https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126731
Bug ID: 126731
Summary: `std::format("{}", std::float16_t)` does not agree
with `std::to_chars`
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: libstdc++
Assignee: unassigned at gcc dot gnu.org
Reporter: 121539739 at qq dot com
Target Milestone: ---
For a floating-point argument with no presentation type and no precision,
[format.string.std] specifies the output as:
> equivalent to `to_chars(first, last, value)`
so `std::format("{}", v)` and `std::to_chars(first, last, v)` have to produce
the same characters. For `std::float16_t` and `std::bfloat16_t` they do not.
```cpp
#include <cassert>
#include <charconv>
#include <format>
#include <iterator>
#include <stdfloat>
#include <string>
int main() {
std::float16_t h = 0.1f16;
char buf[32];
auto [ptr, ec] = std::to_chars(buf, std::end(buf), h);
assert(ec == std::errc{}); // ok
assert(std::format("{}", h) == std::string(buf, ptr));
// "0.099975586" "0.1" fail
}
```
detailed: https://godbolt.org/z/WeYd3o5fh
