External/fmt
8
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

* Eliminate warnings

Browse Source 
* Possibly fix bizarre ambiguity on OSX's AppleClang for the detail::count_digits() function according to https://github.com/gabime/spdlog/issues/914
This commit is contained in:
Filip Matracki 2021-10-20 18:08:55 +02:00
parent 1067285501
commit 7cc94ba0bc
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
include/fmt/chrono.h
View File

@ -876,6 +876,8 @@ template <typename T> struct make_unsigned_or_unchanged<T, true> {
};
#if FMT_USE_CONSTEXPR
// Taken from MSVC <chrono> STL implementation
// Copyright (c) Microsoft Corporation 2021
template <class Duration> FMT_CONSTEXPR unsigned int get_fractional_width() {
static_assert(Duration::period::num > 0 && Duration::period::den > 0,
"Numerator and denominator can't be less than 1.");
@ -938,14 +940,16 @@ To fmt_safe_duration_cast(std::chrono::duration<FromRep, FromPeriod> from) {
#endif
template <class Duration> struct subsecond_helper {
static constexpr std::uintmax_t fractional_width =
static constexpr std::uint64_t fractional_width =
#if FMT_USE_CONSTEXPR
get_fractional_width<Duration>();
#else
width<Duration::period::num, Duration::period::den>::value;
width<Duration::period::num, Duration::period::den>::value > 18
? 6
: width<Duration::period::num, Duration::period::den>::value;
#endif
static constexpr std::uintmax_t fractional_width_pow10 =
static constexpr std::uint64_t fractional_width_pow10 =
#if FMT_USE_CONSTEXPR
pow10(fractional_width);
#else
@ -1204,11 +1208,12 @@ struct chrono_formatter {
const auto tmpval = std::chrono::duration<Rep, Period>(val);
#endif
using subsec_helper = detail::subsecond_helper<duration_Rep>;
const std::uintmax_t subseconds =
const std::uint64_t subseconds =
subsec_helper::get_subseconds(tmpval).count();
if (subseconds > 0) {
*out++ = '.';
const auto num_digits = detail::count_digits(subseconds);
const std::uint32_t num_digits =
detail::count_digits(static_cast<std::uint64_t>(subseconds));
if (subsec_helper::fractional_width > num_digits) {
out = std::fill_n(out, subsec_helper::fractional_width - num_digits,
'0');