division... strange behaviour i dont understand

[clqrq]

can please someone explain why i get the message-box?
im damn shure 500 divided by 300 doesnt equal to zero ;-\

__________________________________
unsigned short uiNumbers = 500;
unsigned float ufFactor = 300.0;
ufFactor = ufFactor / uiNumbers;
if (ufFactor == 0) AfxMessageBox("0");

 

[Ian Collins]

can please someone explain why i get the message-box?
im damn shure 500 divided by 300 doesnt equal to zero ;-\

__________________________________
unsigned short uiNumbers = 500;
unsigned float ufFactor = 300.0;

What's an unsigned float?

 

[Murali Krishna]

What's an unsigned float?

yes, unsigned float that is not compiling on my compiler.

ufFactor = ufFactor / uiNumbers;

any way, that's not 500 / 300.

it is 300 / 500. Should lead to 0.6 if ufFactor is float.
and 0 if it is integral.

-- Murali Krishna

 

[Zara]

can please someone explain why i get the message-box?
im damn shure 500 divided by 300 doesnt equal to zero ;-\

__________________________________
unsigned short uiNumbers = 500;
unsigned float ufFactor = 300.0;
ufFactor = ufFactor / uiNumbers;
if (ufFactor == 0) AfxMessageBox("0");

unsigned float?

Anyway, it seems thta ufFactor (which should have the value 300.0/500)
is converted to int (thus, rounded to 0) before being compared.

Try :

if (ufFactor == 0.0) AfxMessageBox("0");

This is an tricky comparison, as comparison with doubles (and floats)
is subject to lots of problems due to rounding. You should take a look
at this list, siblig list comp.lang.c++.moderated and cousin lists
comp.lang.c and comp.lang.c.moderated about floating point comparisons

Regards,

Zara

 

[wanzi8552]

can please someone explain why i get the message-box?
im damn shure 500 divided by 300 doesnt equal to zero ;-\

__________________________________
unsigned short uiNumbers = 500;
unsigned float ufFactor = 300.0;
ufFactor = ufFactor / uiNumbers;
if (ufFactor == 0) AfxMessageBox("0");

Because unsigned float is not a defined type in C++, the ufFactor will
then be automatically converted to unsigned int.

P.S. Any decent compiler should be able to spot the problem and give
some warnings.

 

[Ian Collins]

Because unsigned float is not a defined type in C++, the ufFactor will
then be automatically converted to unsigned int.

No, it's just plain wrong and should fail to compile.

 

[Pete Becker]

unsigned float?

Anyway, it seems thta ufFactor (which should have the value 300.0/500)
is converted to int (thus, rounded to 0) before being compared.

If ufFactor is really some floating point-type, then 0 would be
converted to a floating-point type, not the other way around.

Try :

if (ufFactor == 0.0) AfxMessageBox("0");

This is an tricky comparison, as comparison with doubles (and floats)
is subject to lots of problems due to rounding.

There's nothing tricky involved in checking whether a value is 0, and
there are no floating-poitn roundoff issues in this code. Most likely,
"unsigned float" was actually "unsigned int".

 

[Jim Langston]

can please someone explain why i get the message-box?
im damn shure 500 divided by 300 doesnt equal to zero ;-\

No, but 300 divided by 500 is 0 in integer math.

 

[Victor Bazarov]

No, but 300 divided by 500 is 0 in integer math.

WTF is "unsigned float"?

V

 

[Jim Langston]

WTF is "unsigned float"?

A syntax error in his code which has already been discussed.

 

[Victor Bazarov]

A syntax error in his code which has already been discussed.

Ah... Thanks. Sorry to butt in at the middle of the thread.