[stdlib] Implement native grapheme breaking for String #37864

Azoy · 2021-06-10T18:12:43Z

This is a draft and will see many updates, but wanted to put this up to get some benchmark readings.

Resolves: rdar://52194063, https://bugs.swift.org/browse/SR-9423

Azoy · 2021-06-10T18:13:40Z

@swift-ci Please benchmark

Azoy · 2021-06-10T18:26:50Z

@swift-ci Please test

Azoy · 2021-06-11T16:46:11Z

@swift-ci Please benchmark

Azoy · 2021-06-23T07:02:03Z

@swift-ci please benchmark

milseman

This is a quick preliminary review, so many of my suggestions are vague and might not actually make sense. But, hopefully it gives you something to chew on or encourages some refactoring or clarifying.

Looking good so far though!

stdlib/public/core/GraphemeIterator.swift

stdlib/public/core/UnicodeGraphemeBreakProperty.swift

stdlib/public/core/GraphemeIterator.swift

milseman

Quick review. This is looking a lot cleaner, thanks for scrapping the iterator pattern!

Refactoring out state helps with the clarity quite a bit. It also shows that index does seem like a different bit of state than the others, so we might pull that out and put it back on the walker or somewhere. Something to consider.

stdlib/public/core/GraphemeWalker.swift

Azoy · 2021-10-18T20:05:51Z

@swift-ci please benchmark

Azoy · 2021-10-18T22:48:28Z

@swift-ci please benchmark

Azoy · 2021-10-21T23:31:39Z

@swift-ci please benchmark

stdlib/public/core/GraphemeWalker.swift

…gn strings

use an Iterator approach remove Iterator conformance

more comments addressed fix crlf bug

milseman

Sorry it's hasty, but I wanted to get you some feedback quickly.

stdlib/public/core/GraphemeWalkerUTF8.swift

Parameterize nextBoundary and previousBoundary

Azoy · 2021-10-29T11:06:00Z

@swift-ci please test

milseman

This is very close!

For landing, I think we can simplify a lot of the state that's passed around and mutated, especially when going backwards.

For future work, I'm interested in splitting this between forwards and backwards a little (since I think that will actually simplify the logic), but I don't want to hold up landing this for that. Those comments are written with "Future note:" at the front.

stdlib/public/core/StringGraphemeBreaking.swift

milseman · 2021-10-29T14:01:33Z

stdlib/public/core/StringGraphemeBreaking.swift

+  //                | = We found our starting .extendedPictographic letting us
+  //                    know that we are in an emoji sequence so our initial
+  //                    break question is answered as NO.
+  internal func checkIfInEmojiSequence(


Future note: One thing that isn't clear to me is how this is different than the walking-backwards loop we're doing anyways inside of previousBoundary.

milseman · 2021-10-29T14:02:32Z

stdlib/public/core/StringGraphemeBreaking.swift

+    // GB11
+    case (.zwj, .extendedPictographic):
+      if state.isBackwards {
+        checkIfInEmojiSequence(&state, index)


Future note: If we're going backwards, seems like the return value is "yes, unless you're in an emoji sequence".

milseman · 2021-10-29T14:22:32Z

stdlib/public/core/StringGraphemeBreaking.swift

+    // state variable to false after every decision of 'shouldBreak'. If we
+    // happen to see a rhs .extend or .zwj, then it's a signal that we should
+    // continue treating the current grapheme cluster as an emoji sequence.
+    var enterEmojiSequence = false


Future note: Is there a way to simplify this logic? It seems to only be essential to forwards-state.

stdlib/public/core/StringGraphemeBreaking.swift

Azoy · 2021-10-29T21:49:34Z

@swift-ci please test

Azoy · 2021-10-29T21:49:42Z

@swift-ci please benchmark

milseman

LGTM. Just a couple places for comments. I think we should try to merge this soon. There's more future work we could do, but this is great for now!

