Summary:
@public
Removes the style properties bitmask. We have used this for experimentation, and it's no longer necessary.
This simplifyies the code, and allows us to cut over to `Bitfield.h` more easily.
Reviewed By: astreet
Differential Revision: D16648862
fbshipit-source-id: 17c0899807af976f4ba34db54f8f0f6a3cd92519
Summary:
@public
Our usage of C++ bit fields has lead to quite some problems with different compiler setups. Problems include sign bits, alignment, etc.
Here we introduce a portable implementation as a variadic template, allowing the user to store a number of booleans and enums (defined with `YG_ENUM_SEQ_DECL`) in an unsigned integer type of their choice.
This will replace all usages of bit fields across the Yoga code base.
Differential Revision: D16647801
fbshipit-source-id: 230ffab500885a3ad662ea8f19e35a5e9357a563
Summary: Adds internal API that we can use to conduct experiments.
Reviewed By: SidharthGuglani
Differential Revision: D16340463
fbshipit-source-id: 07a8bb7dbc4a02c5c95f1ad29b18845ab43752cf
Summary: Add tests for measure and baseline events , same as we had in InstrumentationTests for marker based approach
Reviewed By: davidaurelio
Differential Revision: D16074402
fbshipit-source-id: 8b11cd6468420428701fd5be5c57700cbd913d23
Summary: Adds Baseline start and end events to be handled later for instrumentation
Reviewed By: davidaurelio
Differential Revision: D16048790
fbshipit-source-id: 8409dbb633168753a7bf8fab20bc6551d113ddd6
Summary: Using layoutPassStart and LayoutPassEnd events instead of YGMarkerLayout for instrumentation
Reviewed By: davidaurelio
Differential Revision: D16048789
fbshipit-source-id: 041a35bc2cb1b7281ca83cf9d35041b4011cfeb9
Summary: Add tests for layout pass start and end , same as what we had in InstrumentationTests for marker based approach
Reviewed By: davidaurelio
Differential Revision: D16073121
fbshipit-source-id: 838f01cb2a41d2d2764ba7ce2f317147f737b287
Summary: We are now using custom TypedEventTestData for test cases as we need to copy some data passed from the yoga event system
Reviewed By: davidaurelio
Differential Revision: D16090931
fbshipit-source-id: 4d11bdbdd73b67172ad4bba4b294c71f1c24cc10
Summary:
GCC 8.3.0 (and possibly all gcc 7+) identified several warnings for signed unsigned integer comparison. With `-Werror` enabled this broke compiling tests. I suspect the warning is related to google/googletest#683. This diff updates those `ASSERT_EQ` calls that attempt to compare signed and unsigned errors by specifically declaring the literals to be unsigned.
There is also an issue with Buck where it will not link to pthreads. facebook/buck#1443. Adding a `prebuilt_cxx_library` for pthread fixes that issue and the tests will compile and run.
Finally, there was a warning about a missing return after a switch in `InstrumentationTest.cpp`. I added a `return ""` as a default, but it might be better to throw something. Thoughts?
Pull Request resolved: https://github.com/facebook/yoga/pull/895
Reviewed By: davidaurelio
Differential Revision: D15393082
Pulled By: davidaurelio
fbshipit-source-id: 4f13ec2f016af39537c08fb591b188a6a0ed55ce
Summary: Publishing two events will allow us to replace marker functionality completely with events. This also allows us to remove measuring time spent from Yoga itself.
Reviewed By: SidharthGuglani
Differential Revision: D16049810
fbshipit-source-id: 98628a92ed3c94d479e9fbcd53fac90c5f524087
Summary:
Added event NodeLayoutEnd and this is being used now instead of NodeLayout
It will be used later to add more information about caches
Reviewed By: davidaurelio
Differential Revision: D15920935
fbshipit-source-id: c9f5e193bc8cc70d26ff5d84882d483c9b09f67d
Summary:
`/*` is the standard throughout open source code. For example, Firefox uses single /*: https://hg.mozilla.org/mozilla-central/file/21d22b2f541258d3d1cf96c7ba5ad73e96e616b5/gfx/ipc/CompositorWidgetVsyncObserver.cpp#l3
In addition, Rust considers `/**` to be a doc comment (similar to Javadoc) and having such a comment at the beginning of the file causes `rustc` to barf.
