On 10/2/15, Bart Smissaert <bart.smissaert at gmail.com> wrote: > Do you mean that is only valid if there are no rows where all columns are > NULL?
No, I mean that two rows can be identical (not distinct) event if there is a unique index on all columns, if any single column in those two rows is NULL. Example: SQLite version 3.8.11 2015-07-15 23:15:59 Enter ".help" for usage hints. Connected to a transient in-memory database. Use ".open FILENAME" to reopen on a persistent database. sqlite> create table t1(a,b,c); sqlite> create unique index t1u on t1(a,b,c); sqlite> insert into t1(a,b,c) values(1,null,3),(1,null,3),(1,null,3); sqlite> -- D. Richard Hipp drh at sqlite.org