milseman · 2021-10-29T22:05:07Z

stdlib/public/core/StringGraphemeBreaking.swift

-    let cocoa = _object.cocoaObject
+    // GB11
+    case (.zwj, .extendedPictographic):
+      if isBackwards {


Maybe add a comment describing what we're doing and why

milseman · 2021-10-29T22:05:45Z

stdlib/public/core/StringGraphemeBreaking.swift

+    // GB12 & GB13
+    case (.regionalIndicator, .regionalIndicator):
+      if isBackwards {
+        return countRIs(index)


Maybe a comment, or a more descriptive name like checkPairedRIs.

swift-ci · 2021-10-29T23:06:09Z

Performance (x86_64): -O

Regression	OLD	NEW	DELTA	RATIO
LessSubstringSubstring	22	29	+31.8%	0.76x
EqualSubstringSubstringGenericEquatable	22	29	+31.8%	0.76x
LessSubstringSubstringGenericComparable	22	29	+31.8%	0.76x
EqualSubstringSubstring	22	28	+27.3%	0.79x
EqualStringSubstring	22	28	+27.3%	0.79x
EqualSubstringString	22	28	+27.3%	0.79x
StringComparison_longSharedPrefix	293	330	+12.6%	0.89x (?)
SortStringsUnicode	2045	2255	+10.3%	0.91x
SubstringEqualString	142	155	+9.2%	0.92x

Improvement	OLD	NEW	DELTA	RATIO
SubstringTrimmingASCIIWhitespace	441	122	-72.3%	3.61x
StringHasSuffixUnicode	142000	56000	-60.6%	2.54x
StringHasPrefixUnicode	83000	34000	-59.0%	2.44x
CSVParsing.Char	225	140	-37.8%	1.61x
StringUTF16SubstringBuilder	2520	1620	-35.7%	1.56x
CharacterPropertiesPrecomputed	850	610	-28.2%	1.39x
CharacterPropertiesStashed	790	580	-26.6%	1.36x
CharacterPropertiesStashedMemo	930	690	-25.8%	1.35x
LineSink.characters.alpha	81	61	-24.7%	1.33x
Breadcrumbs.MutatedUTF16ToIdx.Mixed	232	193	-16.8%	1.20x
LineSink.characters.complex	428	362	-15.4%	1.18x
Breadcrumbs.MutatedIdxToUTF16.Mixed	235	199	-15.3%	1.18x
RemoveWhereQuadraticString	196	166	-15.3%	1.18x
CharacterPropertiesFetch	2200	1960	-10.9%	1.12x
StringAdder	266	237	-10.9%	1.12x (?)
StringBuilderSmallReservingCapacity	213	190	-10.8%	1.12x
StringBuilder	205	184	-10.2%	1.11x
StringUTF16Builder	220	200	-9.1%	1.10x
StringInterpolationSmall	1150	1070	-7.0%	1.07x (?)
DropWhileAnySequence	1353	1260	-6.9%	1.07x (?)
ArraySetElement	304	284	-6.6%	1.07x (?)

