Incorrect printing of float/double in 7.1.1

[madscientist]

I'm seeing a problem printing with {:f} in 7.1.1.

Here is the test program:

#include <fmt/core.h>

int main()
{
    double dbl = (double)9223372036854775807;
    fmt::print("double {:f}\n", dbl);
    return 0;
}

In 7.0.3 I get reasonable results:

double 9223372036854775808.000000

In 7.1.1 I get something I can't understand at all:

double B446744073709551616.000000

This is on a GNU/Linux system compiled with GCC 10.2.

[madscientist]

I tried 7.1.0 and see the same (incorrect) results.

[madscientist]

Bisecting shows this change in behavior was introduced by SHA 9f312fe

commit 9f312fe87e476d2ef91d347a8d1cd5f7dd9cb730
Author: Victor Zverovich <[email protected]>
Date:   Mon Sep 7 09:34:05 2020 -0700

    Implement fallback FP formatting with given precision (#1526)

[vitaut]

Good catch, thanks! Fixed in 6b7bfed.