Jeff Davis wrote: > There are a number of internal callers of LOWER(), and conceptually > those should all be using CASEFOLD(). Patches attached.
I tried 0001 with a non-UTF8 database and got quickly stuck: create database latin9 encoding='LATIN9' template=template0 locale='fr_FR.ISO-8859-15@euro'; \c latin9 select 'abc' ilike 'bc'; ERROR: Unicode case folding can only be performed if server encoding is UTF8 Presumably the internal calls to casefold should be conditionalized on the encoding being UTF-8? Best regards, -- Daniel Vérité https://postgresql.verite.pro/