Code size: -O

Regression	OLD	NEW	DELTA	RATIO
RC4.o	3425	3665	+7.0%	0.93x

Improvement	OLD	NEW	DELTA	RATIO
UTF8Decode.o	23381	23109	-1.2%	1.01x

Performance (x86_64): -Osize

Regression	OLD	NEW	DELTA	RATIO
SuffixAnySequence	122	1900	+1457.4%	0.06x
SuffixSequence	131	1901	+1351.1%	0.07x
SuffixSequenceLazy	131	1788	+1264.9%	0.07x
SuffixArrayLazy	5	9	+80.0%	0.56x
FlattenListLoop	943	1386	+47.0%	0.68x (?)
LessSubstringSubstring	22	29	+31.8%	0.76x
EqualStringSubstring	22	29	+31.8%	0.76x
EqualSubstringSubstringGenericEquatable	22	29	+31.8%	0.76x
EqualSubstringString	22	29	+31.8%	0.76x
LessSubstringSubstringGenericComparable	22	29	+31.8%	0.76x
EqualSubstringSubstring	23	29	+26.1%	0.79x
StringComparison_longSharedPrefix	295	331	+12.2%	0.89x (?)
UTF8Decode_InitFromCustom_noncontiguous	257	287	+11.7%	0.90x (?)
UTF8Decode_InitFromCustom_noncontiguous_ascii	620	689	+11.1%	0.90x
SuffixAnySequenceLazy	3037	3357	+10.5%	0.90x (?)
UTF8Decode_InitFromCustom_noncontiguous_ascii_as_ascii	717	788	+9.9%	0.91x (?)
RemoveWhereMoveInts	21	23	+9.5%	0.91x (?)
SortStringsUnicode	2055	2245	+9.2%	0.92x
ArraySetElement	296	322	+8.8%	0.92x (?)
StringWalk	3240	3520	+8.6%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
SubstringTrimmingASCIIWhitespace	443	125	-71.8%	3.54x
StringHasSuffixUnicode	142000	56000	-60.6%	2.54x
StringHasPrefixUnicode	83000	34000	-59.0%	2.44x
CSVParsing.Char	226	143	-36.7%	1.58x
StringUTF16SubstringBuilder	2570	1640	-36.2%	1.57x
CharacterPropertiesPrecomputed	860	620	-27.9%	1.39x
CharacterPropertiesStashed	790	580	-26.6%	1.36x
LineSink.characters.alpha	82	61	-25.6%	1.34x
CharacterPropertiesStashedMemo	930	700	-24.7%	1.33x
SuffixAnyCollection	36	30	-16.7%	1.20x (?)
Breadcrumbs.MutatedUTF16ToIdx.Mixed	231	194	-16.0%	1.19x
LineSink.characters.complex	426	359	-15.7%	1.19x
Breadcrumbs.MutatedIdxToUTF16.Mixed	233	199	-14.6%	1.17x (?)
RemoveWhereQuadraticString	199	170	-14.6%	1.17x
CharacterPropertiesFetch	2200	1960	-10.9%	1.12x
StringBuilderSmallReservingCapacity	217	194	-10.6%	1.12x (?)
StringBuilder	210	188	-10.5%	1.12x (?)
StringUTF16Builder	220	200	-9.1%	1.10x
StringAdder	269	245	-8.9%	1.10x (?)
StringInterpolationSmall	1130	1040	-8.0%	1.09x (?)

Code size: -Osize

Regression	OLD	NEW	DELTA	RATIO
Suffix.o	18820	24636	+30.9%	0.76x
RC4.o	3000	3202	+6.7%	0.94x

Improvement	OLD	NEW	DELTA	RATIO
UTF8Decode.o	21726	21407	-1.5%	1.01x

Performance (x86_64): -Onone

Regression	OLD	NEW	DELTA	RATIO
LessSubstringSubstringGenericComparable	49	56	+14.3%	0.88x
EqualSubstringSubstringGenericEquatable	50	56	+12.0%	0.89x (?)
LessSubstringSubstring	52	58	+11.5%	0.90x (?)
EqualSubstringSubstring	52	58	+11.5%	0.90x
EqualStringSubstring	52	58	+11.5%	0.90x
EqualSubstringString	52	58	+11.5%	0.90x

Improvement	OLD	NEW	DELTA	RATIO
StringHasSuffixUnicode	145000	59000	-59.3%	2.46x
StringHasPrefixUnicode	86000	37000	-57.0%	2.32x
SubstringTrimmingASCIIWhitespace	964	579	-39.9%	1.66x
LineSink.characters.alpha	109	88	-19.3%	1.24x
StringUTF16SubstringBuilder	7250	5960	-17.8%	1.22x
CSVParsing.Char	524	434	-17.2%	1.21x
Breadcrumbs.MutatedIdxToUTF16.Mixed	245	209	-14.7%	1.17x (?)
Breadcrumbs.MutatedUTF16ToIdx.Mixed	232	198	-14.7%	1.17x
CharacterPropertiesStashed	1520	1300	-14.5%	1.17x (?)
LineSink.characters.complex	502	434	-13.5%	1.16x
CharacterPropertiesPrecomputed	1960	1730	-11.7%	1.13x (?)
CharacterPropertiesStashedMemo	2230	2000	-10.3%	1.11x (?)
ArrayOfPOD	733	669	-8.7%	1.10x (?)
CharacterPropertiesFetch	2950	2700	-8.5%	1.09x (?)
NSError	476	444	-6.7%	1.07x (?)

