Hi Keith,
Thanks for the reply.
On Mon, Dec 2, 2019 at 6:31 PM Keith Medcalf wrote:
> Well, of course the trigger did not fire. You created an AFTER UPDATE
> trigger, but only did an insert. So of course one would not expect the
> trigger to fire.
You're right- this was a copy/paste error
I have been trying out generated column support in the 3.31.0 prerelease.
Thank you for adding such a useful feature!
When I create a trigger for an update of a generated column the trigger is
successfully created but it never fires. I could not find any mention in
the draft documentation of
Hello,
I have a request for two API additions: sqlite3_vconfig() and
sqlite3_db_vconfig(). The naming convention is borrowed from vprintf() and
friends. The functions would be identical to sqlite3_config() and
sqlite3_db_config() except they would take a va_list instead of variadic
arguments.
On Sat, Sep 29, 2018 at 5:33 PM Richard Hipp wrote:
> On 9/29/18, Stephen F. Booth wrote:
> > A query that ran fine under SQLite 3.24 is substantially slower in 3.25:
>
> Thanks for the data sent off-list
>
> Your work-around is to add a plus sign "+" b
A query that ran fine under SQLite 3.24 is substantially slower in 3.25:
SQLite version 3.24.0 2018-06-04 19:24:41
> .timer on
> select a.id from a join c on a.id = case when c.b_a_name is not null then
c.b_a_id else c.a_id end where a.id in (select a_fts.rowid from a_fts where
a_fts match 'r*'
I have been working on a custom FTS5 tokenizer using SQLite 3.21.0. I
noticed that in fts5CreateTokenizer() if the call to sqlite3_malloc() fails
SQLITE_NOMEM is returned, however xDestroy (if present) is not called. A
similar situation is handled differently in sqlite3_bind_pointer(); in that
6 matches
Mail list logo