Note that some JavaScript tooling requires `/**`. This is OK since JavaScript files were not covered by the linter in the first place, but it would be good to have that tooling fixed too.
Reviewed By: zertosh
Differential Revision: D15640366
fbshipit-source-id: b4ed4599071516364d6109720750d6a43304c089
Summary: Style bits had overlap, because `dimensionBit` was set with an incorrect increment.
Reviewed By: SidharthGuglani
Differential Revision: D15335134
fbshipit-source-id: 370e1a73547d76b0e26bc6ab67acb96d33ddf180
Summary:
Adds `LayoutPassStart` and `LayoutPassEnd` events.
The existing `NodeLayout` event in isolation is not as useful as it could be. Having events that mark start and end of a layout pass are a useful addition.
Differential Revision: D15305467
fbshipit-source-id: 14af6f65e698fb1e3112eb2ffd87a74d31df4840
Summary:
@public
replaces the global node counter with the event-based one for all tests.
Reviewed By: SidharthGuglani
Differential Revision: D15174856
fbshipit-source-id: f4401d502bdbaf3b6e4632a4d985aac260cb35a8
Summary:
@public
Test utility on top of the new event system that maintains a counter of instantiated nodes. Meant to replace the global node counter.
Reviewed By: SidharthGuglani
Differential Revision: D15174855
fbshipit-source-id: 6998472f95a09b8da652257a26596164bdcf43d6
Summary:
@public
Publish two events, `NodeAllocation` and `NodeDeallocation`, in the same places where the global node counter is changed.
Reviewed By: SidharthGuglani
Differential Revision: D15174858
fbshipit-source-id: 6e4e9add88513b9e987189ca5035d76da2a1de55
Summary:
@public
In order to optimise property storage, we have to know how style properties are used in our apps.
Here, we add a bitmask that allows us to track which properties are set explicitely, and use that for our analysis.
Reviewed By: SidharthGuglani
Differential Revision: D14933022
fbshipit-source-id: 1ab8af562b14baba1d02057e527aa36d5c9a7823
Summary:
@public
Change style property accessors to return `Ref` instances instead of references to `CompactValue`.
This will allow to track assignments to properties later on, e.g. for instrumentation or dynamic property storage.
Reviewed By: SidharthGuglani
Differential Revision: D15078961
fbshipit-source-id: 259f05f7d30f093c04bf333c5bd4fb3601b8e933
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:
@public
This allows short methods defined in class declarations to occupy a single line.
The change makes class declarations more readable.
Reviewed By: SidharthGuglani
Differential Revision: D14950012
fbshipit-source-id: 1321949475184181c6cceb86613f730e430763e2
Summary:
@public
Regenerating the “golden master” tests with chrome surfaced different bugs around `align-content`:
- a misunderstanding that values in `align-content` only applied *if there is only one line.* In fact, it applies *every time* a container is set to `flex-wrap: wrap`. Chrome had this wrong, and as such our tests were generated with incorrect parameters.
- empty children growing to the cross axis size of the container, even when `align-content` is different from `stretch`. This was implemented incorrectly in Chrome as well. Here, we fix it with an extra check.
Reviewed By: SidharthGuglani
Differential Revision: D14725402
fbshipit-source-id: a45bebdadb9c694dc0eb7e27cb52b3d247f81c50
Summary:
@public
We use Chrome for generating test cases, which since v67 (or so) for `min-width` and `min-height` has a default value of either `0px` (CSS 2) or `auto` (CSS 3), depending on style properties.
Our setup only allowed for a single default value, and our test cases produce *both.*
This changes the test gen logic to allow for more than one value.
