Rounding to the Nearest Integer

[chris319]

Here is a little routine for rounding off numbers to the nearest integer:

Code: Select all

Procedure Roundoff(number.f)
integer.f = Int(number)
frac.f = number - Int(number)
If number >= 0  
  If frac >= 0.5
    rounded.f = integer + 1
  Else
    rounded = integer
  EndIf
Else
  If frac <= -0.5
    rounded = integer - 1
  Else
    rounded = integer
  EndIf
EndIf  
ProcedureReturn rounded
EndProcedure

[traumatic]

Same result but should be faster:

Code: Select all

Procedure.l Roundoff(number.f)
  If number - Int(number) >= 0.5
    ProcedureReturn number + 0.1
  EndIf
  
  ProcedureReturn number
EndProcedure 

[Michael Vogel]

Hmm...

I would do it like this :

Code: Select all

Procedure.l Roundoff(number.f) 
  If number<0
    ProcedureReturn Int(number-0.5)
  Else
    ProcedureReturn Int(number+0.5)
  EndIf
EndProcedure 
 

Debug Roundoff(3.1)
Debug Roundoff(3.8)
Debug Roundoff(-3.1); !!!

[#NULL]

and as macro

Code: Select all

Macro Roundoff(_number_)
  Int((_number_) - (0 Or (_number_)<0) + 0.5)
EndMacro
 

Debug Roundoff( 3.2) ;  3
Debug Roundoff( 3.8) ;  4
Debug Roundoff(-3.2) ; -3
Debug Roundoff(-3.8) ; -4

Debug Roundoff( 3)   ;  3
Debug Roundoff(-3)   ; -3

btw:
if i change this

Code: Select all

Int((_number_) - (0 Or (_number_)<0) + 0.5)

to this

Code: Select all

Int((_number_) + 0.5 - (0 Or (_number_)<0))

i get wrong results. why

[Matt]

Shouldn't this function be built-in?
I see Round in the manual but it takes an argument to round up or down, there should be one to round up if >=.5 or down otherwise

[Fred]

Yes, it's added for 4.20.

[Matt]

Yes, it's added for 4.20.

[Psychophanta]

@Michael Vogel & @Traumatic & @chris319 & @#NULL:

Code: Select all

Procedure.l Roundoff(number.f)
  ProcedureReturn number
EndProcedure

Debug Roundoff(3.1)
Debug Roundoff(3.8)
Debug Roundoff(-3.1); !!!

Simple

[Blue]

Code: Select all

Procedure.l Roundoff(number.f)
  ProcedureReturn number
EndProcedure

Simple

Bravo, Psychophanta. Very well done.

As you say, so simple...
Yet, I'd never have thought of it all by my lonesome self.

[Michael Vogel]

One point seems to be still a problem, the lack of precision with binary periodic numbers...

Code: Select all

a.f=1.250000; b.d=1.250000
c=a*2
Debug c

c=1.4999*2
Debug c

...the same issue is seen also within the Roundoff macro, of course.

[citystate]

what's wrong with good old

Code: Select all

Int(a.f + 0.5)

[superadnim]

nothing is wrong, thats what my macro actually does and what I've been using since forever

[Michael Vogel]

Of course your codes are all fine

It's just the problem with 0 and 1 - I'm so negative ....

Code: Select all

Debug Roundoff (-1.25*2)

[traumatic]

@Michael Vogel & @Traumatic & @chris319 & @#NULL:

Code: Select all

Procedure.l Roundoff(number.f)
  ProcedureReturn number
EndProcedure

Debug Roundoff(3.1)
Debug Roundoff(3.8)
Debug Roundoff(-3.1); !!!

Simple

hehe, good one!