Code size: -swiftlibs

How to read the data

The tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.

If you see any unexpected regressions, you should consider fixing the
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac mini
  Model Identifier: Macmini8,1
  Processor Name: 6-Core Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

swift-ci · 2021-10-29T23:12:15Z

Build failed
Swift Test Linux Platform
Git Sha - 16abfe5

milseman · 2021-10-30T00:00:01Z

@swift-ci please test linux platform

xwu · 2021-10-30T04:00:56Z

Are the -Osize performance and/or code size regressions in the x86_64 Suffix* benchmarks a problem?

Azoy · 2021-10-30T04:26:39Z

I have no idea why those benchmarks are showing as a regression in this patch (and it seems they showed a 3x improvement with the normalization patch). Considering they don't even touch strings at all, I'm inclined to say they are not an issue, but I could be wrong.

Azoy · 2021-11-01T23:50:54Z

Here are some benchmarks that I ran locally with the StringWalk benchmarks enabled:

Performance (arm64): -O

Regression	OLD	NEW	DELTA	RATIO
ArrayAppendGenericStructs	920	1220	+32.6%	0.75x (?)
CharIteration_utf16_unicodeScalars	1720	2040	+18.6%	0.84x (?)
ParseFloat.Float.Exp	6	7	+16.7%	0.86x (?)
Set.isSubset.Seq.Int.Empty	82	93	+13.4%	0.88x (?)
Set.isStrictSubset.Seq.Int.Empty	82	93	+13.4%	0.88x (?)
ArrayPlusEqualArrayOfInt	180	200	+11.1%	0.90x (?)
ArrayAppendToGeneric	190	210	+10.5%	0.90x (?)
BufferFillFromSlice	10	11	+10.0%	0.91x (?)
StringWalk_punctuated_characters_Backwards	440	480	+9.1%	0.92x (?)
StringWalk_ascii_characters_Backwards	1800	1960	+8.9%	0.92x (?)
DataCreateEmptyArray	650	700	+7.7%	0.93x (?)

Improvement	OLD	NEW	DELTA	RATIO
SubstringTrimmingASCIIWhitespace	314	86	-72.6%	3.65x
StringWalk_utf16_characters	26080	7200	-72.4%	3.62x
StringWalk_utf16_characters_Backwards	27240	10240	-62.4%	2.66x
SuffixCountableRange	13	6	-53.8%	2.17x
StringHasSuffixUnicode	100000	48000	-52.0%	2.08x (?)
Data.append.Sequence.64kB.Count.RE.I	2	1	-50.0%	2.00x (?)
StringHasPrefixUnicode	60000	32000	-46.7%	1.87x (?)
CSVParsing.Char	151	94	-37.7%	1.61x
StringUTF16SubstringBuilder	1620	1090	-32.7%	1.49x (?)
CharacterPropertiesStashed	550	380	-30.9%	1.45x (?)
CharacterPropertiesPrecomputed	690	520	-24.6%	1.33x
LineSink.characters.alpha	52	40	-23.1%	1.30x
ArrayAppendOptionals	920	710	-22.8%	1.30x (?)
StringWalk_tweet_characters_Backwards	4920	4040	-17.9%	1.22x (?)
CharacterPropertiesStashedMemo	800	660	-17.5%	1.21x (?)
LineSink.characters.complex	277	230	-17.0%	1.20x (?)
DataCreateMedium	1200	1000	-16.7%	1.20x (?)
CharacterPropertiesFetch	1510	1350	-10.6%	1.12x (?)
StringWalk_tweet_characters	3320	3000	-9.6%	1.11x (?)
NSArray.bridged.objectAtIndex	229	208	-9.2%	1.10x (?)
ParseInt.UInt64.Hex	243	222	-8.6%	1.09x (?)