Reviewed By: SidharthGuglani
Differential Revision: D14682387
fbshipit-source-id: e76361f5cc0b88f9c2d74a5f3248c66abd6907a7
Summary:
@public
`setBaseLine` was the only place where we used a capitalised *L.* Fixed here.
Reviewed By: SidharthGuglani
Differential Revision: D14152320
fbshipit-source-id: abf54fe7d6088e03775968baa8421c4bf43d6a6e
Summary:
@public
Context-aware cloning functions are an internal Yoga feature that will be used for Yoga’s JNI code.
It will be possible to specify a context when calculating layout, which will be passed on to cloning functions. This will be a private feature.
Reviewed By: SidharthGuglani
Differential Revision: D14149470
fbshipit-source-id: 1d11106e65f9d872d10f191763da001f8d158a61
Summary:
@public
Limit child cloning to layout calculation. This also allows for mixing shared and owned children.
Rationale:
We do allow for shared children if the caller manages themselves. The single known use case is React Native.
So far, we have cloned children eagerly whenever child lists are mutated, or layout is run. This was to allow for a quick check of the owner of any first child, assuming that either *all* or *no* child of a node are shared.
For Yoga/Java, we want to get rid of global weak JNI refs, and these are also used to invoke clone callbacks. We can achieve that goal by switching to an alternative approach, passing additional data to the layout pass. This additional data has to be passed to any configured cloning callback. Therefore, it is desirable to **only call cloning functions during the layout pass.**
The obvious solution seems to be to not uphold the invariant of the first child determining shared/owned state of all siblings, and allow for a mix of shared and own children.
Reviewed By: shergin
Differential Revision: D14136223
fbshipit-source-id: 34490cfeeb2170c99d6ed1b9bdcbcedb316813af
Summary:
@public
Encapsulates node cloning within `YGConfig`.
This is necessary for allowing for context-aware cloning functions, which will ultimately allow for removal of weak global JNI references.
Reviewed By: shergin
Differential Revision: D14132608
fbshipit-source-id: 0dec114c8e172b1e34a4b7fd146c43f13c151ade
Summary:
@public
Context-aware measure and baseline functions are an internal Yoga feature that will be used for Yoga’s JNI code.
It will be possible to specify a context when calculating layout, which will be passed on to baseline and measure functions. This will be a private feature.
Reviewed By: SidharthGuglani
Differential Revision: D14100509
fbshipit-source-id: acf4a030549b2e38d5ce0cd5dbe837864e5ffd81
Summary:
@public
Stricter encapsulation of baseline and measure callbacks withing `YGNode`.
Instead of invoking these callbacks directly (`node->getBaseline()(...)`), they are invoked via methods on `YGNode` (`node->baseline(...)`).
This change will allow us to add the concept of a *Layout Context,* where measure and baseline functions will be able to receive an additional `void *` argument if configured accordingly. This API will be used internally for Yoga’s JNI bindings, to avoid storing a weak JNI reference for each node, and avoid reference table overflows.
Changed API:
- `YGNodeGetMeasureFunc()` -> `YGNodeHasMeasureFunc()`
- `YGNodeGetBaselineFunc()` -> `YGNodeHasBaselineFunc()`
- `YGNode::getMeasure()` -> `YGNode::hasMeasureFunc()` + `YGNode::measure()`
- `YGNpde::getBaseline()` -> `YGNode::hasBaselineFunc()` + `YGNode::baseline()`
Reviewed By: SidharthGuglani
Differential Revision: D14099550
fbshipit-source-id: 2653ab36acc252a9747986bc88d21dac22d8c91b
Summary:
@public
Adds `_pt` and `_percent` user defined literals to create `YGValue` instances.
This allows to create `YGValue`s in the following form:
```
use namespace facebook::yoga::literals;
auto a = 123_pt; // == YGValue{123.0f, YGUnitPoint}
auto b = -12.5_percent; // == YGValue{-12.5f, YGUnitPercent}
```
Reviewed By: SidharthGuglani
Differential Revision: D13942100
fbshipit-source-id: ce1e2f9431c3e2a99c6e11896a712539cc535e0d
Summary:
@public
Extends `InstrumentationTest` with the ability to record more than one marker.
