Hello!
I've prepared a patch, which adds all members missing from
std::numeric_limits<> specializations for integer-class types.
Jonathan, please let me know whether you like these changes
and do not see any bugs or issues with them. From my side, I just want to
say that:
- Since all std::numeric_limits<> specializations for integral types,
defined in //libstdc++-v3/include/std/limits don't inherit from a base
class
providing common data members and member functions,
I also didn't introduce such a base class in
//libstdc++-v3/include/bits/max_size_type.h.
Such implementation has quite a bit of code duplication, but it's like
that on purpose, right?
- I didn't test the traps static data member, because I don't know how to
accurately predict when this compile-time constant should be true and
when it should be false.
Moreover, I saw that the unit-test verifying correctness of the traps
constant
from std::numeric_limits<> specializations for integral types
(//libstdc++-v3/testsuite/18_support/numeric_limits/traps.cc) also
doesn't verify its value.
- In the unit-tests for integer-class types I've defined variable
template
verify_numeric_limits_values_not_meaningful_for<> to avoid code
duplication
and have clear and readable code. I hope this is OK.
Thanks, Mateusz Zych
On Wed, Jul 2, 2025 at 7:30 PM Jonathan Wakely <jwak...@redhat.com> wrote:
> On Wed, 2 Jul 2025 at 17:15, Mateusz Zych wrote:
> >
> > OK, then I’ll prepare appropriate patch with tests and send it when I’m
> done implementing it.
>
> That would be great, thanks. I won't push the initial patch, we can
> wait for you to prepare the complete fix.
>
> Please note that for a more significant change, we have some legal
> prerequisites for contributions, as documented at:
> https://gcc.gnu.org/contribute.html#legal
>
> If you want to contribute under the DCO terms, please read
> https://gcc.gnu.org/dco.html so that you understand exactly what the
> Signed-off-by: trailer means.
>
> Thanks!
>
>
From d7d20d31e549e001f7644ee53899fa8494d0700f Mon Sep 17 00:00:00 2001
From: Mateusz Zych <mte.z...@gmail.com>
Date: Wed, 2 Jul 2025 01:51:40 +0300
Subject: [PATCH] libstdc++: Added missing members to numeric_limits
specializations for integer-class types.
25.3.4.4 Concept weakly_incrementable [iterator.concept.winc]
(5) For every integer-class type I,
let B(I) be a unique hypothetical extended integer type
of the same signedness with the same width as I.
[Note 2: The corresponding
hypothetical specialization numeric_limits<B(I)>
meets the requirements on
numeric_limits specializations for integral types.]
(11) For every (possibly cv-qualified) integer-class type I,
numeric_limits<I> is specialized such that:
- each static data member m
has the same value as numeric_limits<B(I)>::m, and
- each static member function f
returns I(numeric_limits<B(I)>::f()).
17.3.5.3 numeric_limits specializations [numeric.special]
(1) All members shall be provided for all specializations.
However, many values are only required to be meaningful
under certain conditions (for example,
epsilon() is only meaningful if is_integer is false).
Any value that is not meaningful shall be set to 0 or false.
libstdc++-v3/ChangeLog:
* include/bits/max_size_type.h
(numeric_limits<__max_size_type>): New static data members.
(numeric_limits<__max_diff_type>): Likewise.
