Summary: Apparently we can/should not have in RCTConversions because it creates unnecessary dependency to core iOS module.
Reviewed By: mdvacca
Differential Revision: D15055325
fbshipit-source-id: 507f5a40c03b5c261967de4504297d31ecd02783
Summary:
@public
In order to encapsulate property access on `YGStyle`, as a first measure we wrap all fields with accessors.
This will e.g. enable dynamic property storage and instrumentation in the future.
All accessors have a `const` version that allows direct access via `const&`. For mutation, bit fields are wrapped with a custom reference object.
This style allows for the least amount of changes in client code. Property access simply needs appended parens, eg `style.direction` becomes `style.direction`.
Reviewed By: shergin
Differential Revision: D14999096
fbshipit-source-id: fbf29f7ddab520513d4618f5e70094c4f6330b30
Summary:
UAs must adjust border radius values to fit a content box:
>>> Corner curves must not overlap: When the sum of any two adjacent border radii exceeds the size of the border box, UAs must proportionally reduce the used values of all border radii until none of them overlap.
This diff implements that.
Reviewed By: mdvacca
Differential Revision: D15028325
fbshipit-source-id: 368232ffa2fa0409d13759bbbe7fe10f8474c400
Summary:
ShadowTree commits happen concurrently with limited synchronization that only ensures the correctness of the commit from ShadowTree perspective.
At the same time artifacts of the commit () needs to be delivered (also concurrently) to the proper thread and executed in order (not-concurrently). To achieve this we need some synchronization mechanism on the receiving (mounting) side. This class implements this process.
Practically, this diff fixes a problem with glitching UI during the very first render of Fabric screen.
Reviewed By: JoshuaGross
Differential Revision: D15021794
fbshipit-source-id: 62982425300c515e92b91e1e660b45455a5446e9
Summary:
`MountingTransaction` encapsulates all artifacts of `ShadowTree` commit, particularly list of mutations and meta-data.
We will rely on this heavily in the coming diffs.
Reviewed By: JoshuaGross
Differential Revision: D15021795
fbshipit-source-id: 811da7afd7b929a34a81aa66566193d46bbc34f8
Summary:
Registries, providers, providers of registries, registres of providers. All that can be really confusing, but that represents best the constraints and desires that we have:
* We need to be able to register components on-the-fly (so we need a mechanism that will propagate changes);
* We don't want to register ComponentDescriptors separately from ComponentView classes;
* C++ not always gives us abstractions that we want (e.g. pointers to constructors).
After the change, we can remove the whole Buck target that has a bunch of handwritten boilerplate code.
There is a still room for polish and removing some concepts, types or classes but this diff is already huge.
Reviewed By: JoshuaGross
Differential Revision: D14983906
fbshipit-source-id: ce536ebea0c905059c7a4d644dc25233e2809761
Summary:
ComponentDescriptorProvider represents unified way to create a particular descriptor.
Now all ComponentViews (which support RCTComponentViewProtocol) expose a `ComponentDescriptorProvider` which will allow creating and registering ComponentDescriptor instances for all visual components automatically as a part of ComponentView registration process.
Don't panic, everything is still being as explicit as it always was, no magic involved; we just will have only one registration step instead of two parallel.
That also opens a way to register components on the fly.
Reviewed By: JoshuaGross
Differential Revision: D14963488
fbshipit-source-id: 9e9d9166fabaf7b30b35b8647faa6e3a19cd2435
Summary: If some class or category override/implement one of those methods, it implies that subclasses must call super for them. `NS_REQUIRES_SUPER` allows to enforce this constraint.
Reviewed By: JoshuaGross
Differential Revision: D14896804
fbshipit-source-id: a481f16e1f7ab901d70deb6486f2ca1cf19dd8d7
Summary:
Previously we could call `invalidateLayer` twice during a mounting transaction (one for update-props, one for update-layout-metrics), that was sub-optimal.
Now, with `finalizeUpdates:` we can do it only once. That should save us some CPU cycles.
