`to_chars()` error with `chars_format::fixed` and defaulted precision

[k3DW]

This happens when the absolute value is less than 1/10.

This is fine

char buffer[1024]{};
auto num = decimal32{ 1000000, -7 };
to_chars(std::begin(buffer), std::end(buffer), num, chars_format::fixed);

This crashes in a memset() call

char buffer[1024]{};
auto num = decimal32{ 999999, -7 };
to_chars(std::begin(buffer), std::end(buffer), num, chars_format::fixed);

The error is coming from "charconv.hpp". It's trying to memset() with a size of static_cast<std::size_t>(-1), inside the function detail::to_chars_fixed_impl().

else if (num_leading_zeros > precision)
{
    *first++ = '0';
    *first++ = '.';
    std::memset(first, '0', static_cast<std::size_t>(precision));
    return {first + precision, std::errc()};
}

It looks like real_precision is defined at the top of the function but isn't used other than checking the buffer size. I'd hazard a guess that all uses of precision in this function should actually say real_precision.

[mborland]

Have you tried current develop? This looks similar to: #793

[k3DW]

It looks like I was a little out of sync. I hadn't pulled since the review started. Thanks!