Make CompactValue internal detail of yoga::Style
#41776
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D51775346 |
Base commit: 3854735 |
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 14, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 15, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 15, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 16, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 16, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 16, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 16, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 17, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 17, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 17, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 17, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
force-pushed
the
export-D51775346
branch
from
6acb691 to
c69d022
Compare
|
This pull request was exported from Phabricator. Differential Revision: D51775346 |
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
force-pushed
the
export-D51775346
branch
from
c69d022 to
c8bb5fa
Compare
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
|
This pull request was exported from Phabricator. Differential Revision: D51775346 |
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/react-native
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
Summary: X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
added a commit
to NickGerleman/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346
NickGerleman
force-pushed
the
export-D51775346
branch
from
c8bb5fa to
bc154bc
Compare
|
This pull request was exported from Phabricator. Differential Revision: D51775346 |
|
This pull request has been merged in 6e71851. |
facebook-github-bot
pushed a commit
to facebook/litho
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346 fbshipit-source-id: c618af41b4882b4a227c917fcad07375806faf78
facebook-github-bot
pushed a commit
to facebook/yoga
that referenced
this pull request
Dec 19, 2023
Summary: X-link: facebook/react-native#41776 Pull Request resolved: #1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346 fbshipit-source-id: c618af41b4882b4a227c917fcad07375806faf78
Othinn
pushed a commit
to Othinn/react-native
that referenced
this pull request
Jan 9, 2024
Summary: Pull Request resolved: facebook#41776 X-link: facebook/yoga#1492 # Summary In preparation to replace `CompactValue`, this fully encapsulates it as an implementation detail of `yoga::Style`. The internal API now always operates on `Style::Length`, converted to `YGValue` at the public API boundary. In the next step, we can plug in a new representation within `Style`, which should enable 64 bit values, and lower memory usage. # Test Plan 1. Existing tests (inc for style, invalidation, CompactValue) pass 2. Check that constexpr `yoga::isinf()` produces same assembly under Clang as `std::isinf()` 3. Fabric Android builds 4. Yoga benchmark does style reads # Performance Checking whether a style is defined, then reading after, is a hot path, and we are doubling any space style lengths take in the stack (but not long-term on the node). After a naive move, on one system, the Yoga benchmark creating, laying out, and destroying a tree, ran about 8-10% slower in the "Huge nested flex" example. We are converting in many more cases instead of doing undefined check, but operating on accessed style values no longer needs to do the conversion multiple times. I changed the `CompactValue` conversion to YGValue/StyleLength path to check for undefined as the common case (since we always convert, instead of calling `isUndefined` directly on CompactValue. That seemed to get the difference down to ~5-6% when I was playing with it then. We can optimistically make some of this up with ValuePool giving better locality, and fix this more holistically if we reduce edge and value resolution. On another machine where I tested this, the new revision went the opposite direction, and was about 5% faster, so this isn't really a cut and dry regression, but we see different characteristics than before. # Changelog [Internal] Reviewed By: rozele Differential Revision: D51775346 fbshipit-source-id: c618af41b4882b4a227c917fcad07375806faf78
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
In preparation to replace
CompactValue, this fully encapsulates it as an implementation detail ofyoga::Style.The internal API now always operates on
Style::Length, converted toYGValueat the public API boundary.In the next step, we can plug in a new representation within
Style, which should enable 64 bit values, and lower memory usage.Changelog: [Internal]
Differential Revision: D51775346