Reviewed By: JoshuaGross
Differential Revision: D14896802
fbshipit-source-id: b6572fb2a4fa48a12b1d1c29144cd7c67f6cff80
Summary: We use this pattern already and seems we use it more. Those two functions introduce a "semantical" wrappers for this context, so now there is no need to think which exact `__bridge ***` qualifier we should use, so it's much less error-prone.
Reviewed By: JoshuaGross
Differential Revision: D14896800
fbshipit-source-id: 85b86bfcefdad5aff0375e7172769df86c001506
Summary:
The new method is being called right after all update methods were called for a particular component view.
It is useful for performing updates that require knowledge of several independent aspects of the compound mounting change (e.g. props *and* layout constraints).
Reviewed By: JoshuaGross
Differential Revision: D14896801
fbshipit-source-id: 485d8c97d81d7a2ad7a7afc209094c328da6ef3c
Summary:
This diff replaces all MountItem classes with a bunch of static C functions that do the same job as classes did.
Seems, originally we overestimated the complexity of MountItem classes and that they ended up being notably trivial. Now, maintaining that even longer would mean paying for abstractions and allocations that we don't really need and writing a lot of tedious code.
Besides that, the one particular change that will be introduced in the coming diffs is not particularly fit very well in the existing class-based model.
This change also should save us many hundreds of allocations and atomic counters bumps, so maybe we can get a millisecond-or-two win.
This diff does not introduce any practical behavioral/logical changes in the mounting layer.
Reviewed By: mdvacca
Differential Revision: D14893764
fbshipit-source-id: 6f1247923ae36f29c12a7d358e2d496cf6c3e298
Summary:
This allows an unsupported component to be rendered as a "unimplemented view" for better visualization of which component is missing. It is off by default, but configurable in the component factory.
For now, the layout simply follows regular <View />, which means the width/height etc is based on the react component styling. The side effect is that components with 0 height/width won't show up at all.
Reviewed By: mdvacca
Differential Revision: D14869656
fbshipit-source-id: f31e012fb7dc1c64fcc431ea5aa45079a23a618e
Summary:
In the case where the first rootView we load enables Fabric, the UIManagerBinding ended up calling AppRegistry (JS) too early, before the bridge was fully ready. This means AppRegistry wasn't set up yet, causing redbox.
To fix this, RuntimeExecutor impl should ask the bridge to queue the callback instead of asking the MessageQueueThread directly. MessageQueueThread only works well IFF the bridge is alive, but it has no knowledge about whether the bridge is alive or not. To support this, we add `invokeAsync` impl to the RCTCxxBridge.
Reviewed By: JoshuaGross
Differential Revision: D14868547
fbshipit-source-id: 3b3f2b9150e930b4a2c71012242305241fc6dbed
Summary:
Main change is to the property diffing - we now use the last known props set on the view rather than the default props to compute the diff. This requires exposing a `getProps` method on all view components which should be fine I think.
I also realized that in more complex animations with multiple nodes, the node that the animation starts on might not be connected to a view, so we don't know if it's fabric just based on that, so we have to do a recursive search through the children to find if there are any that are associated with a fabric view to decide we should start the animation immediately. Unfortunately there can still be a timing gap here since the animated API is async and the uimanager API is sync - I'll need to change the animated API to be sync to completely fix this.
Reviewed By: shergin
Differential Revision: D14732028
fbshipit-source-id: 882c056b0b63aa576f8e42439be405cf7fb3147a
Summary:
Otherwise reloading from metro deadlocks like this:
`[RCTSurfaceRegistry _stopAllSurfaces]` is calling `[RCTSurfaceRegistry enumerateWithBlock]` which locks `_mutex` and then we call `surfaceForRootTag` which then deadlocks on the same mutex:
https://pxl.cl/tmm1
Reviewed By: shergin
Differential Revision: D14679843
fbshipit-source-id: 9f4ecdfa7a79fcf7f3fc2ce437d4399b00910f26
Summary:
If size is zero, the `CGContext` would invalid, so we need to filter zero things.
cc. shergin
[iOS] [Fixed] - Fix invalid CGContext when invalidate layer
Pull Request resolved: https://github.com/facebook/react-native/pull/24195
Differential Revision: D14683396
Pulled By: shergin
fbshipit-source-id: b5b45fb86ca3158284281460cf1d95d1b22eab0d
Summary:
We will use it inside `core` module, so we have to decouple it from `view`.
