Re: [sqlite] The upcoming 3.23.0 release

Dear Richard, the rather complicated script below used to print a tree: node1|node2|node3|node4|node5 1 1|11||| 1|12||| 1|12|121|| 1|13||| and with yesterday's pre-release snapshot it does not yield any result. A clue is this depends on the size of the query, not so much on the logic. If

Re: [sqlite] The upcoming 3.23.0 release

If only all tech manuals were written like that, I'd be doing much MUCH more entertaining things with my life. Mind you, I love my job as is On Fri, Mar 23, 2018 at 1:14 PM, Simon Slavin wrote: > > You'll get mixed up, as you know > You're not sure what you know > So

Re: [sqlite] The upcoming 3.23.0 release

On 23 Mar 2018, at 3:09pm, R Smith wrote: > select blue, one from foo where one is blue or blue is true; > -- blue | one > -- | > -- 1 | 1 > > Is the blue one true? or the true one blue? or is this the one true

Re: [sqlite] The upcoming 3.23.0 release

On 2018/03/23 3:59 PM, David Raymond wrote: sqlite> create table foo (true boolean); sqlite> insert into foo (true) values (false); sqlite> select * from foo; true 0 sqlite> select 1 is true; 1 is true 1 sqlite> select 1 is true from foo; 1 is true 0 sqlite> select 1, true, 1 is true

Re: [sqlite] The upcoming 3.23.0 release

"Moral: Don't use identifiers "true" or "false" as column names." Oh, I'm in full agreement there. But as you said: "developers who use SQLite tend to be very creative and wind up using SQLite in ways that we never thought to test." So I tried to think up some psychotic/pathological cases to

Re: [sqlite] The upcoming 3.23.0 release

On 3/23/18, David Raymond wrote: > > sqlite> create table foo (true boolean); > > sqlite> insert into foo (true) values (false); > Then you get a non-empty set from: SELECT * FROM foo WHERE true IS false; Moral: Don't use identifiers "true" or "false" as column

Re: [sqlite] The upcoming 3.23.0 release

: Thursday, March 22, 2018 3:09 PM To: General Discussion of SQLite Database; sqlite-dev Subject: [sqlite] The upcoming 3.23.0 release SQLite version 3.23.0 will probably be released soon, in early April. For a summary of changes see https://sqlite.org/draft/releaselog/3_23_0.html Please download the late

Re: [sqlite] The upcoming 3.23.0 release

Eric Grange wrote: >> Add the LEFT JOIN strength reduction optimization that converts a LEFT >> JOIN into an ordinary JOIN > > A question on this one: I have been using LEFT JOIN for many years (and not > juste in SQLIte) in cases where a JOIN could have been used as a way to > "hint" query

Re: [sqlite] The upcoming 3.23.0 release

> Add the LEFT JOIN strength reduction optimization that converts a LEFT JOIN into an ordinary JOIN A question on this one: I have been using LEFT JOIN for many years (and not juste in SQLIte) in cases where a JOIN could have been used as a way to "hint" query optimizers which tables to scan

Re: [sqlite] The upcoming 3.23.0 release

> Le 22 mars 2018 à 20:09, Richard Hipp a écrit : > > ... > Please download the latest Pre-release Snapshot > (https://sqlite.org/download.html) and test out the latest SQLite in > your applications. Report any issues, either to this mailing list, or > directly to me at

Re: [sqlite] The upcoming 3.23.0 release

On 22 Mar 2018, at 7:09pm, Richard Hipp wrote: > For a summary of changes see > https://sqlite.org/draft/releaselog/3_23_0.html "causes the database connection D to disconnection from database S"

[sqlite] The upcoming 3.23.0 release

SQLite version 3.23.0 will probably be released soon, in early April. For a summary of changes see https://sqlite.org/draft/releaselog/3_23_0.html Please download the latest Pre-release Snapshot (https://sqlite.org/download.html) and test out the latest SQLite in your applications. Report any