Summary: this diff exposes the rawProps folly::dynamic map received by JS as part of the Props class
Reviewed By: shergin
Differential Revision: D9093559
fbshipit-source-id: 5f5bc4924aebb6bcc24c7a82ce1a59593d44450e
Summary:
@public
Now it accepts `const ShadowNode &` instead of `std::shared_ptr<const ShadowNode>` which is more reasonable (and more performant) becasue the function must not retain ownershipt.
Reviewed By: mdvacca
Differential Revision: D9073921
fbshipit-source-id: c24c475615e0f81b3e004e118dea7565d8e757b4
Summary:
@public
To avoid unnecessary copying of `shared_ptr`s inside ShadowNodeFragment, now we store them as `const &` references.
Reviewed By: mdvacca
Differential Revision: D8988388
fbshipit-source-id: 0b3582e57ce7577b8fa819392bf33f34e1a60b59
Summary:
@public
Now we use same data structure to specify a shape of shadow node as we use in ShadowNode (sub)clases.
Reviewed By: mdvacca
Differential Revision: D8988387
fbshipit-source-id: 475298b2c71ee7ee2b197db009f7b8313b54f5df
Summary:
@public
When we copy-construct ShadowNode, we don't need to retain a source shadow node, so there is no need to pass it as a `shared_ptr`. Passing an argument to constructor as `const &` is also more idiomatic in C++.
Reviewed By: mdvacca
Differential Revision: D8988384
fbshipit-source-id: 1279d9185fa1b4b82fd26e3040bd62fa9495b4d3
Summary:
@public
ShadowNode class is designed to share `props` and `children` objects between instances. Given that all *Props classes are immutable, it's very easy to share them and we do this from the day one. The `children_` collection is more tricky though because ShadowNode class has a couple of mutation methods. Previously, we dealt with it very simply by copying the whole vector in constructors, and that was far from optimal. Now we store a special flag that indicates that the children list is shared among nodes, and we clone this before the first mutation.
Sharing a `shared_ptr` should be much more efficient (cost of atomic refcount increment) than instantiating whole new collection (an allocation).
Reviewed By: mdvacca
Differential Revision: D8988386
fbshipit-source-id: cb2f6b2fccac70a35e070a1aa108d135f77cd041
Summary:
@public
This diff changes a way how we specify a shape of newly created and/or cloned of ShadowNode. Previously we pass those values as a list of arguments, now those values are coupled into a new data structure called ShadowNodeFragment. All that makes suppose to make code much more easy to read and maintain, this is especially important because we want to add a couple of new entities in this set.
Reviewed By: mdvacca
Differential Revision: D8988389
fbshipit-source-id: 1835f646e1ecc6a1f413feaf1900f3d3ad0ebc05
Summary:
@public
Previously, all ConcreteShadowNode subclasses had to override `getComponentName()` function to specialize a name of the component. And often it was all that those subclasses do. Now, it's a template argument; and many ShadowNode classes can be created as oneliners via *just* specializing ConcreteShadowNode template.
Unfortunately, C++ does not allow to use `std::string`s or string literals as template arguments, but it allows to use pointers. Moreover, those pointers must point to some linked data, hence, those values must be declared in .cpp (not .h) files. For simplicity, we put those constants in Props classes, (but this is not a strong requirement).
Reviewed By: mdvacca
Differential Revision: D8942826
fbshipit-source-id: 4fd517e2485eb8f8c20a51df9b3496941856d8a5
Summary:
@public
Now, Sealable is already error-preventing-only mechanism, no business logic relies on that.
So, it makes sense to make it debug-only to illuminate possible performance impact.
Reviewed By: mdvacca
Differential Revision: D8923597
fbshipit-source-id: 80aa9097c4b719e91de73ac59f38d3a4751f0b06
Summary:
@public
We need that because gonna add much more event-related stuff, so it deserves separate buck target.
Reviewed By: mdvacca
Differential Revision: D8831547
fbshipit-source-id: 616581b39b425a49302d5f7f86267e62b0d58389
Summary:
@public
We don't need to maintain an order of this collection, so using `unordered_map` is more appropriate.
Reviewed By: mdvacca
Differential Revision: D8826946
fbshipit-source-id: f6890097cc5d6a1e06f6b2cfd1b7d68a388da461
Summary:
@public
In most cases callsite knows probable index of replacing child node, hence it makes sense to provide this info to `replaceChild` to illuminate O(n) search in most cases.