As part of this, I added some comments, changed `const Float &` to just `Float` and put the implementation into `.cpp` file.
Reviewed By: mdvacca
Differential Revision: D14593952
fbshipit-source-id: 80f7746f4fc5b95febc8df9f5a9c0386a6425c88
Summary:
Put `prepareForRecycle` to last before enqueue to recycle pool, ensure only call it when count lower than RCTComponentViewRegistryRecyclePoolMaxSize.
cc. shergin .
[General] [Changed] - Reorder prepareForRecycle before adding recycle pool
Pull Request resolved: https://github.com/facebook/react-native/pull/24025
Differential Revision: D14536843
Pulled By: shergin
fbshipit-source-id: 82a816e2c0afb5a6bb72637d7d55d6a4fda708af
Summary: Several things need to ironed out: 1) LocalState in Fabric C++, 2) setting dimensions of BottomSheet component to 0,0 for parent.
Reviewed By: shergin
Differential Revision: D14426167
fbshipit-source-id: 45a90a7971c87672872108a9e360926b4a6095f0
Summary:
This fixes a few memory leaks in fabrics handling of colors for borders, when using CGColorRef's we must be diligent about releasing the memory back.
[iOS] [Fixed] - Border style memory leaks
Pull Request resolved: https://github.com/facebook/react-native/pull/23815
Differential Revision: D14431250
Pulled By: shergin
fbshipit-source-id: dc663c633ae24809cb4841800d31a6ac6eeb8aa5
Summary:
I measured the performance of UIMananger methods and found that `create` method spends a significant amount of time on preallocating views (around 10 ms on my device). Interestingly, this time was actually spent on dispatching operations on the main thread, not on performing those operations. That makes me think about think about the preallocation problem one more time.
(Here and later I discuss preemptive optimistic preallocation views on iOS, but the issues are more-less applicable for all platforms.)
BTW, what's view preallocation? – In React Native we believe that we can get significant TTI wins instantiating platform views ahead of time while JavaScript thread is busy doing reconciliation and the main one is idle.
So, seems the current approach has those downsides:
* We spend too much time on dispatching only (jumping threads, creation Objective-C blocks, managing them in GCD, incrementing atomic refcountes inside `ShadowView` and so on). That's wasteful.
* We don't have much information during preallocations that can help prepare something to save time in the future. We don't know the future size of the component, so we cannot e.g. pre-draw a border or even allocate memory for a future drawing.
* We pre-allocate to much. At the moment of component creation, we don't know will this component be filtered out by view-flattening infra or not, so we always allocate a view for it. That's ~30% more views that we actually need.
* We don't stop allocating (or dispatching instructions for that) after the recycle pool is already filled in. That's also wasteful.
Why is this so bad and how can we fix it properly?..
I thought about this problem for months. And I think the answer is trivial: We don't know the exact shape of the interface until we finish creating it, and there is no way to overcome this problem on the client side. In the ideal world, the server size (or hardcoded manifest somewhere) tells the renderer (at the moment where we started navigating to it) how many views of which type will some surface use.
And until our world is not so perfect, I think we can get a significant performance improvement doing simple preallocation of 100 regular views, 30 text views, and 30 image views during rendering the first React Native surface. So I hardcoded that.
Reviewed By: JoshuaGross
Differential Revision: D14333606
fbshipit-source-id: 96beeb58b546258de1b8fd58550e0ae412b78aa9
Summary:
Right now we rely on the Paper UIManager to update animated node graphs - this hooks us into `RCTSurfacePresenter` in the same way so we are no longer reliant on Paper. Should also help with complex ordering corner cases with pre vs. post operations and restoring defaults when nodes are removed. More info:
https://github.com/facebook/react-native/pull/11819/files
Note that we don't have a way to differentiate animation nodes related to fabric views vs. paper views, so if paper and fabric are both rendering updates simultaneously it's possible they could get processed by the wrong callback. That should be very rare, rarely cause problems even if it does happen, and won't be a problem at all in a post-Paper world.