Signed-off-by: Mateusz Zych <mte.z...@gmail.com>
---
libstdc++-v3/include/bits/max_size_type.h | 83 +++++++++++++++++++
.../std/ranges/iota/max_size_type.cc | 31 +++++++
2 files changed, 114 insertions(+)
diff --git a/libstdc++-v3/include/bits/max_size_type.h b/libstdc++-v3/include/bits/max_size_type.h
index 73a6d141d5b..30c5b124767 100644
--- a/libstdc++-v3/include/bits/max_size_type.h
+++ b/libstdc++-v3/include/bits/max_size_type.h
@@ -38,6 +38,7 @@
#include <ext/numeric_traits.h>
#include <bit> // __bit_width
#include <numbers>
+#include <limits> // __glibcxx_integral_traps
// This header implements unsigned and signed integer-class types (as per
// [iterator.concept.winc]) that are one bit wider than the widest supported
@@ -775,10 +776,27 @@ namespace ranges
static constexpr bool is_signed = false;
static constexpr bool is_integer = true;
static constexpr bool is_exact = true;
+ static constexpr bool is_bounded = true;
+ static constexpr bool is_modulo = true;
+ static constexpr bool traps = __glibcxx_integral_traps;
+ static constexpr int radix = 2;
static constexpr int digits
= __gnu_cxx::__int_traits<_Sp::__rep>::__digits + 1;
static constexpr int digits10
= static_cast<int>(digits * numbers::ln2 / numbers::ln10);
+ static constexpr int max_digits10 = 0;
+ static constexpr int min_exponent = 0;
+ static constexpr int min_exponent10 = 0;
+ static constexpr int max_exponent = 0;
+ static constexpr int max_exponent10 = 0;
+ static constexpr bool is_iec559 = false;
+ static constexpr bool has_infinity = false;
+ static constexpr bool has_quiet_NaN = false;
+ static constexpr bool has_signaling_NaN = false;
+ static constexpr bool has_denorm_loss = false;
+ static constexpr bool tinyness_before = false;
+ static constexpr float_denorm_style has_denorm = denorm_absent;
+ static constexpr float_round_style round_style = round_toward_zero;
static constexpr _Sp
min() noexcept
@@ -791,6 +809,30 @@ namespace ranges
static constexpr _Sp
lowest() noexcept
{ return min(); }
+
+ static constexpr _Sp
+ denorm_min() noexcept
+ { return 0; }
+
+ static constexpr _Sp
+ epsilon() noexcept
+ { return 0; }
+
+ static constexpr _Sp
+ round_error() noexcept
+ { return 0; }
+
+ static constexpr _Sp
+ infinity() noexcept
+ { return 0; }
+
+ static constexpr _Sp
+ quiet_NaN() noexcept
+ { return 0; }
+
+ static constexpr _Sp
+ signaling_NaN() noexcept
+ { return 0; }
};
template<>
@@ -802,9 +844,26 @@ namespace ranges
static constexpr bool is_signed = true;
static constexpr bool is_integer = true;
static constexpr bool is_exact = true;
+ static constexpr bool is_bounded = true;
+ static constexpr bool is_modulo = false;
+ static constexpr bool traps = __glibcxx_integral_traps;
+ static constexpr int radix = 2;
static constexpr int digits = numeric_limits<_Sp>::digits - 1;
static constexpr int digits10
= static_cast<int>(digits * numbers::ln2 / numbers::ln10);
+ static constexpr int max_digits10 = 0;
+ static constexpr int min_exponent = 0;
+ static constexpr int min_exponent10 = 0;
+ static constexpr int max_exponent = 0;
+ static constexpr int max_exponent10 = 0;
+ static constexpr bool is_iec559 = false;
+ static constexpr bool has_infinity = false;
+ static constexpr bool has_quiet_NaN = false;
+ static constexpr bool has_signaling_NaN = false;
+ static constexpr bool has_denorm_loss = false;
+ static constexpr bool tinyness_before = false;
+ static constexpr float_denorm_style has_denorm = denorm_absent;
+ static constexpr float_round_style round_style = round_toward_zero;
static constexpr _Dp
