how can 0.00665 not equal 0.00665??!!

OK, I must be missing something. How can 0.00665 not equal 0.00665?

I hard code mytemp = 0.00665;
then i read in with cin >> searchValue to be 0.00665

then I test to see if they are equal ( both are double ) and
low and behold the logic says that mytemp is < searchValue!!!!

Even in debug both numbers are identical yet they fail as equal in the logic test:
if ( mytemp == searchValue)...

I will attach the .cpp file.

If I hard code both values into the code then they are equal.

What in the world could be causing this? I have spent all day looking at this and I am at a loss.

Thanks,
Susan

When I run it, the program says that they are equal. I'm using Dev C++ 4. Must be something wrong with your computer. Try restarting it, and if that doesn't work, try kicking it really hard. That always does the trick.

I was afraid of just the answer you gave! I rebooted and still had the same error. Does this mean that my compiler is corrupted in some way and I have to reinstall my compiler?
Susan

Hmm, I don't know what to do except to beat the crap out of your computer. Nice, firm kicks to the computer usualy do the trick for me! :D

Ok ... don't trash your comp just yet. I recompiled this program in visual C++ and had the same problem.

After much frustration I determined the problem.

Keep in mind i'm a noobie, but after ripping the program apart and writing one exactly like it of my own, I could find no difference between your program and mine. Except mine ran correctly.

Then I noticed the one difference. I was not using a namespace in my program.

So I removed the namespace and replaced <iostream> with <iostream.h>, recompiled your program and it ran just fine.

I don't know much about how namespaces work, so eh...

But I hope this helps.

interesting. don't quite know what to make of your observations. i'll make the similar changes and see what develops.

Well, i get the same as you, it works.

guess we need some name space specialist to explain what is going on.

Much thanks for giving me some of my sanity back!

susan

And thanks for sharing your insanity ;)

I actually learned the while loop on the fly.

Yes it would be nice to have namespaces explained in detail. But my best guess is that the namespace treated searchValue as 2 seperate instances of the same variable name.

Just a guess.

> OK, I must be missing something. How can 0.00665 not equal 0.00665?
You cannot compare floats for equality (EVER)

floats are approximations to actual numbers, so whilst you might see only 0.00665, what the machine is actually trying to compare is 0.00665500001 and 0.00665500002
This is why they appear to be different.

You can do two things
1. multiply both floats by (in this example) 10000 and assigning the results to integer variables, so you end up with comparing 665 with 665 (which will always work)

2. subtract the two floats, and check that the difference is small, like so
if ( fabs(a-b) < 0.00001 ) {
&nbsp;&nbsp;&nbsp;&nbsp;// close enough
}

er...

that's interesting math

are floats and doubles the same?

does that really apply to this situation, neither of the values were a result of any computation

one was hard coded and the other was user input, so i'm assuming the cpu will take them at face value

why did it work when the namespace was removed?

well, i did that, sort of. my mistake perhaps was that I returned the value back to float after I rounded to the precision I wanted.

I have another post by susan that is a few below this one and the topic is on rounding floating point numbers.

my floating points, while i view them in debug, both *look* like
0.006650000000
both are exactly the same.
yet when they are logically compared something is amiss.

// interpolating a sorted array
tryHere = lb + ( ( searchVal - dvlb ) / ( dvub - dvlb ) ) * ( ub - lb );

// my attempt to round
temp = ( array [ tryHere ] * dmill );
iarrayRounded = int ( floor ( temp + 0.50 ))
arrayRounded = iarrayRounded / 100000.0; // back to float

I will try and convert to int and compare that way.

Tried it with integers and of course it works just fine. Feel dumb but at least i can go to bed and get some rest.

My family thanks you, as do I,
susan


susan

> are floats and doubles the same?
doubles have many more bits
Look in float.h

> one was hard coded and the other was user input
User input involves computation - lots of multiply by 10.0 and add another digit kinda stuff

> why did it work when the namespace was removed
Pass