Example of bengali grapheme clusters out fo data

[andjc]

The current editors draft has the following text:

For example, the Bangla user-perceived character kshī ক্ষী is composed of four characters: U+0995 BENGALI LETTER KA + U+09CD BENGALI SIGN VIRAMA + U+09B7 BENGALI LETTER SSA + U+09C0 BENGALI VOWEL SIGN II.
Unicode splits these into two grapheme clusters, unless language-specific tailoring is applied. For more information, see our article Character encodings: Essential concepts.

This describes the behavior prior to Unicode 15.1. UAX29 was updated in the Unicode 15.1 release, adding an additional rule GB9c:

Do not break within certain combinations with Indic_Conjunct_Break (InCB)=Linker

For the example 'ক্ষী' , UAX29 revision 41 and earlier would result in two extended grapheme clusters ('ক্', 'ষী') while UAX29 revision 43 onwards results in a single extended grapheme cluster ('ক্ষী'). So behaviour is dependent on version of UAX29 (i.e. version of Unicode supported).