Reviewed By: mdvacca
Differential Revision: D8814809
fbshipit-source-id: 0edf82878a72260365e2757beb3886ad07c7464d
Summary:
@public
This diff consists of many interdependent changes which support one simple idea: YogaLayoutableShadowNode is now using YGNode children to iterate on them (it previously relied on `ShadowNode::getChildren()`). All other changes are just an unavoidable consequence of that. Hence we don't need to filter child nodes every single time when we do layout anymore! The logic around `clone callback` is also drastically simpler now.
The new approach also implies that `LayoutableShadowNode` and `YogaLayoutableShadowNode` don't use `shared_ptr`s to refer to ShadowNode objects because new relationship does not imply ownership. No more `SharedShadowNode` objects in those two classes.
Reviewed By: mdvacca
Differential Revision: D8796159
fbshipit-source-id: 6f52f92d1826f3eb13b2f8a132c3ea77de155d82
Summary:
Removes the concept of instance handle. Instead we pass the event target
to createNode and don't pass it to subsequent clones.
The life time of the event target is managed by native (the event emitter).
It has to be released manually.
Reviewed By: shergin
Differential Revision: D8688330
fbshipit-source-id: e11b61f147ea9ca4dfb453fe07063ed06f24b7ac
Summary:
@public
Most of them are legit issues which should not be compilable anyways (but Clang tolerates thems).
Reviewed By: mdvacca
Differential Revision: D8655539
fbshipit-source-id: 645729fb9d6a120ce1ab2b07542abcdacd72320d
Summary:
@public
Suddenly, it is not supported on Android.
Luckelly `folly:to<std::string>()` is as good as `std::to_string()`.
Reviewed By: mdvacca
Differential Revision: D8655538
fbshipit-source-id: 2b3b970f6a261253aaa6b22dba8338dc66b7195d
Summary:
@public
Otherwise, it can mess with implementation for `int`s and causes some errors where `float` implementation was requested but `int` was applied.
Reviewed By: mdvacca
Differential Revision: D8601752
fbshipit-source-id: cfe51b7785ff29ee4ad88f0f1cbfed335557d5ef
Summary:
@public
If some prop has `std::vector` type, it possible that on JS side we want to pass just one element of the array.
And in this case we sometimes drop array initialization (`[]`) part, so instead of passing `[{x:1, y:1}]` we pass `{x:1, y:1}`.
This diff adds support for that.
Reviewed By: mdvacca
Differential Revision: D8526572
fbshipit-source-id: 33d4369ac48cac3eb1c534f477d8259e76e0c547
Summary:
@public
After reading about move-semantic and rvalue refs I realized that we (I) definitely overuse `auto &&` (aka universal reference) construction. Even if this is harmless, does not look good and idiomatic.
Whenever I used that from a semantical point of view I always meant "I need an alias for this" which is actually "read-only reference" which is `const auto &`.
This is also fit good to our policy where "everything is const (immutable) by default".
Hence I change that to how it should be.
Reviewed By: fkgozali
Differential Revision: D8475637
fbshipit-source-id: 0a691ededa0e798db8ffa053bff0f400913ab7b8
Summary:
@public
... and we initalize this in Surface.
We need this for requesting images with proper size/pixel-density, setup proper parameters for rasterizing CALayer's and rounding layout metric values.
Then we have to figure out how to wire this up with YGConfig.
Reviewed By: fkgozali
Differential Revision: D8475639
fbshipit-source-id: cec7af581b94efb4595dcf3f232252ce87a1fde3
Summary:
@public
There are some race conditions between VM objects getting deallocated and the instanceHandle held by the eventEmitter can point to deallocated memory space, causing undefined behavior like a crash.
For now, keep a strong ref to the eventTarget inside EventEmitter to avoid that scenario. This is a temporary workaround.
Reviewed By: shergin
Differential Revision: D8576785
fbshipit-source-id: 87ef36f716270ceca906b32bb86e0046ceaca19e
Summary:
Now, if `fromDynamic` is defined for some type, `fromDynamic` for `std::vector` of this type is also will be defined.
We need this for parsing `ImageSources` (a vector of `ImageSource`) type.
Reviewed By: fkgozali
Differential Revision: D8473508
fbshipit-source-id: d8dc8e3a3273f35b76c7132c553130762f768394
Summary: Calling the event emitters on the main thread seems to be problematic, so let's dispatch it via the JS thread. This requires some changes to make "eventTarget" single-use because otherwise the binding would need to synchronize the actual JS call with the act of releasing the target.
Reviewed By: shergin
Differential Revision: D8375291
fbshipit-source-id: bd2b42731176ae209f4a19c232309c163fb1c01b
Summary:
* numbers in JS are doubles in native land, since there's no notion of int or int64 in JS - so simply convert numbers to int instead of assuming it's int
* the parsing of Yoga props with `'...%'` string value has a bug: it should be copying the number instead of the `%`
Reviewed By: shergin
Differential Revision: D8370873
fbshipit-source-id: 44e9e3f0530c000c963e8e9ca66e8b0a48d80bcd
Summary:
Using `EventHandlers` name was a bad idea, and I cannot tolerate it anymore.
