Summary:
@public
Allow to pass an optional completion block to the bridge JS calls.
The block will be called from the JS thread, after the javascript has finished
running and the returned calls have been processed/dispatched to the native modules.
Test Plan: Added `testCallbackIsCalledOnTheRightTime` to `RKBatchedBridgeTests`
Summary:
D2052669 introduced a block for objects that had to be invalidated on the main
thread, but after the JS thread objects, but the block was being dispatched on
the JS thread.
@public
Test Plan:
I added `RCTAssertMainThread()` to the `mainThreadInvalidate` block, it was
crashing on reload, but now it should work as expected.
Summary:
The parent RCTBridge no longer tracks the JS loading since that has been handed off to the RCTBatchedBridge. To make the `loading` property accurate again, just expose the batch bridge's loading property from the parent bridge (note: I didn't make it KVO-compliant).
Fixes#1199
Closes https://github.com/facebook/react-native/pull/1200
Github Author: James Ide <ide@jameside.com>
Test Plan: Imported from GitHub, without a `Test Plan:` line.
Summary:
Add the RCTBatchedBridge object to the notification's userInfo for consistency with RCTJavaScriptDidLoadNotification, and set the target object to `_parentBridge`.
cc @nicklockwood @tadeuzagallo
Closes https://github.com/facebook/react-native/pull/1243
Github Author: James Ide <ide@jameside.com>
Test Plan: Imported from GitHub, without a `Test Plan:` line.
Summary:
After the RCTBatchedBridge refactor, the `-[NSNotificationCenter addObserver]` and `[removeObserver]` calls got divided between RCTBridge and RCTBatchedBridge. This diff does two things:
- Moved `removeObserver` out of RTCBatchedBridge and into `-[RCTBridge invalidate]`
- Moved `addObserver` from `bindKeys` to `setUp`. This is so that `-[RCTBridge reload]` will re-add the observer after invalidating and removing the observer
cc @tadeuzagallo
Closes https://github.com/facebook/react-native/pull/1212
Github Author: James Ide <ide@jameside.com>
Test Plan: Imported from GitHub, without a `Test Plan:` line.
Summary:
The RCTRootView creates a underlying RCTRootContentView that was deallocated when
the bridge modules were deallocated. That doesn't work when the bridge is held.
@public
Test Plan:
Launch Groups, put a breakpoint on `-[RCTRootContentView dealloc]`, enter and
leave a group page. It should be called now.
Summary:
This provides a way to get notified when a bridge fails to load JS, allowing apps to handle the error.
Closes https://github.com/facebook/react-native/pull/1085
Github Author: James Ide <ide@jameside.com>
Test Plan: run the UIExplorer app with the packager server not running, and verify that the notification is posted.
Summary:
cc @a2 @nicklockwood
This diff introduces a new macro called `RCT_EXPORT_NAMED_METHOD`, which is like `RCT_EXPORT_METHOD` but lets you choose the name of the method in JS. This diff is backwards compatible with the `RCT_EXPORT_METHOD` and legacy `RCT_EXPORT` macros.
The entries in the data segment now contain `__func__`, the Obj-C selector signature, and the JS name. If the JS name is `NULL`, we take the legacy `RCT_EXPORT` code path. If the JS name is an empty string, we use the Obj-C selector's name up to the first colon (that is, the behavior of `RCT_EXPORT_METHOD`).
Since there are three values in each data segment entry, the macros now specify 1-byte alignment. Without the byte alignment, the compiler defaults to 2-byte alignment meaning that each entry takes up 4 bytes instead of 3. The extra byte isn't a concern but being explicit about the alignment should reduce compiler surprises.
Closes https://github.com/facebook/react-native/pull/802
Github Author: James Ide <ide@jameside.com>
Test Plan: Imported from GitHub, without a `Test Plan:` line.
Tadeu Zagallo
Christopher Chedeau
James Ide
Nick Lockwood
Bryce Redd
Alex Kotliarskyi
Alex Akers
Spencer Ahrens
Ben Hiller