Re: [sqlite] Behavior change around triggers between 3.19 and 3.30

2020-01-30 Thread Dominique Devienne
On Thu, Jan 30, 2020 at 3:38 PM Graham Holden wrote: > Thursday, January 30, 2020, 12:24:40 PM, Dominique Devienne > wrote: > > The strange thing though, is that I can't repro on a small example. > > Despite using not_there in the trigger, and doing DML and ALTER TABLE, > > still doesn't fail

Re: [sqlite] Behavior change around triggers between 3.19 and 3.30

2020-01-30 Thread Graham Holden
Thursday, January 30, 2020, 12:24:40 PM, Dominique Devienne wrote: > The strange thing though, is that I can't repro on a small example. > Despite using not_there in the trigger, and doing DML and ALTER TABLE, > still doesn't fail the same way as in production. What could be the cause? > --DD

Re: [sqlite] Behavior change around triggers between 3.19 and 3.30

2020-01-30 Thread Dominique Devienne
On Thu, Jan 30, 2020 at 1:09 PM Richard Hipp wrote: > On 1/30/20, Dominique Devienne wrote: > > My first question would be to ask whether there's a pragma or > > compile-time option to get back to the old behavior? > > Did you try "PRAGMA legacy_alter_table=ON;"? BINGO!!! Thanks a bunch

Re: [sqlite] Behavior change around triggers between 3.19 and 3.30

2020-01-30 Thread Richard Hipp
On 1/30/20, Dominique Devienne wrote: > > My first question would be to ask whether there's a pragma or > compile-time option to get back to the old behavior? Did you try "PRAGMA legacy_alter_table=ON;"? > > Second, any idea when this was introduced? > People have been requesting enhanced

[sqlite] Behavior change around triggers between 3.19 and 3.30

2020-01-30 Thread Dominique Devienne
BEFORE 3.19.3 2017-06-08 14:26:16 0ee482a1e0eae22e08edc8978c9733a96603d4509645f348ebf55b579e89636b AFTER 3.30.1 2019-10-10 20:19:45 18db032d058f1436ce3dea84081f4ee5a0f2259ad97301d43c426bc7f3dfalt2 Every 18 to 24 months we upgrade SQLite in a large commercial software suite. Such a recent upgrade