onTextChange to keep the naming more consistent (see [this naming justification](https://twitter.com/notbrent/status/709445076850597888)).
This is split in 2 commits for easier review, one for iOS and one for android.
The iOS implementation simply checks if the content size has changed everytime we update it and fire the callback, the only small issue was that the content size had several different values on initial render so I added a check to not fire events before the layoutSubviews where at this point the value is g
Closes https://github.com/facebook/react-native/pull/8457
Differential Revision: D3528202
Pulled By: dmmiller
fbshipit-source-id: fefe83f10cc5bfde1f5937c48c88b10408e58d9d
Summary:
So `PanReponder.onPanResponderRelease/onPanResponderTerminate` receive a `gestureState` object containing a `onPanResponderTerminate.vx/vy` property. On Android and iOS, they appear to be orders of magnitude different, which appear to be due to the different scale of timestamps that are used when generating touch events.
This pull request fixes the timestamps to be milliseconds on both platforms (since I assume iOS is the more authoritative one, and is the one that `react-native-viewpager`'s vx thresholds written written to compare against.)
As far as I can tell, the RN code doesn't use the `vx/vy` properties, so they should be okay. And looks like the RN code only cares about relative values of `startTimestamp/currentTimestamp/previousTimestamp` though, so should be fine too. it's quite possible there will be downstream android breakage with this change, particularly for those who are already compensating for the RN discrepancy.
Closes https://github.com/facebook/react-native/pull/8199
Differential Revision: D3528215
Pulled By: davidaurelio
fbshipit-source-id: d81732e50a5ece2168e8347309d8d52a0db42951
Summary:
So `PanReponder.onPanResponderRelease/onPanResponderTerminate` receive a `gestureState` object containing a `onPanResponderTerminate.vx/vy` property. On Android and iOS, they appear to be orders of magnitude different, which appear to be due to the different scale of timestamps that are used when generating touch events.
This pull request fixes the timestamps to be milliseconds on both platforms (since I assume iOS is the more authoritative one, and is the one that `react-native-viewpager`'s vx thresholds written written to compare against.)
As far as I can tell, the RN code doesn't use the `vx/vy` properties, so they should be okay. And looks like the RN code only cares about relative values of `startTimestamp/currentTimestamp/previousTimestamp` though, so should be fine too. it's quite possible there will be downstream android breakage with this change, particularly for those who are already compensating for the RN discrepancy.
Closes https://github.com/facebook/react-native/pull/8199
Differential Revision: D3528215
Pulled By: dmmiller
fbshipit-source-id: cbd25bb7e7bb87fa77b661a057643a6ea97bc3f1
Summary: D3516692 incorrectly assumed that the Spacing object was immutable. This is not true as it can change in a measure pass on the shadow thread while the UI is displaying some other data. This fixes it to just pull the values we need and pass them instead of the object.
Reviewed By: foghina
Differential Revision: D3523004
fbshipit-source-id: 056baecf60a9c2f2494d173a3c60396a93b97d42
Summary: Text was not correctly respecting padding. We would account for it when measuring, but then not actually apply the padding to the text. This adds support for proper padding
Reviewed By: andreicoman11
Differential Revision: D3516692
fbshipit-source-id: 9a0991d89e9194c0e87af0af56c6631a6b95700a
Summary: We prefer not showing the accurate time information in the exception message of time drift, in order to be able to group this kind of exception only by exception message.
Reviewed By: foghina
Differential Revision: D3515390
fbshipit-source-id: f38bf3091ba749d4daeae24f7d42653b76c3289c
Summary: This removes `node_modules/react` from the list of directories that are used for haste module resolutions. Modules required from React are now imported with `require('react/lib/…')`.
Reviewed By: astreet
Differential Revision: D3509863
fbshipit-source-id: 32cd34e2b8496f0a6676dbe6bb1eacc18124c01e
Summary:
Idle WebSocket connections get reset after a few minutes of inactivity. To prevent this, most WebSocket implementations offer sending special ping messages. This PR adds a method `sendPing()` to WebSocket. Ping payloads are not supported.
Manual testing can be done by adding `connection.on('ping', _ => console.log('Received ping'));` to a ws connection or using a packet sniffer while sending pings.
Closes https://github.com/facebook/react-native/pull/8505
Differential Revision: D3516260
Pulled By: dmmiller
fbshipit-source-id: cfebf5899188ae53254d5be6b666a9075e0eed89
Summary: These are helpful for development and shouldn't affect anything when DEBUG=false
Reviewed By: lexs
Differential Revision: D3515015
fbshipit-source-id: 0bd5ff833f90fea8e70b3103eb1f9bc803bef27c
Summary:
This should revert back to using old bridge by default until we fix gradle script to build new bridge for OSS correctly
Closes https://github.com/facebook/react-native/pull/8539
Differential Revision: D3514263
Pulled By: bestander
fbshipit-source-id: db9e0232b9bde27206814212dedc487c366e3511
Summary: Move ImageSource out of ReactImageView so it can be used by other views dealing with images. Make ResourceDrawableIdHelper static in the process.
Reviewed By: dmmiller
Differential Revision: D3505103
fbshipit-source-id: 776a5b22a6934eeaa370826cb71c00cd937d3187
Summary:
Currently, DevTools only work under ios (although this is undocumented!),
because the JavaScriptEngine initialization process skips setupDevTools() on
android.
DevTools work fine with Android, as tested on 0.26, 0.27, and 0.28 using
Nuclide's inspector.
For reference, the relevant issue on react-devtools.
Closes https://github.com/facebook/react-native/pull/8095
Reviewed By: bestander
Differential Revision: D3492788
Pulled By: andreicoman11
fbshipit-source-id: 1eda9196d7125da19a8d7baaab22b61b744ca629
Summary:
Found a couple of places where we were copy-pasting the logic `ResourceDrawableIdHelper` had. This class was only available on the image module and it had package visibility. I moved it to its own module so that we can easily use it from others.
This diff is pretty simillar to 54ed44628d but it fixes a bug due to which we had to revert it.
Reviewed By: andreicoman11
Differential Revision: D3499062
fbshipit-source-id: f912f57e5ac21a9f30fe42067c784f49fa46ed48
Summary:
ReactTextView occasionally crashes when `setText` is called. Doing some cursory research, it looks like this could be a bug in Android. We also suspect it might be related to removeClippedSubviews though.
The LayoutParams don't actually matter because RN controls layout, but on occasion (haven't narrowed down what this is...) `mLayout` is non-null and triggers relayout during `setText`, which fails because the LayoutParams are null. jesseruder came up with this fix and it appears to be working well.
Stack trace: http://pastebin.com/P8VbxvPz
Closes https://github.com/facebook/react-native/pull/7011
Differential Revision: D3508097
fbshipit-source-id: 12b4aa11e42112c8ba19a1af325e3ee9a232d08f
Janic Duplessis
Chris Hopman
Emil Sjolander
Philipp von Weitershausen
sathya
Konstantin Raev
Michał Gregorczyk
Tim Yung
Andy Street
Tadeu Zagallo
Clement Genzmer
Alexander Blom
Mike Lambert
Marc Horowitz
Dave Miller
Siqi Liu
David Aurelio
danielbasedow
Felix Oghina
Franklyn Tackitt
James Ide
Andrew Y. Chen
