Standalone Fixes #403

mborland · 2021-12-11T14:12:02Z

Moves the discussion from the now merged #386

[ci skip]

ckormanyos · 2021-12-11T16:59:01Z

Hi @mborland thanks for creating this PR for further progress.

In our previous thread, you made a comment regarding using a multiprecision type for __int128. I think this could be potentially problematic (although I'm not sure). Problems may arise because the particular type unsigned __int128 is being actually needed as cpp_int's class-internal double-limb-type.

So you might end up using a class that has not even yet been synthesized in order to build up class-internals?

I could be wrong on this one though since i'm not exactly sure how the intent is that you had mentioned.

ckormanyos · 2021-12-11T17:20:45Z

Hi @mborland and @jzmaddock

The noexcept culprit seems to have been found. Thanks John!

From our previous chat, @jzmaddock mentions...

perhaps you could test making that method noexcept only when BOOST_NO_CXX17_IF_CONSTEXPR is not defined?

The modification around line 1272 in cpp_int.hpp fixes the noexcept warnings here.

There are a handful of other little changes in my branch for warning fixes on conversion and sign conversion (these are optional). My CI in the unrelated project ran green.

[ci skip]

mborland · 2021-12-11T17:28:12Z

@ckormanyos if you look at the latest commit boost::multiprecision::int128_type is just an alias to __int128. Let me know if it suppresses the warnings for you like it should.

…one_fixes [ci skip]

ckormanyos · 2021-12-12T08:08:23Z

look at the latest commit boost::multiprecision::int128_type is just an alias to __int128. Let me know if it suppresses the warnings for you like it should

Hi @mborland. Thank you.

Your recent commits eliminate almost all the warnings. For lack of a better communication method, I simply included full warnings under subjection to -Wall -Wextra -pedantic below.

The two cases are:

develop branch
your standalone branch.

Almost all warnings are gone from your branch with a couple remaining, that I did not, yet at the moment, analyze.

DEVELOP BRANCH

chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$ g++ -Wall -Wextra -pedantic -Os -march=native -std=c++11 -I/mnt/c/boost/modular_boost/boost/libs/multiprecision/include -I/mnt/c/boost/modular_boost/boost/libs/math/include -I/mnt/c/boost/modular_boost/boost/libs/config/include ./test.cpp -o test.exe
In file included from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/number_base.hpp:18,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/is_variable_precision.hpp:9,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/precision.hpp:9,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/number.hpp:11,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/cpp_dec_float.hpp:29,
                 from ./test.cpp:61:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:31:18: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   31 | struct is_signed<__int128> : public std::integral_constant<bool, true> {};
      |                  ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:33:27: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   33 | struct is_signed<unsigned __int128> : public std::integral_constant<bool, false> {};
      |                           ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:35:20: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   35 | struct is_unsigned<__int128> : public std::integral_constant<bool, false> {};
      |                    ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:37:29: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   37 | struct is_unsigned<unsigned __int128> : public std::integral_constant<bool, true> {};
      |                             ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:39:20: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   39 | struct is_integral<__int128> : public std::integral_constant<bool, true> {};
      |                    ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:41:29: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   41 | struct is_integral<unsigned __int128> : public std::integral_constant<bool, true> {};
      |                             ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:43:22: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   43 | struct is_arithmetic<__int128> : public std::integral_constant<bool, true> {};
      |                      ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:45:31: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   45 | struct is_arithmetic<unsigned __int128> : public std::integral_constant<bool, true> {};
      |                               ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:47:22: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   47 | struct make_unsigned<__int128>
      |                      ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:49:26: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   49 |    using type = unsigned __int128;
      |                          ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:52:31: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   52 | struct make_unsigned<unsigned __int128>
      |                               ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:54:26: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   54 |    using type = unsigned __int128;
      |                          ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:57:20: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   57 | struct make_signed<__int128>
      |                    ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:59:17: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   59 |    using type = __int128;
      |                 ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:62:29: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   62 | struct make_signed<unsigned __int128>
      |                             ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/traits/std_integer_traits.hpp:64:17: warning:
ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   64 |    using type = __int128;
      |                 ^~~~~~~~
In file included from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/generic_interconvert.hpp:9,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/number.hpp:12,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/cpp_dec_float.hpp:29,
                 from ./test.cpp:61:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/default_ops.hpp: In function ‘void boost::multiprecision::default_ops::eval_karatsuba_sqrt(Backend&, const Backend&, Backend&, Backend&, size_t)’:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/default_ops.hpp:1691:25: warning: ISO C++ does not support ‘__int128’ for ‘a’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                         ^
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/default_ops.hpp:1691:30: warning: ISO C++ does not support ‘__int128’ for ‘b’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                              ^
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/default_ops.hpp:1691:35: warning: ISO C++ does not support ‘__int128’ for ‘c’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                                   ^
In file included from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/number.hpp:16,
                 from /mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/cpp_dec_float.hpp:29,
                 from ./test.cpp:61:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp: At global scope:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp:23:49: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   23 | std::size_t hash_value(const unsigned __int128& val);
      |                                                 ^~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp:25:47: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   25 | inline std::size_t hash_value(const __int128& val)
      |                                               ^~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp: In function ‘std::size_t boost::multiprecision::detail::hash_value(const __int128&)’:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp:27:43: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   27 |    return hash_value(static_cast<unsigned __int128>(val));
      |                                           ^~~~~~~~
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp: At global scope:
/mnt/c/boost/modular_boost/boost/libs/multiprecision/include/boost/multiprecision/detail/hash.hpp:44:56: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   44 | inline std::size_t hash_value(const unsigned __int128& val)
      |                                                        ^~~
chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$

