 |
 |
Archives of the TeradataForum
Message Posted: Thu, 16 May 2002 @ 18:56:53 GMT
| Subj: | | Re: Rounding in Teradata |
| |
| From: | | Geoffrey Rommel |
Calvin Kline wrote:
| | The main "problem" is the exact decimal (base 10) fractions are repeating fractions when represented in the binary system (base 2).
And the binary representation does not round in the same way as the decimal representation. | |
This is true for binary floating-point numbers (e.g., IEEE or S/390 floating-point). It is not true for decimal fields in Teradata,
since they are being stored with full precision. Division will always present difficulties, since in any base you choose some rational
numbers will require repetends. The Web page you pointed out, however, is very good. (It's by Mike Cowlishaw, the inventor of REXX.)
--wgr
| |