Code size: -O

Performance (arm64): -Osize

Regression	OLD	NEW	DELTA	RATIO
SuffixSequenceLazy	117	828	+607.7%	0.14x
SuffixAnySequence	117	802	+585.5%	0.15x
SuffixSequence	117	802	+585.5%	0.15x
DataCreateEmptyArray	800	950	+18.7%	0.84x (?)
DataToStringEmpty	400	450	+12.5%	0.89x (?)
SuffixAnySequenceLazy	1633	1814	+11.1%	0.90x (?)
DataCreateSmallArray	1650	1800	+9.1%	0.92x (?)
StringWalk_ascii_characters_Backwards	1880	2040	+8.5%	0.92x (?)
StringWalk_punctuated_characters_Backwards	480	520	+8.3%	0.92x (?)
StringWalk_punctuatedJapanese_characters_Backwards	480	520	+8.3%	0.92x (?)
SortArrayInClass	58036	62725	+8.1%	0.93x (?)

Improvement	OLD	NEW	DELTA	RATIO
SubstringTrimmingASCIIWhitespace	317	88	-72.2%	3.60x
StringWalk_utf16_characters	26200	7280	-72.2%	3.60x
StringWalk_utf16_characters_Backwards	27160	10280	-62.2%	2.64x
StringHasSuffixUnicode	101000	47000	-53.5%	2.15x (?)
StringHasPrefixUnicode	60000	32000	-46.7%	1.87x (?)
CSVParsing.Char	158	101	-36.1%	1.56x
StringUTF16SubstringBuilder	1670	1130	-32.3%	1.48x (?)
CharacterPropertiesStashedMemo	840	580	-31.0%	1.45x (?)
CharacterPropertiesStashed	630	440	-30.2%	1.43x (?)
SIMDReduce.Int8x16.Cast	43	31	-27.9%	1.39x (?)
CharacterPropertiesPrecomputed	680	500	-26.5%	1.36x
LineSink.characters.alpha	57	45	-21.1%	1.27x
StringWalk_tweet_characters_Backwards	5120	4240	-17.2%	1.21x (?)
LineSink.characters.complex	294	247	-16.0%	1.19x (?)
CharacterPropertiesFetch	1550	1380	-11.0%	1.12x (?)
StringWalk_tweet_characters	3560	3240	-9.0%	1.10x (?)
ConvertFloatingPoint.MockFloat64ToInt64	433	402	-7.2%	1.08x (?)
FindString.Rec3.Substring	75	70	-6.7%	1.07x (?)

Code size: -Osize

Performance (arm64): -Onone

Regression	OLD	NEW	DELTA	RATIO
Breadcrumbs.MutatedUTF16ToIdx.ASCII	2	3	+50.0%	0.67x (?)
ArrayOfGenericPOD2	902	1029	+14.1%	0.88x (?)
String.replaceSubrange.String	12	13	+8.3%	0.92x (?)
RawBufferCopyBytes	12	13	+8.3%	0.92x (?)