Reviewed By: shergin
Differential Revision: D14336760
fbshipit-source-id: 1c6a72fa67d5fedbaefb21cd4d7e5d75484f4fae
Summary: Use the codegen for the Slider component with the new `inferfaceOnly` option
Reviewed By: TheSavior
Differential Revision: D14295981
fbshipit-source-id: 0482572892fbcffada43c7c6fbf17e70546300b8
Summary: This diff removes the "isLayoutable" parameter from SchedulerDelegate.schedulerDidRequestPreliminaryViewAllocation. This now can be infered from the shadowView parameter
Reviewed By: shergin
Differential Revision: D14296481
fbshipit-source-id: b200504f9c2bef41f0a70257f1f5a274fbe97cbb
Summary:
Update Props during pre-allocation avoiding re-updating the same props during mounting
MobileLab test showed an improvement of:
MARKETPLACE_YOU_TTI_SUCCESS: -3.34% = 58ms
Reviewed By: shergin
Differential Revision: D14252170
fbshipit-source-id: 1f4e9ad5dcecbc06651fa065135ffeed4892d984
Summary:
We are now generating the native cpp files for ActivityIndicatorView via Buck.
Deleting the hand written files and switching over.
Reviewed By: TheSavior
Differential Revision: D14247446
fbshipit-source-id: 63a6df3254e4184de6c8abb9ea2c89654ad54398
Summary:
This is a back-out of D14214844, we noticed that this regressed TTI for Marketplace You screen running in Fabric
Original commit changeset: b81005f2bf49
Reviewed By: JoshuaGross
Differential Revision: D14247897
fbshipit-source-id: de0cea92b437b2fbcd075f0d6a0066156800e3f0
Summary: The changes allows to get the State object on mounting layer and initiate the updates.
Reviewed By: mdvacca
Differential Revision: D14217185
fbshipit-source-id: 370644e06e350932e93c39adbe46544b071c28b3
Summary: That's bummer that we have to do it, but it's actually reasonable. Files in `core` and `events` depend on each other creating circular dependencies and other similar hard problem.
Reviewed By: mdvacca
Differential Revision: D14195022
fbshipit-source-id: 96a44ae28631cc9ccd7d7de72a94526f9e0dd12a
Summary:
This diff changes the pre-allocation of Images update the props on the ImageViews during the creation of ShadowNodes instead of during rendering.
The purpose of this change is to optimize TTI.
Reviewed By: shergin
Differential Revision: D14214844
fbshipit-source-id: b81005f2bf494f62f421dc24846e1561e13b9a87
Summary:
Not super clean, but not terrible.
Unfortunately this still relies on the old Paper UIManager calling delegate methods to flush the operations queues. This will work for Marketplace You since Paper will be active, but we need to fix this, along with Animated Events which don't work at all yet.
Random aside: it seems like taps are less responsive in fabric vs. paper, at least on iOS. There is a sporadic delay between the touches event coming in nativly to the JS callback invoking the native module function to start the animation - this will need some debugging.
Reviewed By: shergin
Differential Revision: D14143331
fbshipit-source-id: 63a17eaafa1217d77a532a2716d9f886a96fae59
Summary:
CALayer will crash if we pass NaN or Inf values.
It's unclear how to detect this case on cross-platform manner holistically, so we have to do it on the mounting layer as well.
NaN/Inf is a kinda valid result of some math operations. Even if we can (and should) detect (and report early) incorrect (NaN and Inf) values which come from JavaScript side, we sometimes cannot backtrace the sources of a calculation that produced an incorrect/useless result.
Besides that, I will investigate why the crash is actually happening, so we might need to fix something in layout engine. But, it general, we cannot capture all errors like that, so we need to have it here anyway.
Reviewed By: JoshuaGross, mmmulani
Differential Revision: D14126058
fbshipit-source-id: 807e5a223bdef48af9a3b7210803863431e8c507
Summary: Use the new copyright header format used elsewhere in the React Native repository.
