what wronge with using floor(+0.5) as round?

Question

4.00/5 (1 vote)

See more:

I have to round value and store it, because the c++ does not has round function I use this:

C++

floor(x+0.5)

but when I use it in my code it does not work:

C#

void algo(float xx1, float yy1, float xx2, float yy2, float xx[], float yy[])
{
   double mm, m1, m2;
   m1 = yy2 - yy1;
   m2 = xx2 - xx1;
   mm = m1 / m2;
   yy[0] = yy1;
   xx[0] = xx1;

   if((mm>=-1)&&(mm<=1))
   {
      for(int i=0; i<=xx2; i++)
      {
         yy[i + 1] = floor(yy[i] + mm + 0.5);
         xx[i + 1] = xx[i] + 1;
      }
   }
   else
   {
      for(int i=0; i<=xx2; i++)
      {
         xx[i + 1] = floor(xx[i] + (1/mm) + 0.5);
         yy[i + 1] = yy[i] + 1;
      }
   }
}

Posted 13-Apr-12 1:25am

sami-syria

Updated 13-Apr-12 9:14am

fjdiewornncalwe

v2

Add a Solution

Comments

[no name] 13-Apr-12 8:31am

What do you mean by does not work? Can you give examples of input and output?

sami-syria 13-Apr-12 8:57am

this function is part of lines drawing algorithms
when run this code without floor with this in put:xx1=2,yy1=2 xx2=7,yy2=5
I get this : xx=2 3 4 5 6 7
yy=2 2.6 3.2 3.8 4.4 5
but when I use floor I get this:
xx= 2 3 4 5 6 7
yy= 2 3 4 5 6 7

thank you

4 solutions

Solution 4

thank you all for your kindness the problem was solved...the program has 3 functions for 3 line drawing Algorithms: simple, digital differential analyzer and Bresenham's Algorithm to compare the execution time for each one.

thank you again

Posted 15-Apr-12 11:44am

sami-syria

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

Espen Harlinn · Accepted Answer · 2012-04-13T02:16:00

Solution 1

You'll find a pretty good answer here:
http://www.cplusplus.com/forum/articles/3638/[^]

Best regards
Espen Harlinn

Posted 13-Apr-12 2:16am

Espen Harlinn

Comments

CPallini 13-Apr-12 8:22am

Good one, 5.

Espen Harlinn 13-Apr-12 8:24am

Thank you Pallini :-D

sami-syria 13-Apr-12 8:59am

Thank you for your interest sir
i well try it

VJ Reddy 14-Apr-12 21:23pm

Good reference. +5

Espen Harlinn 15-Apr-12 4:39am

Thanks VJ :-D

sami-syria 15-Apr-12 17:13pm

thank you Mr.Espen it was very useful link.

Espen Harlinn 15-Apr-12 17:23pm

Well, glad you found it useful :-D

User 59241 · Accepted Answer · 2012-04-13T08:48:00

Solution 2

If you put this in the debugger which is the FIRST thing to do when debugging a program you will find floor() is working properly.

The line

yy[i+1]=floor(yy[i]+mm+0.5);

does completely different things when you use floor() because yy[i+1] is dependent on yy[i].

You are generating a different series altogether.

Your algorithm is incorrect and needs work.

Posted 13-Apr-12 8:48am

User 59241

Updated 13-Apr-12 8:49am

v2

Comments

Aescleal 14-Apr-12 5:33am

All the talk of rounding and Bresenham's algorithm aside it's nice that someone spotted a potential code error. I was sitting here thinking "hang on, what he's trying should work."

A 5 to you Sir.

[no name] 14-Apr-12 8:47am

Thanks for your kind remarks.

Espen Harlinn 15-Apr-12 18:37pm

Oh, I'm nearly blushing - I feel it's a bit embarrassing that I didn't spot that one, because now that you've mentioned it - it kind of leaps of the screen :)
Nicely spotted!

[no name] 15-Apr-12 20:10pm

Thanks - most bugs are like that.

nv3 · Accepted Answer · 2012-04-13T08:51:00

Espen has already contributed a good link regarding the rounding. Thanks Espen, that got my 5.

The reason I post this solution is to make you aware that there is better algorithm for line drawing that you might not be aware of. It was invented by Jack E. Bresenham and is named after him.

Bresenham's Algorithm in Wikipedia

The nice thing about this algorithm is that it runs completely with integer operations. And that adds tremendously to its speed. Every conversion between floating-point format and integer is always a relatively slow operation. Or in your case you need a floating-point add (+0.5) a function call to floor, and probably later on when it comes to the real drawing a conversion to int. So for serious graphics work I would suggest you take a look at Bresenham's algorithm. It's not hard to convert the pseudo-code given on the Wiki page into C code.

Back to the subject of rounding: I have found the implicit conversion from double to int by the C runtime system sometimes a very slow operation. That was true VC++ V6; don't know if that is still so in newer versions. As a good work-around I found somewhere on CodeProject the following code (sorry that I cannot give credit to the author; might have been the following article: Floating-Point utilities):

inline long RoundInt (double a)
{
    long retval;
    __asm fld a
    __asm fistp retval
    return retval;
}

This little inline function translates into only two machine instructions and is according to my measurements way faster than floor (x + 0.5) and the following implicit conversion to int.

Hope that helped you a little further along the way.

what wronge with using floor(+0.5) as round?

4 solutions

Solution 1

Solution 2

Solution 3

Solution 4

Add your solution here

Preview 0

Existing Members

...or Join us