Instead of overwriting any previously recorder marker, we collect them in a `std::deque`.
`std::deque` was chosen, because it keeps pointers to elements stable.
Reviewed By: SidharthGuglani
Differential Revision: D13880777
fbshipit-source-id: f6f6606ac8625bdd90c06459fdb967c3478265e7
Summary:
@public
Adds the maximum number of measure cache entries in use to the metrics for `YGMarkerLayout`
Reviewed By: SidharthGuglani
Differential Revision: D13844731
fbshipit-source-id: fa66dbf1b7a1799494f72ecc17dfaef04d0b56e4
Summary:
@public
Evolves setup and helpers in `InstrumentationTest` to avoid repetition across test cases, and allow for more readable test by hiding default values where they don’t matter.
Reviewed By: SidharthGuglani
Differential Revision: D13839521
fbshipit-source-id: 7f7ad49fec84e0bbb09ad746dd8c28bd34de25b2
Summary:
@public
Adds marker meta data to `YGMarkerLayout`: the number of measures, and the numbers of layouts for all nodes in the tree.
Reviewed By: SidharthGuglani
Differential Revision: D13838975
fbshipit-source-id: d575c26a3d5a4f0b66834eb6bedecadc3f3ca265
Summary:
@public
Adds a class for triggering markers.
This calls `startMarker()` on construction, and `endMarker()` on destruction, thus being usable like a "scope guard": the object is instantiated, and automatically destroyed when going out of scope.
Reviewed By: SidharthGuglani
Differential Revision: D13817589
fbshipit-source-id: fd88884af970c1c0933d9ca6843f3f8f5d28b9e6
Summary:
After this diff D13403925 that got rid of `-ffast-math` we started to have a very odd behavior on Yoga when using release builds.
After digging a while we found that certain set of conditions on O2 and O3 optimization levels was causing Origami to set some `INFINITE` values on Yoga.
We found the root of the problem and fix it on Origami side. But I'm wondering if guarding agains `INFINITE` on Yoga side would be good too. Since now Yoga it's not using `-ffast-math` anymore, and I think infinite is not a a valid value anywhere on Yoga side, it seems to support the reason to guard against it.
I'm happy to abandon this diff if you guys think this is not a good solution.
Reviewed By: davidaurelio
Differential Revision: D13679319
fbshipit-source-id: 061448fea9d1a8496362ff07dc46b546e7f1ffa3
Summary:
@public
`CompactValue` represents a `YGValue` in 32bits instead of 64. This comes at the cost of a range limitation, as one exponent bit is borrowed for the unit.
*Undefined* and *Auto* have no magnitude, and are represented as *NaN* values.
The data structure is meant to be used as a field type on `YGStyle` to save memory.
This is header-only for efficient inlining.
Reviewed By: jackerghan, aCorrado
Differential Revision: D13187211
fbshipit-source-id: 16e3ffad592e38e2493e4f7c8b952d372e449846
Summary:
@public
Replace `YGFloatOptional::getValue()` with `YGFloatOptional::unwrap()`.
`YGFloatOptional::getValue()` has the unfortunate property of calling `std::exit` if the wrapped value is undefined.
Here, we eliminate the method, and just call `.unwrap()` everywhere.
Reviewed By: shergin
Differential Revision: D13439608
fbshipit-source-id: 5ae82b170537d0a10c301412567a7a66fd50bab4
Summary:
Replaces `YGUnwrapFloatOptional` with `YGFloatOptional::unwrap`.
This leads to more idiomatic C++, and to less function call nesting, thus increasing readability.
Reviewed By: SidharthGuglani
Differential Revision: D13439604
fbshipit-source-id: 33b43c08d725c253c359959e7cbbd83fd6bd9ba4
