I did a little Python script to go find what the data was that was actually being stored. They're each getting stored as serial type 6: "Value is a big-endian 64-bit twos-complement integer."
So apparently unique indexes consider uniqueness based on the *stored* value, whereas distinct queries, grouping, etc work on the value *after* it has been converted back to the column's declared type. -----Original Message----- From: sqlite-users [mailto:sqlite-users-boun...@mailinglists.sqlite.org] On Behalf Of James K. Lowden Sent: Thursday, November 01, 2018 3:23 PM To: sqlite-users@mailinglists.sqlite.org Subject: Re: [sqlite] Bug: float granularity breaking unique contraint? INSERT INTO TestReal values (9223372036854775807);INSERT INTO TestReal values (9223372036854775807 - 1);INSERT INTO TestReal values (9223372036854775807 - 2);INSERT INTO TestReal values (9223372036854775807 - 3);sqlite> ...> ...> ...> ...> _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