min() noexcept
@@ -817,6 +876,30 @@ namespace ranges
static constexpr _Dp
lowest() noexcept
{ return min(); }
+
+ static constexpr _Dp
+ denorm_min() noexcept
+ { return 0; }
+
+ static constexpr _Dp
+ epsilon() noexcept
+ { return 0; }
+
+ static constexpr _Dp
+ round_error() noexcept
+ { return 0; }
+
+ static constexpr _Dp
+ infinity() noexcept
+ { return 0; }
+
+ static constexpr _Dp
+ quiet_NaN() noexcept
+ { return 0; }
+
+ static constexpr _Dp
+ signaling_NaN() noexcept
+ { return 0; }
};
template<>
diff --git a/libstdc++-v3/testsuite/std/ranges/iota/max_size_type.cc b/libstdc++-v3/testsuite/std/ranges/iota/max_size_type.cc
index 4739d9e2f79..fbd783bdf1a 100644
--- a/libstdc++-v3/testsuite/std/ranges/iota/max_size_type.cc
+++ b/libstdc++-v3/testsuite/std/ranges/iota/max_size_type.cc
@@ -352,6 +352,9 @@ static_assert(numeric_limits<max_size_t>::is_specialized);
static_assert(!numeric_limits<max_size_t>::is_signed);
static_assert(numeric_limits<max_size_t>::is_integer);
static_assert(numeric_limits<max_size_t>::is_exact);
+static_assert(numeric_limits<max_size_t>::is_bounded);
+static_assert(numeric_limits<max_size_t>::is_modulo);
+static_assert(numeric_limits<max_size_t>::radix == 2);
// We can't unconditionally use numeric_limits here because __int128 is an
// integral type only in GNU mode.
#if __SIZEOF_INT128__
@@ -379,6 +382,9 @@ static_assert(numeric_limits<max_diff_t>::is_specialized);
static_assert(numeric_limits<max_diff_t>::is_signed);
static_assert(numeric_limits<max_diff_t>::is_integer);
static_assert(numeric_limits<max_diff_t>::is_exact);
+static_assert(numeric_limits<max_diff_t>::is_bounded);
+static_assert(!numeric_limits<max_diff_t>::is_modulo);
+static_assert(numeric_limits<max_diff_t>::radix == 2);
static_assert(numeric_limits<max_diff_t>::digits
== numeric_limits<max_size_t>::digits - 1);
static_assert(numeric_limits<max_diff_t>::digits10
@@ -400,6 +406,31 @@ static_assert(max_diff_t(max_size_t(1)
<< (numeric_limits<max_size_t>::digits-1))
== numeric_limits<max_diff_t>::min());
+template <typename integer_class>
+constexpr bool verify_numeric_limits_values_not_meaningful_for = true
+ && (numeric_limits<integer_class>::max_digits10 == 0)
+ && (numeric_limits<integer_class>::min_exponent == 0)
+ && (numeric_limits<integer_class>::min_exponent10 == 0)
+ && (numeric_limits<integer_class>::max_exponent == 0)
+ && (numeric_limits<integer_class>::max_exponent10 == 0)
+ && !numeric_limits<integer_class>::is_iec559
+ && !numeric_limits<integer_class>::has_infinity
+ && !numeric_limits<integer_class>::has_quiet_NaN
+ && !numeric_limits<integer_class>::has_signaling_NaN
+ && !numeric_limits<integer_class>::has_denorm_loss
+ && !numeric_limits<integer_class>::tinyness_before
+ && (numeric_limits<integer_class>::has_denorm == std::denorm_absent)
+ && (numeric_limits<integer_class>::round_style == std::round_toward_zero)
+ && (numeric_limits<integer_class>::denorm_min() == 0)
+ && (numeric_limits<integer_class>::epsilon() == 0)
+ && (numeric_limits<integer_class>::round_error() == 0)
+ && (numeric_limits<integer_class>::infinity() == 0)
+ && (numeric_limits<integer_class>::quiet_NaN() == 0)
+ && (numeric_limits<integer_class>::signaling_NaN() == 0);
+
+static_assert(verify_numeric_limits_values_not_meaningful_for<max_size_t>);
+static_assert(verify_numeric_limits_values_not_meaningful_for<max_diff_t>);
+
// Verify that the types are structural types and can therefore be used
// as NTTP types.
template<max_size_t V> struct Su { static_assert(V*V == V+132); };
--
2.48.1
