@@ -95,6 +95,22 @@ export function compareUtf8Strings(left: string, right: string): number {
9595 // used to represent code points greater than 0xFFFF which have 4-byte UTF-8 representations
9696 // and are lexicographically greater than the 1, 2, or 3-byte representations of code points
9797 // less than or equal to 0xFFFF.
98+ //
99+ // An example of why Case 2 is required is comparing the following two Unicode code points:
100+ //
101+ // |-----------------------|------------|---------------------|-----------------|
102+ // | Name | Code Point | UTF-8 Encoding | UTF-16 Encoding |
103+ // |-----------------------|------------|---------------------|-----------------|
104+ // | Replacement Character | U+FFFD | 0xEF 0xBF 0xBD | 0xFFFD |
105+ // | Grinning Face | U+1F600 | 0xF0 0x9F 0x98 0x80 | 0xD83D 0xDE00 |
106+ // |-----------------------|------------|---------------------|-----------------|
107+ //
108+ // A lexicographical comparison of the UTF-8 encodings of these code points would order
109+ // "Replacement Character" _before_ "Grinning Face" because 0xEF is less than 0xF0. However, a
110+ // direct comparison of the UTF-16 code units, as would be done in case 1, would erroneously
111+ // produce the _opposite_ ordering, because 0xFFFD is _greater than_ 0xD83D. As it turns out,
112+ // this relative ordering holds for all comparisons of UTF-16 code points requiring a surrogate
113+ // pair with those that do not.
98114 const length = Math . min ( left . length , right . length ) ;
99115 for ( let i = 0 ; i < length ; i ++ ) {
100116 const leftChar = left . charAt ( i ) ;
0 commit comments