Improvement	OLD	NEW	DELTA	RATIO
StringWalk_utf16_characters	29720	10360	-65.1%	2.87x
StringWalk_utf16_characters_Backwards	31560	13960	-55.8%	2.26x
StringHasSuffixUnicode	102000	49000	-52.0%	2.08x (?)
StringHasPrefixUnicode	61000	33000	-45.9%	1.85x (?)
SubstringTrimmingASCIIWhitespace	648	401	-38.1%	1.62x (?)
ArrayPlusEqualArrayOfInt	220	170	-22.7%	1.29x (?)
CharacterPropertiesStashed	1010	830	-17.8%	1.22x (?)
CSVParsing.Char	334	275	-17.7%	1.21x (?)
StringUTF16SubstringBuilder	4390	3710	-15.5%	1.18x (?)
LineSink.characters.alpha	71	61	-14.1%	1.16x (?)
CharacterPropertiesPrecomputed	1430	1230	-14.0%	1.16x (?)
LineSink.characters.complex	330	285	-13.6%	1.16x (?)
CharacterPropertiesStashedMemo	1700	1500	-11.8%	1.13x (?)
CharacterPropertiesFetch	1940	1760	-9.3%	1.10x (?)
StringWalk_tweet_characters_Backwards	13680	12680	-7.3%	1.08x (?)
DataReplaceSmall	1500	1400	-6.7%	1.07x (?)

Code size: -swiftlibs

How to read the data

The tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.

If you see any unexpected regressions, you should consider fixing the
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: MacBook Pro
  Model Identifier: MacBookPro17,1
  Total Number of Cores: 8 (4 performance and 4 efficiency)
  Memory: 16 GB

This comment has been minimized.

Sign in to view

Azoy force-pushed the native-grapheme-breaking branch 3 times, most recently from 753875e to 17c277f Compare June 22, 2021 16:57

swiftlang deleted a comment from swift-ci Jun 22, 2021

Azoy force-pushed the native-grapheme-breaking branch from 17c277f to c05e787 Compare June 22, 2021 18:12

swiftlang deleted a comment from swift-ci Jun 23, 2021

This comment has been minimized.

Sign in to view

Azoy force-pushed the native-grapheme-breaking branch from c05e787 to 716d6f6 Compare August 10, 2021 05:41

Azoy mentioned this pull request Sep 8, 2021

[utils] Add a Unicode data generator util package #39213

Merged

Azoy force-pushed the native-grapheme-breaking branch from 716d6f6 to 9004029 Compare September 24, 2021 23:36

Azoy force-pushed the native-grapheme-breaking branch from 9004029 to effbcb2 Compare October 4, 2021 07:48

milseman reviewed Oct 8, 2021

View reviewed changes

Azoy force-pushed the native-grapheme-breaking branch from effbcb2 to aa8da8f Compare October 14, 2021 21:34

Azoy marked this pull request as ready for review October 17, 2021 23:42

milseman reviewed Oct 18, 2021

View reviewed changes

Azoy changed the title ~~[WIP][stdlib] Implement native grapheme breaking for String~~ [stdlib] Implement native grapheme breaking for String Oct 18, 2021

Azoy force-pushed the native-grapheme-breaking branch from aa8da8f to c385938 Compare October 18, 2021 20:04

Azoy force-pushed the native-grapheme-breaking branch from c385938 to b3bb9ac Compare October 18, 2021 22:47

This comment has been minimized.

Sign in to view

WowbaggersLiquidLunch reviewed Oct 24, 2021

View reviewed changes

stdlib/public/core/GraphemeWalker.swift Outdated Show resolved Hide resolved

Azoy added 7 commits October 25, 2021 13:05

Implement GraphemeWalker that does native grapheme breaking

f61754b

Bridged strings use native grapheme breaking for forward strides

c4175d1

Implement bidirectional native grapheme breaking for native and forei…

2f7405a

…gn strings

Remove ICU's grapheme breaking support

89b890c

Use UnicodeScalarView to implement GraphemeWalker

af10a9b

use an Iterator approach remove Iterator conformance

Incorporate Michael's feedback

004ac2c

more comments addressed fix crlf bug

Try bringing back some old fast paths

ec31e3f

Azoy force-pushed the native-grapheme-breaking branch from 2d04fd3 to 58fb22c Compare October 25, 2021 23:18

milseman reviewed Oct 26, 2021

View reviewed changes

stdlib/public/core/GraphemeWalkerUTF8.swift Outdated Show resolved Hide resolved

Parameterize nextBoundary and previousBoundary

85352c2

Parameterize nextBoundary and previousBoundary

Azoy force-pushed the native-grapheme-breaking branch from 58fb22c to 85352c2 Compare October 29, 2021 04:54