The worst part of it is that when you have a collection of `EventHandlers` objects you cannot use plural word to describe it because `EventHandlers` is an already plural word.
And, this object is actually an event emitter, the thing on which we call events.
Reviewed By: fkgozali
Differential Revision: D8247723
fbshipit-source-id: b3303a4b9529bd6d32bb8ca0378287ebefaedda8
Summary:
During transforming raw prop (`rawProps[<prop>]`) to typed props (`MyProps.<prop>`) we can face three different cases:
* `rawProps` collection has proper serialized value for the key. In this case, we have to set a new value of the typed prop converting value using `fromDynamic`.
* `rawProps` collection does not have value for the key. In this case, we have to copy a value from source prop (`sourceValue`).
* `rawProps` collection has `null` value for the key. This is the special case which means that the prop was removed from the particular component instance and we have to reset it to some *default* value (which is *not* the same as `sourceValue`). Now the default value of the `defaultValue` (sic!) argument is a default value of the type of the value (which may be different from logical default value).
We didn't handle the last case previously and this caused crashes (and unexpected behavior) because `fromDynamic` often cannot handle `null` value.
And yes, all this mean that we also have to update all `convertRawProp` call sites where logical default values are not equal to type-specific default values. This is a potential error-prone place, especially because now we have to specify logical default values in two places (in a prop declaration and in a parameterized constructor). And seems there is no way to avoid that without performance loss (because both of those places are basically constructors).
My hope is that codegen (where default values are also defined in JavaScript) will help with it eventually.
Reviewed By: fkgozali
Differential Revision: D8247652
fbshipit-source-id: 2cbe65f5f5cccd7a0d34aaa19e385aacebfe8cb1
Summary:
`LayoutableShadowNode.cpp` includes `"LayoutableShadowNode.h"` as well as `<fabric/core/LayoutContext.h>`. In turn, `LayoutContext.h` then includes `<fabric/core/LayoutableShadowNode.h>`. `LayoutContext.h` doesn't actually require `LayoutableShadowNode.h`, but this unnecessary inclusion can cause duplicate definition errors if the two include paths don't map to exactly the same file. This patch removes the unnecessary include.
The CI's build system should cover the testing needed.
[INTERNAL] [MINOR] [fabric] - Remove an unnecessary include in fabric/core/layout.
Closes https://github.com/facebook/react-native/pull/19548
Differential Revision: D8313337
Pulled By: shergin
fbshipit-source-id: 2e01e29ff25131543d9a8601483c2e716c7437be
Summary:
This is the first attempt to implement some base part of event dispatching pipeline from end-to-end.
Even when it is working, all this is still incomplete and generally up in the air. We are still messing proper implementation of event queue, priority, and synchronization of react reconciliation process with event scheduling.
Reviewed By: fkgozali
Differential Revision: D8212271
fbshipit-source-id: 92f9427d14726441c70ffff294ac95eeb004152a
Summary:
In order to dispatch event, `EventHandlers` must also know react tag. So we have to store it inside.
We plan to illuminate this requirement (and `tag` from `EventHandlers`) eventually.
Reviewed By: fkgozali
Differential Revision: D8211685
fbshipit-source-id: 2064c0f4a7869cbf4d2c92d0349f4ee3998cb8f5
Summary:
Nothing actually changed besides type names... which actually helps me found an issue in FabricUIManager!
Now there is no a single `void *` in Fabric/C++ and JavaScript bindings. Yay!
Reviewed By: fkgozali
Differential Revision: D8191420
fbshipit-source-id: b1eb60b6bc34dd25ab200aab854ffbd7ccf5b15d
Summary:
It's maybe not so important/crucial, but this thing bothers me a lot.
We use raw opaque `EventTarget`, `InstanceHandle` and `EventHandler` pointers in application layer quite a lot and we don't have any kind of type-safety here. I believe all those opaque types should be represented as named scalar types which compiler at least can differentiate at compile time.
So I propose introducing named aliases for them which will point to particular empty `struct`s. This will allow us to tag types properly in all functions and methods and ensure that we pass right values as right arguments.
Again, they are *just aliases*, which are effectively still `void *`, no any additional logic or names are involved.
Unfortunately, those nice type names are already taken by `JSIFabricUIManager` local anonymous namespace (even if they are inside anonymous namespace we cannot use them https://stackoverflow.com/questions/3673353/anonymous-namespace-ambiguity). I think it's fair to rename them because... it's local. And we already use `Wrapper` suffix for them anyways.
