Re: [sqlite] PRAGMA reverse_unordered_selects=true results in row not being fetched

Hi Donald, You are right. Sorry for not mentioning that. I was working on trunk and using Linux. The bug has already been fixed: https://www.sqlite.org/src/info/bc7d2c1656396bb4 Best, Manuel On Wed, May 15, 2019 at 3:13 PM Donald Griggs wrote: > On Tue, May 14, 2019 at 10:18 AM Manuel Rigger

Re: [sqlite] PRAGMA reverse_unordered_selects=true results in row not being fetched

On Tue, May 14, 2019 at 10:18 AM Manuel Rigger wrote: > ... > > The query does not fetch a row. However, without the PRAGMA statement one > row is fetched. > It probably helps the devs to specify the version you were testing. For me, your example returns a single row of "1" in each case. I'm

[sqlite] PRAGMA reverse_unordered_selects=true results in row not being fetched

Hi everyone, I found a curious bug, which I could reproduce only with a very specific statement sequence: PRAGMA reverse_unordered_selects=true; CREATE TABLE t1 (c0, c1); CREATE TABLE t2 (c0 INT UNIQUE); INSERT INTO t1(c0, c1) VALUES (0, 0), (0, NULL); INSERT INTO t2(c0) VALUES (1); SELECT 1,

[sqlite] PRAGMA reverse_unordered_selects=true results into row not being fetched

Hi everyone, I discovered what I think is a bug, as demonstrated below: CREATE TABLE t0(c0); CREATE TABLE t1(c0 INTEGER PRIMARY KEY); PRAGMA reverse_unordered_selects=true; INSERT INTO t1(c0) VALUES (0); INSERT INTO t0(c0) VALUES ('a'); SELECT * FROM t1, t0 WHERE t1.c0 < t0.c0; I would expect