On 11/30/16, Chris Locke <chrisjlo...@gmail.com> wrote: > I recently had this problem. Values stored as real values. Had to check > records in the database to see if any value had changed, and needed > updating. Even though all values in my code were singles, I had bad > rounding problems where (as an example) 0.1+2.2 did not equal 2.3 in the > database. Aargh. > Storing as integers is the way to go.
Just to be clear, this is a property of binary floating-point numbers, not a quirk of SQLite. Using the IEEE 64-bit floating point format, there is no way to represent values 0.1, 2.2, and 2.3. The closest you can get are the following: 0.1: 0.1000000000000000055511151231257827021181583404541015625 2.2: 2.20000000000000017763568394002504646778106689453125 2.3: 2.29999999999999982236431605997495353221893310546875 If you add the first numbers you get: 2.300000000000000266453525910037569701671600341796875 which is not equal to the third number. Points to remember: (1) Floating point numbers are usually approximations, not exact values. (2) Never compare two floating point numbers for equality -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
