On Tue, Jul 10, 2012 at 11:33 AM, Igor Tandetnik <[email protected]>wrote:
> On 7/10/2012 11:24 AM, Ryan Johnson wrote: > >> On 10/07/2012 9:58 AM, >> [email protected] wrote: >> >>> SQLite version 4.0.0 2012-07-07 12:21:48 >>> Enter ".help" for instructions >>> Enter SQL statements terminated with a ";" >>> sqlite> create table t(a); >>> sqlite> insert into t values(123); >>> sqlite> insert into t values(123.0); >>> sqlite> insert into t values(12323); >>> sqlite> insert into t values(12323.0); >>> sqlite> select * from t group by 1; >>> 123.0 >>> 12323 >>> 12323.0 >>> sqlite> >>> >> This is correct: 123 != 123.0 != "123" because they all have different >> types (int/float/string). >> > > Could you explain why 123 and 123.0 got grouped together, but 12323 and > 12323.0 did not? Shouldn't the same logic apply? It's this inconsistency > that looks qutie suspicious. > SQLite4 is still using "double" internally for floating point computations. (Yes, there are plans to change that but it has not happened yet.) The bug above is due to rounding errors in the doubles, specifically rounding errors that occur when doing the key encoding. > -- > Igor Tandetnik > > > ______________________________**_________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**users<http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users> > -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
