Announcement

**Edwin Knoppert** · 9 Jun 2008, 03:34 PM

DEFLNG A-Z... one of the most terrible BASIC statements ever..

Y is a long integer which can not hold decimals.
Local Y as Double would do..
VAL() returns an Extended-currency (@@)

**Paul Purvis** · 9 Jun 2008, 04:37 PM

removed

**Michael Mattias** · 9 Jun 2008, 04:50 PM

For money?

How about...

Code:

N = ROUND(M, 2)

?????

**Edwin Knoppert** · 9 Jun 2008, 04:51 PM

Not using # or @@ in the values .5 and 10 may have impact on results!
Eventually make it a variable or assign the symbols.

I also had trouble with color calculation where the result is a quad.

2 ^ ... for example is not good enough and is an overflow.

**Paul Purvis** · 11 Jun 2008, 01:43 AM

my math is broken
i need to multiply F times the number of decmials of variable B
see the line "my math is broken"
this program does not work yet for negative numbers

i am trying to round for fraction of 5 up
it does not have be money either

just another rounding routine

Code:

#COMPILE EXE
#DIM ALL

FUNCTION MONEYROUND(BYVAL A AS EXT,BYVAL B AS LONG) AS EXT
LOCAL D AS QUAD
LOCAL C AS QUAD
LOCAL E AS QUAD
LOCAL F AS QUAD
D=(10^(B+2))
E=(FIX(((FIX(D*A)) +(SGN(A)*(5&&)))))\10
PRINT E
F=((E-(E-FIX(E*.1)*10))\10)*10
PRINT F
IF E-F>4 THEN F=F+10
PRINT F
FUNCTION=F/(FIX(10^(b+1)))'*SOMETHING HERE    my math is broken
END FUNCTION

FUNCTION PBMAIN () AS LONG
LOCAL M AS EXT
LOCAL N AS EXT
M=12.15555
N=MONEYROUND(M,4)
PRINT N
WAITKEY$

END FUNCTION

added:
removed variable D and C, and placed code for negative numbers

THIS IS STILL CODE BEING TESTED

Code:

#COMPILE EXE
#DIM ALL

FUNCTION MONEYROUND(BYVAL A AS EXT,BYVAL B AS LONG) AS EXT
LOCAL E AS QUAD
LOCAL F AS QUAD
LOCAL isign AS LONG
isign=SGN(a)
IF isign=-1 THEN a=ABS(a)
E=(FIX(((FIX((10^(B+2))*A))+5&&)))\10
PRINT E
F=((E-(E-FIX(E*.1)*10))\10)*10
PRINT F
IF E-F>4 THEN F=F+10
PRINT F
'if statement to allow for a negative number
IF isign=-1 THEN
  FUNCTION=(F/(FIX(10^(b+1))))*-1'*SOMETHING HERE    my math is broken
  ELSE
  FUNCTION=F/(FIX(10^(b+1))) '*SOMETHING HERE    my math is broken
END IF
END FUNCTION

FUNCTION PBMAIN () AS LONG
LOCAL M AS EXT
LOCAL N AS EXT
M=-12.15555
N=MONEYROUND(M,4)
PRINT N
WAITKEY$

END FUNCTION

**John Gleason** · 11 Jun 2008, 12:18 PM

Paul, this wasn't as straightforward as I was thinking it would be.

yow. Also, this rounds negatives more negative. Is that correct? It can be modified easily to round them less negative too.

see new code here

Code:

#COMPILE EXE
#DIM ALL

FUNCTION roundUp(BYVAL A AS EXT, BYVAL B AS LONG) AS EXT
   LOCAL x AS EXT, neg AS LONG

'more bugs found, needs more testing. Will repost later...
Reposted at link above 
[URL="http://www.powerbasic.com/support/pbforums/showthread.php?p=287657#post287657"]here[/URL]

END FUNCTION

**Guy Dombrowski** · 11 Jun 2008, 09:26 PM

Rounding

Since VAL($) will give an extended Currency value why not use it directly ?
The rounding will be automatic and always right for both positive and negative fractions.