milseman reviewed Oct 29, 2021

View reviewed changes

Implement Michael's suggestions

16abfe5

milseman approved these changes Oct 29, 2021

View reviewed changes

Azoy merged commit 5a0bbb9 into swiftlang:main Nov 1, 2021

Azoy deleted the native-grapheme-breaking branch November 1, 2021 23:52

Azoy mentioned this pull request Nov 18, 2021

[SR-9423] Swift Native Grapheme Breaking #51887

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[stdlib] Implement native grapheme breaking for String #37864

[stdlib] Implement native grapheme breaking for String #37864

Azoy commented Jun 10, 2021 •

edited

Loading

Azoy commented Jun 10, 2021

Azoy commented Jun 10, 2021

This comment has been minimized.

Azoy commented Jun 11, 2021

This comment has been minimized.

Azoy commented Jun 23, 2021

This comment has been minimized.

milseman left a comment

milseman left a comment

Azoy commented Oct 18, 2021

Azoy commented Oct 18, 2021

This comment has been minimized.

Azoy commented Oct 21, 2021

This comment has been minimized.

milseman left a comment

Azoy commented Oct 29, 2021

milseman left a comment

milseman Oct 29, 2021

milseman Oct 29, 2021

milseman Oct 29, 2021 •

edited

Loading

Azoy commented Oct 29, 2021

Azoy commented Oct 29, 2021

milseman left a comment

milseman Oct 29, 2021

milseman Oct 29, 2021

swift-ci commented Oct 29, 2021

swift-ci commented Oct 29, 2021

milseman commented Oct 30, 2021

xwu commented Oct 30, 2021 •

edited

Loading

Azoy commented Oct 30, 2021

Azoy commented Nov 1, 2021

[stdlib] Implement native grapheme breaking for String #37864

[stdlib] Implement native grapheme breaking for String #37864

Conversation

Azoy commented Jun 10, 2021 • edited Loading

Azoy commented Jun 10, 2021

Azoy commented Jun 10, 2021

This comment has been minimized.

Azoy commented Jun 11, 2021

This comment has been minimized.

Azoy commented Jun 23, 2021

This comment has been minimized.

milseman left a comment

Choose a reason for hiding this comment

milseman left a comment

Choose a reason for hiding this comment

Azoy commented Oct 18, 2021

Azoy commented Oct 18, 2021

This comment has been minimized.

Azoy commented Oct 21, 2021

This comment has been minimized.

milseman left a comment

Choose a reason for hiding this comment

Azoy commented Oct 29, 2021

milseman left a comment

Choose a reason for hiding this comment

milseman Oct 29, 2021

Choose a reason for hiding this comment

milseman Oct 29, 2021

Choose a reason for hiding this comment

milseman Oct 29, 2021 • edited Loading

Choose a reason for hiding this comment

Azoy commented Oct 29, 2021

Azoy commented Oct 29, 2021

milseman left a comment

Choose a reason for hiding this comment

milseman Oct 29, 2021

Choose a reason for hiding this comment

milseman Oct 29, 2021

Choose a reason for hiding this comment

swift-ci commented Oct 29, 2021

Performance (x86_64): -O

Code size: -O

Performance (x86_64): -Osize

Code size: -Osize

Performance (x86_64): -Onone

Code size: -swiftlibs

swift-ci commented Oct 29, 2021

milseman commented Oct 30, 2021

xwu commented Oct 30, 2021 • edited Loading

Azoy commented Oct 30, 2021

Azoy commented Nov 1, 2021

Performance (arm64): -O

Code size: -O

Performance (arm64): -Osize

Code size: -Osize

Performance (arm64): -Onone

Code size: -swiftlibs

Azoy commented Jun 10, 2021 •

edited

Loading

milseman Oct 29, 2021 •

edited

Loading

xwu commented Oct 30, 2021 •

edited

Loading