BorlandTalk.com

Johan Elmsäter · Guest

I have a problem with double when adding decimal values.

the doubles are assigned with:
double start = (double) StrToFloat(TEdit1->Text);
double inc = (double) StrToFloat(TEdit2->Text);
double stop = (double) StrToFloat(TEdit3->Text);
start could be 0.001 and inc the same. Stop is for example 0.009

When I add inc to start the value isn't 0.001 it could be something like
0.00100000001 instead
and when comparing the value with stop when it should be 0.009 instead it is
0.009000005

What can I do to fix this problem???

Regards Johan

Analian · Guest

Don't relay on double's valus being exactly what you expect (:
Instead of the standart comparing routines you could use:

const double eps = 0.0000001; // you can coustomly choose the precision

// instead of ==
int equal (double a, double b) {
return fabs (a - b) < eps;
}
// instead of >
a > b transforms in a > b + eps
// instead of <
a < b transforms in a < b - eps

Hope this helps.

Analian

qyte · Guest

Johan Elmsäter wrote:

Palle Meinert · Guest

The VCL has a function suited for comparing floating points values. Check
the SameValue function.

/Palle

Johan Elmsäter · Guest

Thanks for all your answers. I have taken a deeper look at it...
and the following code:

double start = 0.01;
double stop = 0.3;
double step = 0.01;
double current = (double) start;
printf("start %20.20g, current: %20.20g, stop: %20.20grn
",start,current,stop);

Generates the following output:
start 0.010000000000000000208, current: 0.010000000000000000208, stop:
0.2999999
999999999889

I have known about problems with floating point values for a long time but
that you can't
make a simple assignment correctly is just too much...

But thanks for your answers though!
Best Regards Johan

"Palle Meinert" <No (AT) Spam (DOT) com> wrote

qyte · Guest

Johan Elmsäter wrote:

Remy Lebeau (TeamB) · Guest

"Johan Elmsäter" <johan.elmsater (AT) bulletinboard (DOT) nu> wrote

Johan Elmsäter · Guest

I changed the code and added my whole testexample and it still doesn't work.
When the loop is finished current should have the value 0.31 but it doesn't
it has the
value of 0.3000000000000000999
The <= operator shows that the textoutput is probably right!

TestDouble() {
double start = 0.01;
double stop = 0.3;
double step = 0.01;
double current = (double) start;
printf("start %20.20lg, current: %20.20lg, stop: %20.20lgrn
",start,current,stop);
while(current <= stop) {
current += step;
printf("current: %20.20lg, stop: %20.20lgrn ",current,stop);
}
}

"Remy Lebeau (TeamB)"