Code:

#COMPILE EXE
FUNCTION PBMAIN () AS LONG

 m$="123.4567"
 n$="-123.4567"
 [email protected]@=VAL(m$)
 [email protected]@=VAL(n$)
 PRINT [email protected]@,[email protected]@
 WAITKEY$

END FUNCTION

**Paul Purvis** · 12 Jun 2008, 12:34 AM

i first was working with the dollars and it would be great(fantastic) to have a money rounding function, but it seemed to me being a younger pb programmer(it is nice to be younger at something) that maybe i should try my hand at floating point rounding for many different financial math routines.
there are many math functions dealing in the money that have to be very correct such as the APR on borrowed money that is Federally regulated and the there is not much tolerance in the results that can be computed.
and a lot of those computations that can not be done with a currency variable, or at least, i would think so from previous experience in calculating the APR.
also there are also laws with certain limits that can not be exceeded in many financial situations and i can see where using currency only values in calculations would make some results unacceptable.

i have been so brain dead from working on too many projects, i need to take the F variable and reduced it to certain decimal places after getting it to a whole number with out a loop.

for those looking at my math i could possibly had used the MOD statement to get the 1st digit between 0 and 9, but without testing, i thought maybe some other math might have been faster, but maybe not.

so can some body help with this

i might of just found the fix
i need somebody to shoot this routine out the sky
i placed code that should allow for negative numbers

FUNCTION=F/(FIX(10^(b+1)))

i made the change above.

**Paul Purvis** · 12 Jun 2008, 01:08 AM

GUY
sorry
your function will fail
using 123.455, the 5 is going to be the problem and i did not understand the 10 as mentioned earlier until today

values returned should be 123.46
but you will get will be 123.45

Code:

#COMPILE EXE
FUNCTION PBMAIN () AS LONG

 m$="123.455"
 n$="-123.455
 [email protected]@=VAL(m$)
 [email protected]@=VAL(n$)
 PRINT [email protected]@,[email protected]@
 WAITKEY$

END FUNCTION

added:
my code failed too
use 12.1149
rounded to two decimal, resulting with 12.12

**Michael Mattias** · 12 Jun 2008, 06:58 AM

I do some floating point rounding in this post:
Telco Benchmark Program 5-20-04

**John Gleason** · 12 Jun 2008, 07:03 AM

Paul, it's funny but that's the reason my code was failing too, the 49's rounding up. I finally got it to work in every case of two decimal places. Now to get those pesky other few thousand.

**Guy Dombrowski** · 12 Jun 2008, 07:43 AM

Rounding

Paul,

What version of PBCC are you using ?
My rounding results are correct.

Code:

#COMPILE EXE
FUNCTION PBMAIN () AS LONG

 m$="123.455": n$="-123.455": o$="12.1149"
 [email protected]@=VAL(m$): [email protected]@=VAL(n$): [email protected]@=VAL(o$)
 PRINT [email protected]@,[email protected]@,[email protected]@
 'Result shown are : 123.46  -123.46  12.11
 WAITKEY$

END FUNCTION

**John Gleason** · 12 Jun 2008, 07:52 AM

Originally posted by Michael Mattias View Post

I do some floating point rounding in this post:
Telco Benchmark Program 5-20-04

Michael, for p = ROUND(p,2), is it a compiler bug that even digits preceding the 5 round down, for instance, .185 rounds to .18, and odds round up, eg. .475 rounds to .48? PB ver. 8.04 differs from your .065 rounding to .07 result (8.04 rounds it to .06)

**John Gleason** · 12 Jun 2008, 08:02 AM

Guy, try the following:

Code:

#COMPILE EXE
FUNCTION PBMAIN () AS LONG

 m$="123.465": n$="-123.465": o$="12.1149"
 [email protected]@=VAL(m$): [email protected]@=VAL(n$): [email protected]@=VAL(o$)
 PRINT [email protected]@,[email protected]@,[email protected]@
 'Result shown are : 123.46  -123.46  12.11
 WAITKEY$

END FUNCTION

Results should be 123.47 -123.47 12.11

**jcfuller** · 12 Jun 2008, 09:19 AM

I'm Not sure this is relevant.

James

Rounding - Wikipedia

http://en.wikipedia.org/wiki/Rounding#Round-to-even_method

**Paul Purvis** · 12 Jun 2008, 04:11 PM

James to me, that was very helpful.
i guess with what i am reading.
the only way for me then maybe is just to walk the digits furtherest from the right back to the desired digit.

**John Gleason** · 12 Jun 2008, 06:44 PM

Redo of earlier post with extensive test code built in. PB ver. 8.04. Sry, I don't have CC. Try your technique Paul, and the two versions can be tested against each other for accuracy.

Code:

#COMPILE EXE
#DIM ALL
'-----------------------------------------------------------------------
'Rounds last digit of 5 always up, eg. 24.425 rounds to 24.43 if B = 2. Runs in ~ 415 ticks
'-----------------------------------------------------------------------
FUNCTION roundUp(BYVAL a AS EXT, BYVAL b AS LONG) AS EXT
   LOCAL x, y AS EXT, neg AS LONG
   LOCAL z AS QUAD       '<< has to be quad! if it's EXT, z MOD 5 = 5 occasionally. Don't know why.
                            
