[DUG] Need help in format function

[John Bird]

Tax calculations are tricky, as there are rules about either truncating to 
whole dollars or down to whole cents.  These are not maths functions, you 
may have to make your own.

I have seen people do all sorts of tricks with formating floats as strings 
and truncating, and converting back but to my mind the best is to calculate 
in cents - which is what the rules are actually specifying.  ie convert the 
dollars to integer or int64 cents, do the calculation and convert the result 
back to float.   Otherwise you start getting oddities where Net + GST <> 
Gross when the amounts are printed to dollars and cents.

Here is an example with GST calculations - guaranteeing that Net + GST = 
Gross when printed as dollars and cents to 2 decimal places..:

//IRD guidelines are that GST calculated amount is rounded down to nearest 
cent
procedure xcGSTCalc(aGSTRatePercent:double;var aNetVal:Double;var 
aGSTVal:Double;var aInclVal:Double);
var
lGSTRate:int64;       //gst rate as cents per 100 dollars (ie per 10000 
cents)=GSTRate*100
lNetCents,lGSTCents,lInclCents:int64;
begin
  lGSTRate:=trunc(aGSTRatePercent*100);         //eg 1250
  lNetCents:=round(aNetVal*100);
  lGSTCents:=(lNetCents * lGSTRate) div 10000;  //1250 div 10000 = .125
                                                //note integer division 
truncates fractions
  lInclCents:=lNetCents+lGSTCents;
  //return to float - this should guarantee nothing after first 2 decimal 
places
  aNetVal:=lNetCents/100;
  aGSTVal:=lGSTCents/100;
  aInclVal:=lInclCents/100;
end;

similar rules apply to with tax calculations - there is usually some 
truncating to whole dollars or cents involved.

Incidentally this stuff fools the IRD too - for years the printed tax tables 
have had some errors because they didn't follow their own truncating and 
rounding rules for some calculations - and the printed tax tables did not 
agree with their online tax calculator which was correct.

John