On Mon, Aug 12, 2013 at 4:34 PM, Scott Hess <sh...@google.com> wrote:
> I had been attempting to write some code like this: > > CREATE TABLE t (x); > CREATE INDEX t_x ON t(x); > -- bunch of operations over a long period. > -- now we want to run an expiration pass: > BEGIN; > ALTER TABLE t RENAME TO t_old; > CREATE TABLE t (x); > CREATE INDEX t_x ON t(x); -- (*) > -- Bunch of code like: > INSERT INTO t (x) SELECT x FROM t_old WHERE (complicated); > DROP TABLE t_old; > COMMIT; > > Unfortunately, at (*) this throws: > Error: index t_x already exists > > Is there something really obvious I'm missing, here? > Does the index really have to be named "t_x". Can you make up some nonsense name containing a lot of randomness - a name that is very unlikely to exist in the current schema? Like say: "t_x_b27289506cc846621a648dc9e95bba85". SQLite doesn't care what the name of the index is, unless you use the "INDEXED BY" clause on a query. Are you doing that? -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
