The others explained why. The following is the workaround:

for (my $i10=0; $i10<=100; $i10++) {
   my $i = $i10 / 10;
   print "$i, \n";
}
[download]

or the easier to read

for (0..100) {
   my $i = $_ / 10;
   print "$i, \n";
}
[download]

Always use integers for loop counters.

If you need to compare two decimal numbers, check out Comparing Reals using a Tolerance.

See Re: Re: Re: Bug? 1+1 != 2.

It's a standard problem when dealing with real numbers on a computer. Most real numbers can't be represented accurately using the binary representations used in computers. You'll therefore just get an approximation (as your example shows).

There are two options for dealing with it. Either stick with real numbers and make sure that everything is rounded to the correct number of decimal places before displaying it to the user or use integer arithmatic whenever possible.

See also perlfaq4.

for ( $i=0; $i<=10; $i+=0.1 ) {
    printf( "%3.1f\n", $i );
}
[download]

To expand on what others have said, a computer can only exactly represent numbers with an integer powers of 2 in the denominator (1/2, 1/4, 1/8, etc.). Any other denominator will result in a representation that continues indefinitely. The representation is either truncated or rounded, but in any event it is inexact. For example, 0.1 (decimal) is represented as 0.0001100110010011001... in binary. Much of the time, the tiny error is insignificant. However, these errors accumulate when you use repeated calculations, and can eventually cause complete loss of precision.