Summary:D3092867 / 1d3db4c5dc caused deadlock when chrome debugging was turned on, so it was reverted as D3128586 / 144dc30661.
The reason: I was calling `[_bridge dispatchBlock:^{ [self flushEventsQueue]; } queue:RCTJSThread];` from main thread and expecting it will `dispatch_async` to another,
since a held lock was being accessed the dispatched block and was released after the dispatch.
Turns out `RCTWebSocketExecutor` (which is used when chrome debugger is turned on) executes all blocks dispatched this way to `RCTJSThread` synchronously on the main thread.
This resulted in a deadlock. The "dispatched" block was trying to acquired lock which held by the same thread in the dispatching phase.
A fix for this is pretty simple. We will release the lock before dispatching the block.
However it's not super straightforward to see this won't introduce some race condition in a case with two threads where we would end up with events not being processed.
My thinking why that shouldn't happen goes like this: We could get in a bad state if `flushEventsQueue` would run on JS thread while `sendEvent:` is running on MT.
(I don't have a specific example how, maybe it's not possible. However when I show this case is safe we know we are good.)
The way how locking is setup in this diff the only possible scenario where these two threads would execute in these methods concurrently is JS holding the lock and MT going to enqueue another block on JS thread (since that's outside of "locked" zone).
But this scenarion can never happen, since if MT is about to enqueue the block on JS thread it means there cannot be a not yet fully executed block on JS thread.
Therefore nothing bad can happen.
So this diff brings back the reverted diff and adds to it the fix for the deadlock.
Reviewed By: javache
Differential Revision: D3130375
fb-gh-sync-id: 885a166f2f808551d7cd4e4eb98634d26afe6a11
fbshipit-source-id: 885a166f2f808551d7cd4e4eb98634d26afe6a11
Summary:Previously, (mostly touch and scroll) event handling on iOS worked in a hybrid way:
* All incoming coalesce-able events would be pooled and retrieved by js thread in the beginning of its frame (all of this happens on js thread)
* Any non-coalesce-able event would be immediately dispatched on a js thread (triggered from main thread), and if there would be pooled coalesce-able events they would be immediately dispatched at first too.
This behavior has a subtle race condition, where two events are produced (on MT) in one order and received in js in different order.
See https://github.com/facebook/react-native/issues/5246#issuecomment-198326673 for further explanation of this case.
The new event handling is (afaik) what Android already does. When an event comes we add it into a pool of events and dispatch a block on js thread to inform js there are events to be processed. We keep track of whether we did so, so there is at most one of these blocks waiting to be processed. When the block is executed js will process all events that are in pool at that time (NOT at time of enqueuing the block).
This creates a single way of processing events and makes it impossible to process them in different order in js.
The tricky part was making sure we don't coalesce events across gestures/different scrolls. Before this was achieved by knowing that gestures and scrolls start/end with non-coalesce-able event, so the pool never contained events that shouldn't be coalesced together. That "assumption" doesn't hold now.
I've re-added `coalescingKey` and made touch and scroll events use it to prevent coalescing events of the same type that should remain separate in previous diffs (see dependencies).
On top of it it decreases latency in events processing in case where we get only coalesce-able events. Previously these would be processed at begging of the next js frame, even when js would be free and could process them sooner. This delay is done, since they would get processed as soon as the enqueued block would run.
To illustrate this improvement let's look at these two systraces.
Before: https://cloud.githubusercontent.com/assets/713625/14021417/47b35b7a-f1d3-11e5-93dd-4363edfa1923.png
After: https://cloud.githubusercontent.com/assets/713625/14021415/4798582a-f1d3-11e5-8715-425596e0781c.png
Reviewed By: javache
Differential Revision: D3092867
fb-gh-sync-id: 29071780f00fcddb0b1886a46caabdb3da1d5d84
fbshipit-source-id: 29071780f00fcddb0b1886a46caabdb3da1d5d84
Summary: This was previously removed in D2884587, but we will need it going forward. See D3092867 for reasons why it's necessary again.
Reviewed By: javache
Differential Revision: D3092848
fb-gh-sync-id: 0d10dbac4148fcc8e035d32d8eab50f876d99e88
fbshipit-source-id: 0d10dbac4148fcc8e035d32d8eab50f876d99e88
Summary:
Currently only scroll events are send through `sendEvent`, and all of them are can be coalesced. In future (further in the stack) touch events will go through there as well, but they won't support coalescing.
In order to ensure js processes touch and scroll events in the same order as they were created, we will flush the coalesced events when we encounter one that cannot be coalesced.
public
___
//This diff is part of a larger stack. For high level overview what's going on jump to D2884593.//
Reviewed By: nicklockwood
Differential Revision: D2884591
fb-gh-sync-id: a3d0e916843265ec57f16aad2f016a79764dcce8
Summary:
I want to use the `RCTEvent` protocol for touch events as well. That's why I'm removing not very well defined `body` property and replacing it with `arguments` method, which will return an array that will be passed directly to the js call.
I think this makes sense because there is no unified arguments format for all events and and the called js method (`moduleDotMethod`) is already event specific.
This way touch events and scroll events can result in calling a completely different js function with a completely different arguments (what they indeed currently do).
public
___
//This diff is part of a larger stack. For high level overview what's going on jump to D2884593.//
Reviewed By: nicklockwood
Differential Revision: D2884590
fb-gh-sync-id: 2c1885c3414e255d8572c0fbbbfe62a23d94dd06
Summary:
This property was never used, so I'm removing it.
public
___
//This diff is part of a larger stack. For high level overview what's going on jump to D2884593.//
Reviewed By: javache
Differential Revision: D2884587
fb-gh-sync-id: acd5e576cd13a02e77225f3b308232f8331d3b61
Summary:
`RCTBaseEvent` was never used. This diff removes it.
public
___
//This diff is part of a larger stack. For high level overview what's going on jump to D2884593.//
Reviewed By: javache
Differential Revision: D2884585
fb-gh-sync-id: 66a6afcda3b5baec7f768682da215570f6d33bb1
Summary:
Our events all follow a common pattern, so there's no good reason why the configuration should be so verbose. This diff eliminates that redundancy, and gives us the freedom to simplify the underlying mechanism in future without further churning the call sites.
Summary:
This introduces event counts to make sure JS doesn't set out of date values on
native text inputs, which can cause dropped characters and can mess with
autocomplete, and obviates the need for the input buffering which added lag and
complexity to the component. Made sure to test simulated super-slow JS text
event processing to make sure characters aren't dropped, as well as typing
obviously correctable words and making sure autocomplete works as expected.
TextInput is now a controlled input by default without causing any issues for
most cases, so I removed the `controlled` prop.
Fixes selection state jumping by restoring it after setting new text values, so
highlighting the middle of some text in the new ReWrite example and hitting
space will replace that selection with an underscore and keep the cursor at a
sensible position as expected, instead of jumping to the end.
Ads `maxLength` prop to support the most commonly needed syncronous behavior:
preventing the user from typing too many characters. It can also be used to
prevent users from continuing to type after entering special characters by
changing it to the current length after a regex match. Made sure to verify it
works well with pasted input (including in the middle of existing text),
truncating it and collapsing the selection the same way it does on the web.
Fixes bug in TextEventsExample where it wouldn't show the submit and end events,
even though there were firing correctly.
Martin Kralik
Alex Kotliarskyi
Dave Sibiski
Pieter De Baets
Nick Lockwood
Spencer Ahrens
Tadeu Zagallo
Ben Alpert