Reviewed By: fkgozali
Differential Revision: D8181151
fbshipit-source-id: 9b55b43fb671a56b32a862ac54f78d528e1188ce
Summary: This implements `EventHandlers` abstract class (aka "Events Guy") which encapsulates `eventDispatcher` and `instanceHandle` (and ownership of future `eventTarget`), all of this as part of existing {ShadowNode + Props + LayoutMetrics + LocalData + Descriptor + (and now) EventHandlers} infra. (We don't plan to add anything else to this model. Ever.)
Reviewed By: fkgozali
Differential Revision: D8053351
fbshipit-source-id: 1dd9ccbcbe5a2eb284b59ea351dc8beca645e8bf
Summary:
Now ConcreteComponentDescriptor can infer `ComponentName` from `ShadowNodeT` automatically,
so in the most cases we even don't need to create a subclass of that.
Reviewed By: fkgozali
Differential Revision: D8016965
fbshipit-source-id: d910597093c9f4c9f32ca06cb3ef1b12538b9543
Summary: We don't use them at all; moreover they complicate adding/changing signatures of those methods (because arguments with defaults must be grouped at the end and some arguments cannot have defaults).
Reviewed By: fkgozali
Differential Revision: D7981456
fbshipit-source-id: d7dd098e83630d1ab3342d2ca52ade9c4e27b2c3
Summary:
A few fixes:
* missing include: folly/Optional.h
* switch folly::Optional's `has_value()` to `hasValue()` for now until folly is upgraded to newer version
* fix up import for RCTTextAttributes.h
* fix up includes for "conversions.h" to use namespaced includes
Reviewed By: mmmulani
Differential Revision: D8021149
fbshipit-source-id: d3955986d3ab6b1d9b61ac1e385767893ce57e5e
Summary: Oh, my! No more `#define`s related to props conversions and debug-printing.
Reviewed By: fkgozali
Differential Revision: D7958250
fbshipit-source-id: 86950070c55f134aa3a575b9fd68fc90d865cf44
Summary:
Same as previous one.
Adopting template-generated `convertRawProp` and `debugStringConvertibleItem` functions in `core` module.
Note, to do so we have to change signatures of some conversions functions to make them more overloading-friendly.
Reviewed By: fkgozali
Differential Revision: D7958243
fbshipit-source-id: 500ee420d9aa562ee3c5810ef625e06541eda8fb
Summary:
Same as previous one.
Adopting template-generated `convertRawProp` and `debugStringConvertibleItem` functions in `graphics` module.
Note, to do so we have to change signatures of some conversions functions to make them more overloading-friendly.
Reviewed By: fkgozali
Differential Revision: D7958252
fbshipit-source-id: 0f33a2e6aad60befacee31486acdb9b6114d3e07
Summary:
We have to have automatic treatment for `optional` types. So, if we can process type `T` we can also automatically process `optional<T>.`
Support for optional allows us to not introduce new types (with embedded special "undefined" value) or pollute existing pure types (with special "undefined" value). (A lot of examples of those types can be found in AttributedString module.)
Reviewed By: fkgozali
Differential Revision: D7958249
fbshipit-source-id: 21af526a17dd0329e1262020cab8ecb902316654
Summary:
This diff opens a diffstack where we migrate the generation of all prop conversions (convertRawProp) and pretty-printing (debugStringConvertibleItem) functions to C++ templates (instead of using `#define`s).
So, this diff implements base versions of those functions as templated functions.
For now we still need #define-based version, but eventually, we will get rid of it.
Reviewed By: fkgozali
Differential Revision: D7958247
fbshipit-source-id: 24346297c1bd17e8054758f0eb84698eebfa21e2
Summary:
This diff contains several tight to each other changes (which can/should not be split into several diffs):
* The props parsing/conversion process was de-virtualized: we don't use virtual `apply` method to parse props anymore. Instead, we use old-fashioned constructors.
* All fields of Props classes which represent props values were marked as `const` which make impossible to modify them after the objects were created (even if we have non-const value-of/pointer-to the whole Props object). Those fields are also `public` now.
* All custom handwritten getters were removed (because we don't need them anymore).
So, now we don't need all those custom getters which makes code much more compact, performant and codegen-friendly.
Reviewed By: fkgozali
Differential Revision: D7901245
fbshipit-source-id: 9f4b1fd2da64bf963b63215ed3bd74b9d3c58dd5
David Vacca
Valentin Shergin
Taras Tsugrii
Sebastian Markbage
Kevin Gozali
Robert Sayre