Add support for _BitInt on clang

[Arghnews]

Issue #4007

Only enabled for clang >= 14 as that's all that supports _BitInt at the moment

We need a forwarding reference for map to be a "more specialized" match; const T&, T etc. are ambiguous.

I'm not too familiar, would there be a better way to do this by adding an entry into enum class type for _BitInts? Or this is fine?

---

Fyi this worked in version 9 because of this catch-all overload:

fmt/include/fmt/core.h

Lines 1506 to 1516 in a337011

	template <typename T, typename U = remove_cvref_t<T>,
	FMT_ENABLE_IF(!is_string<U>::value && !is_char<U>::value &&
	!std::is_array<U>::value &&
	!std::is_pointer<U>::value &&
	!has_format_as<U>::value &&
	(has_formatter<U, Context>::value ||
	has_fallback_formatter<U, char_type>::value))>
	FMT_CONSTEXPR FMT_INLINE auto map(T&& val)
	-> decltype(this->do_map(std::forward<T>(val))) {
	return do_map(std::forward<T>(val));
	}

Which was then subsequently narrowed down to extended FP types (which "broke" the accidental support for this):

fmt/include/fmt/base.h

Lines 1527 to 1537 in 6a192f8

	// is_fundamental is used to allow formatters for extended FP types.
	template <typename T, typename U = remove_const_t<T>,
	FMT_ENABLE_IF(
	(std::is_class<U>::value || std::is_enum<U>::value ||
	std::is_union<U>::value || std::is_fundamental<U>::value) &&
	!has_to_string_view<U>::value && !is_char<U>::value &&
	!is_named_arg<U>::value && !std::is_integral<U>::value &&
	!std::is_arithmetic<format_as_t<U>>::value)>
	FMT_MAP_API auto map(T& val) -> decltype(FMT_DECLTYPE_THIS do_map(val)) {
	return do_map(val);
	}

[vitaut]

Thanks for the PR!

[vitaut]

This can be simplified to

#  if FMT_CLANG_VERSION >= 1400

[vitaut]

Let's move this out of the mapper, next to the _BitInt detection above.

[vitaut]

This should be defined outside of conditional block so that the check could be used on all platforms.

[vitaut]

Defining formatters for _BitInt should be disallowed.

[vitaut]

_BitInt should be converted to appropriate integer type.

[Arghnews]

Are we assuming the maximum width for a _BitInt is 128 bits here, so we can widen to any of the native integer types (including __int128)?

I think officially this is what clang supports according to BITINT_MAXWIDTH for c++ (although from the link in that 5th bullet point, it seems this is more due to a limitation than a design choice. Doesn't seem to be defined for 18 wherever I try it either)

Although for C(23), it supports up to 8388608
https://godbolt.org/z/jzfGYoY6a

@vitaut

[vitaut]

I think we can assume that for now and in any case for smaller sizes we should convert to integers for efficiency instead of going through the formatter extension.

[vitaut]

@Arghnews, do you plan to update this PR or shall we close it for now (can be reopened later)?

[Arghnews]

@Arghnews, do you plan to update this PR or shall we close it for now (can be reopened later)?

@vitaut Hi, sorry been busy and now I'm on holiday for 2 weeks, but will update once back and implement your feedback. I'd say leave it open and I'll update in 2-3 weeks, but it's up to you (can reopen of course). Cheers!

[vitaut]

No hurry. Let's keep it open and enjoy your holiday!