Don't break a flex line on floating-point rounding noise - #2022
Open
jwatzman wants to merge 1 commit into
Open
Conversation
A content-sized wrapping row sizes itself to the max-content sum of its items, then recovers the space available to those items by subtracting its padding and border back off (calculateAvailableInnerDimension). In float32 that add-then-subtract round-trip can land a fraction of an ulp below the sum it came from. For an item with a measure function the measurement cache still returns the max-content measurement in that situation: oldSizeIsMaxContentAndStillFits accepts a cached width up to the inexactEquals epsilon (1e-4) wider than the space now available. calculateFlexLine then compared that basis against availableInnerMainDim with a bare `>`, so the basis the cache had just accepted as fitting read as overflow and the line broke. Because the container had already been sized for a single line, the result was a box one line tall with its items laid out on two, overflowing its own padding and border. In React Native this reproduced on both iOS and Android and was extremely sensitive to the measured text width, since only some values leave a residue inside the window. Make the overflow test tolerant of the same epsilon the cache uses, so the two agree. Items without a measure function are unaffected: they take their exact style width in both passes and never enter that cache path. The regression test sweeps fractional measured widths whose content sum sits just under 128 while the outer width sits just over it, so adding padding and border crosses a float32 binade and the subtraction cannot always recover the original value. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
facebook-github-tools
Bot
added
the
Shared with Meta
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
A content-sized wrapping row sizes itself to the max-content sum of its items, then recovers the space available to those items by subtracting its padding and border back off (calculateAvailableInnerDimension). In float32 that add-then-subtract round-trip can land a fraction of an ulp below the sum it came from.
For an item with a measure function the measurement cache still returns the max-content measurement in that situation: oldSizeIsMaxContentAndStillFits accepts a cached width up to the inexactEquals epsilon (1e-4) wider than the space now available. calculateFlexLine then compared that basis against availableInnerMainDim with a bare
>, so the basis the cache had just accepted as fitting read as overflow and the line broke.Because the container had already been sized for a single line, the result was a box one line tall with its items laid out on two, overflowing its own padding and border. In React Native this reproduced on both iOS and Android and was extremely sensitive to the measured text width, since only some values leave a residue inside the window.
Make the overflow test tolerant of the same epsilon the cache uses, so the two agree. Items without a measure function are unaffected: they take their exact style width in both passes and never enter that cache path.
The regression test sweeps fractional measured widths whose content sum sits just under 128 while the outer width sits just over it, so adding padding and border crosses a float32 binade and the subtraction cannot always recover the original value.