> Cannot happen. If not executed within an implicit transaction, each > SQL statement is effectively wrapped in a transaction all of its own.
So executing the insert/update/delete statement places RESERVED lock right from the start before executing nested selects? Pavel On Tue, Aug 18, 2009 at 1:54 PM, Dan Kennedy<[email protected]> wrote: > > On Aug 19, 2009, at 12:35 AM, Pavel Ivanov wrote: > >> You can always do >> >> insert into table (id, count) values (?1, (select count(*) + 1 from >> table where id = ?1)) >> >> Though I'd be cautious about race condition that seems like possible >> here when after select returned some value and before insert was made >> another process made another insert with the same id... > > Cannot happen. If not executed within an implicit transaction, each > SQL statement is effectively wrapped in a transaction all of its own. > > Dan. > > _______________________________________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
