Heenan, Lambert
Lambert.Heenan at AIG.com
Fri May 2 08:28:53 CDT 2008
<quote> # VB6 Round rounds 1/2 to nearest the even number (aka 'Banker's rounding') Round(1.5) => 2 Round(2.5) => 2 !! should be 3 in 'Other people's rounding' </quote> The "should be 3" comment gets me. Why would anyone wish to use a rounding method that skews the results in one direction? For once the bankers might have it right, IMHO. By alternately rounding up and down in a systematic way you reduce skew. Which begs the question as to why Ccur() gets it wrong... ? ccur(111.11115),ccur(1111.11115), ccur(11111.11115) 111.1111 1111.1111 11111.1112 ? ccur(111.11105),ccur(1111.11105), ccur(11111.11105) 111.1111 1111.111 11111.111 Lambert -----Original Message----- From: accessd-bounces at databaseadvisors.com [mailto:accessd-bounces at databaseadvisors.com] On Behalf Of Gustav Brock Sent: Friday, May 02, 2008 2:57 AM To: accessd at databaseadvisors.com Subject: Re: [AccessD] Rounding with CCur conversion Hi Asger First, this: CCur(111.11115) returns here 111.1112 as expected. Don't know why you get a different result. Then, the CCur, CInt, CLng and CByte functions perform Banker's rounding while Int and Fix don't. Further, Round is known to be buggy, so avoid that for any serious purpose. If you are interested in rounding and how to optimise this regarding speed while still being accurate, read on here: http://www.xbeat.net/vbspeed/c_Round.htm Note that the absolutely simplest (but slowest) method to perform 100% correct (mathematical) rounding is ... Format! /gustav >>> ab-mi at post3.tele.dk 02-05-2008 01:55 >>> Hi group, I just noticed that the algorithm in VB for rounding seems to be different for the Round-function and the CCur-function. The Round-function use "Banker's rounding", also known as "Dutch rounding". The CCur-function behaves differently: Round(1.11115,4) --> 1.1112 Round(11.11115,4) --> 11.1112 Round(111.11115,4) --> 111.1112 CCur(1.11115) --> 1.1112 CCur(11.11115) --> 11.1112 CCur(111.11115) --> 111.1111 ?!! What's going on in the last conversion? Is this a bug or does the CCur-function have a special algorithm? TIA Asger -- AccessD mailing list AccessD at databaseadvisors.com http://databaseadvisors.com/mailman/listinfo/accessd Website: http://www.databaseadvisors.com