Not sure about replacing the collation sequence - does not sound easier than recreating the table.
You could just add a view: sqlite> CREATE TABLE test_table (ID INTEGER PRIMARY KEY, ...> ExternalID2 INTEGER, ...> ExternalID INTEGER, ...> Value ); sqlite> INSERT INTO "test_table" VALUES(1007,1,37,'-5'); sqlite> INSERT INTO "test_table" VALUES(1044,4,37,'-10'); sqlite> INSERT INTO "test_table" VALUES(1081,2,37,'-20'); sqlite> INSERT INTO "test_table" VALUES(1118,3,37,'-1'); sqlite> INSERT INTO "test_table" VALUES(1155,5,37,'-7'); sqlite> INSERT INTO "test_table" VALUES( 2044,4,37,'fred'); sqlite> INSERT INTO "test_table" VALUES( 3044,4,37,'bill'); sqlite> sqlite> create view test_view as ...> select ID, ExternalID2, ExternalID, cast( value as integer ) as Value ...> from test_table where cast( Value as text)=cast(Value as integer); sqlite> sqlite> select * from test_view; 1007|1|37|-5 1044|4|37|-10 1081|2|37|-20 1118|3|37|-1 1155|5|37|-7 sqlite> sqlite> select min( Value ) from test_view; -20 sqlite> select max( Value ) from test_view; -1 Rgds, Simon 2008/8/15 Dennis Volodomanov <[EMAIL PROTECTED]>: > >> Declaring the column as integer does not prevent you from storing strings: >> >> > Yes, except for a small problem of updating all live databases with the > new column type. I don't think I can update the column type without > recreating the table, right? It's not hard, so if it comes down to this, > then I guess I'll do it. Or if replacing the collation sequence is not > too hard, I'd rather go that route. > > Thank you, > > Dennis > > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