Reviewed By: shergin
Differential Revision: D14091706
fbshipit-source-id: b27b8e6bcdf2f3d9402886dbc6b68c305150b7d5
Summary:
Sometimes, when we deal with ImageRequest and ImageResponseObserverCoordinator we subscribe for status (or access the coordinator) without owning an ImageRequest. In those cases, we have to retain the coordinator explicitly.
For those cases, ImageRequest now exposes `ImageResponseObserverCoordinator` as a `std::shared_ptr`.
Eg, concretely in the code, `completionBlock` and `progressBlock` copied a raw pointer to the observer inside which can lead to a crash when ImageRequest is being deallocated before we received an image data.
Reviewed By: JoshuaGross
Differential Revision: D14072079
fbshipit-source-id: e10120bc05bf685e288f7b3d69092714dcd91d43
Summary: Apparently, I haven't modify all files in D14018903. Here is the last (I hope) chunk.
Reviewed By: JoshuaGross
Differential Revision: D14058288
fbshipit-source-id: b21950cdd1aa9aa55b0c72fac0f8b44e4a7d131c
Summary:
Trivial.
If you have troubles with rebasing on top of this revision, run this on your diff:
$ find */*.h */*.mm */*.cpp */*.m -exec clang-format -style=file -i {} \;
Reviewed By: JoshuaGross
Differential Revision: D14018903
fbshipit-source-id: fd0ce2da0e11954e683385402738c701045e727c
Summary:
There is no reason to allocate views ahead of time on the main thread.
There is a chance that this view will not be mounted and we are not saving any time because it's a sequential process anyway (because we are doing it on the main thread). Moreover, the switching context can only slowdown JS execution.
Reviewed By: JoshuaGross
Differential Revision: D14019433
fbshipit-source-id: 83ac05a91e4b70cb382a55d6687752480984404e
Summary: Recycling and dealloc were not implemented at all before for Slider, so I've taken a first stab at it. It's a little more complex than I initially thought, due to things I don't 100% understand about UISlider as well as Fabric, so I've left a TODO note to fix this at some point. We should be aware that view recycling doesn't appear to be working the way I would expect currently though.
Reviewed By: shergin
Differential Revision: D13965475
fbshipit-source-id: fd18a219cead770b63b514fdc868e23214e735b7
Summary: Don't use shared_ptr in this case, it's not needed.
Reviewed By: shergin
Differential Revision: D13965413
fbshipit-source-id: ec98c13f53c7d558a0cb68cea0f97568dd202cd8
Summary: The biggest change is that (1) the image proxy/observer code from the Image component has been generalized, (2) the four image props for the Slider component are fully supported, (3) a handful of props that were ignored or buggy on iOS now perform as expected.
Reviewed By: shergin
Differential Revision: D13954892
fbshipit-source-id: bec8ad3407c39a1cb186d9541a73b509dccc92ce
Summary:
This diff adds performance loggers for Fabric in Android to be able to compare current version or RN with Fabric
This is the summary of Points and Annotations:
- **UIManager_CommitStart**: time that React starts the commit (react tree is ready to start rendering in native)
- **UIManager_LayoutTime**: this is the time it takes to calculate layout in yoga
- **UIManager_FabricFinishTransactionTime**: Time it takes transform "C++ mutationInstructions" into "Java MountItems" and cross boundaries from C++ to Java (including serialization of data) (THIS IS ONLY FABRIC)
- **UIManager_DispatchViewUpdates**: time right before RN moves the mount operations to the Queue that is going to be processed in the next tick UI thread
- **UIManager_BatchRunStart**: time right before the mountItems are going to be process in the UI Thread
- **UIManager_BatchedExecutionTime**: time it took to run batched mountItems (usually layout and prop updates on views)
- **UIManager_NonBatchedExecutionTime**: time it took to run non-batched mountItems (usually creation of views)
Reviewed By: fkgozali
Differential Revision: D13838337
fbshipit-source-id: 0a707619829e7d95ce94d9305ff434d1224afc46