   IF a < 0 THEN
      a = -a
      neg = 1
   END IF

   x = ROUND(a, b)      'in 8.04 (and probably most recent CC ver.) round to even, or banker's rounding
   IF x >= a THEN
      IF neg = 1 THEN x = -x
      FUNCTION = x
      EXIT FUNCTION
   END IF

   y = 10 ^ (b + 1)
   z = a * y            'move decimal point to end
                   
   IF z MOD 5 = 0 THEN  'see if last digit is a 5
      x = z + 5         'round up if it is
      IF neg = 1 THEN
         FUNCTION = ROUND(-x / y, b)
      ELSE
         FUNCTION = ROUND( x / y, b)
      END IF
      EXIT FUNCTION
   END IF

   IF neg = 1 THEN
      FUNCTION = -x     'use from ROUND calculation already done above
   ELSE
      FUNCTION = x
   END IF

END FUNCTION

FUNCTION PBMAIN () AS LONG
   LOCAL M AS EXT, rot, iter, ii AS LONG
   LOCAL n AS QUAD
   RANDOMIZE
                  
   FOR ii = 1 TO 100000000      'test a hundred million roundUp's against ROUND's to be quite sure it is correct
      INCR rot                  'rotator
                                'note: use only 1 of following 2 lines at a time for +/- testing.
      M = ROUND(RND * 100, rot) 'get a random  number with rotator mantissa digits, eg. if rot = 3,  46.321 might result
'     M = ROUND(RND * -100, rot)'get a -random number with rotator mantissa digits, eg. if rot = 3, -46.321 might result
      IF (iter AND &h0ffffff) = 0 THEN BEEP 'beep occasionally to let you know it's testing.
      INCR iter
                                         'note: use only 1 of following 2 lines at a time for +/- testing.
      IF roundUp(M, rot - 1) <= M THEN   'use to test for positive number errors
'     IF roundUp(M, rot - 1) > M THEN    'use to test for negative number errors
         n = 10 ^ (rot)
         IF M * n MOD 10 < 5.5 AND M * n MOD 10 > 4.5 THEN    'if last digit is a 5 then function has an error!
            ? STR$(M - roundUp(M, rot-1), 18) & $CRLF & STR$(M, 18) & $CRLF & STR$(roundUp(M, rot-1), 18) & _
              $CRLF & STR$(ROUND(M, rot-1), 18) & $CRLF & STR$(iter) & $CRLF & STR$(rot),48,"Error1"
         ELSEIF ROUND(M, rot - 1) <> roundUp(M, rot - 1) THEN 'they otherwise should be =, or again it's a error!
            ? STR$(M - roundUp(M, rot-1), 18) & $CRLF & STR$(M, 18) & $CRLF & STR$(roundUp(M, rot-1), 18) & _
              $CRLF & STR$(ROUND(M, rot-1), 18) & $CRLF & STR$(iter) & $CRLF & STR$(rot),48,"Error2"
         END IF
     END IF
    IF rot = 16 THEN rot = 0
   NEXT
END FUNCTION

**Guy Dombrowski** · 12 Jun 2008, 06:58 PM

Rounding

Paul,

You are right, my function did fail with those numbers.
I tought extended currency could do it automaticaly but I was wrong.

Here is a technique I used all the time with QB45 to round double precision numbers.

Code:

#COMPILE EXE
FUNCTION PBMAIN () AS LONG

m$="123.465": n$="-123.465": o$="12.1149"
[email protected]@=ABS(VAL(m$))+.0001: IF VAL(m$)<0 THEN [email protected]@[email protected]@*-1
[email protected]@=ABS(VAL(n$))+.0001: IF VAL(n$)<0 THEN [email protected]@[email protected]@*-1
[email protected]@=ABS(VAL(o$))+.0001: IF VAL(o$)<0 THEN [email protected]@[email protected]@*-1

 PRINT [email protected]@,[email protected]@,[email protected]@
 'Result shown are : 123.47  -123.47  12.11
 WAITKEY$

END FUNCTION

**Paul Purvis** · 13 Jun 2008, 01:59 AM

the method i suggested using a lengthy string, then rounding up to the decimal from right to left did not produce no better results that using the compiler built in round statment.
rethinking
paul

Announcement

VAL rounds a variable

VAL rounds a variable

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment