Thanks for the bug report. The ticket is https://www.sqlite.org/src/tktview/7b3328086a5c1
Your analysis of the problem seems spot on. I'm guessing you spent a long time debugging his.... Your efforts are appreciated. On 10/2/16, Clemens Ladisch <[email protected]> wrote: > Hi, > > if the transfer optimization needs to check if the destination table is > empty, and if the optimization is actually used, and if the destination > table uses AUTOINCREMENT, then the generated code does not update the > sqlite_sequence table. > > After these statements, the row in dst should have rowid 2, but the > rowid 1 got reused: > > create table src(a integer primary key autoincrement, b unique); > create table dst(a integer primary key autoincrement, b unique); > insert into src default values; > insert into dst select * from src; > -- select * from sqlite_sequence where name = 'dst'; is empty > delete from dst; > insert into dst default values; > select a from dst; > 1 > > (reported by Matthew Moisen on <http://stackoverflow.com/q/39820552/11654>) > > > Regards, > Clemens > _______________________________________________ > sqlite-users mailing list > [email protected] > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
