Aha. Got it. After PRAGMA case_sensitive_like =1
the optimizer might use the index with BINARY collation. I was under the mistaken impression that LIKE() adapted itself to the column's collation. But I see that it is not possible to define one column to use BINARY collation and another column to use NOCASE collation and have the LIKE operator be case-sensitive when used with the BINARY column and case-insensitive when used with the NOCASE column-- at least not without a PRAGMA change. Thanks. Tim Romano Igor Tandetnik wrote: > ... LIKE is case insensitive by default, while BINARY collation is case > sensitive. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
