CLDR Ticket #5859(accepted data)
Ambiguous Collation of ll before Middle Dot
|Reported by:||richard.wordingham@…||Owned by:||markus|
|Component:||collation||Data Locale:||cy es sq|
Albanian, Spanish (traditional) and Welsh have collation tailorings for 'll', 'lL' and 'LL'. If the root weightings for 'l·' and 'L·' are interpreted as contractions, as implied by allkeys_CLDR.txt and the published definition of FractionalUCA.txt (though that definition is probably wrong - to be dealt with under ticket 5850), the sequence ll· will have weight CE(ll)CE(·), not CE(ll)CE(l|·). To correct this anomaly, the following three prefix rules need to be added for the three locales:
(The use of 'p' is correct for UCA 6.2.0 and CLDR Version 23.)
Applications interpreting prefix rules as equivalent to single contractions will then derive the same collations as implementations interpreting '004C | 00B7' in FractionalUCA.txt as an ICU prefix rule.