STANDALONE BRANCH

chris@DESKTOP-Q6SS99B:~$ cd /mnt/c/Users/User/Documents/Ks/PC_Software/Test
chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$ g++ -Wall -Wextra -pedantic -Os -march=native -std=c++11 -I/mnt/c/boost/boost_multiprecision_standalone/include -I/mnt/c/boost/modular_boost/boost/libs/math/include -I/mnt/c/boost/modular_boost/boost/libs/config/include ./test.cpp -o test.exe
In file included from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/generic_interconvert.hpp:9,
                 from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/number.hpp:12,
                 from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_dec_float.hpp:29,
                 from ./test.cpp:61:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/default_ops.hpp: In function ‘void boost::multiprecision::default_ops::eval_karatsuba_sqrt(Backend&, const Backend&, Backend&, Backend&, size_t)’:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/default_ops.hpp:1691:25: warning: ISO C++ does not support ‘__int128’ for ‘a’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                         ^
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/default_ops.hpp:1691:30: warning: ISO C++ does not support ‘__int128’ for ‘b’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                              ^
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/default_ops.hpp:1691:35: warning: ISO C++ does not support ‘__int128’ for ‘c’ [-Wpedantic]
 1691 |       unsigned __int128 a{}, b{}, c{};
      |                                   ^
In file included from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/number.hpp:16,
                 from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_dec_float.hpp:29,
                 from ./test.cpp:61:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp: At global scope:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp:23:49: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   23 | std::size_t hash_value(const unsigned __int128& val);
      |                                                 ^~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp:25:47: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   25 | inline std::size_t hash_value(const __int128& val)
      |                                               ^~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp: In function ‘std::size_t boost::multiprecision::detail::hash_value(const __int128&)’:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp:27:43: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   27 |    return hash_value(static_cast<unsigned __int128>(val));
      |                                           ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp: At global scope:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/hash.hpp:44:56: warning: ISO C++ does not support ‘__int128’ for ‘val’ [-Wpedantic]
   44 | inline std::size_t hash_value(const unsigned __int128& val)
      |                                                        ^~~
chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$

[ci skip]

ckormanyos · 2021-12-12T11:28:27Z

Thnks @mborland.

Sorry Matt, I missed a few __int128 warnings on -pedantic. My last report was for cpp_dec_float.

I have found the following shown below for cpp_bin_float. Note that this should cover now dec-float, bin-float and cpp_int as well. (I'll run gmp_float and mpfr_floatthrough ASAP.)

STANDALONE BRANCH with cpp_bin_float

chris@DESKTOP-Q6SS99B:~$ cd /mnt/c/Users/User/Documents/Ks/PC_Software/Test
chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$ g++ -Wall -Wextra -pedantic -Os -march=native -std=c++11 -I/mnt/c/boost/boost_multiprecision_standalone/include -I/mnt/c/boost/modular_boost/boost/libs/math/include -I/mnt/c/boost/modular_boost/boost/libs/config/include ./test.cpp -o test.exe
chris@DESKTOP-Q6SS99B:/mnt/c/Users/User/Documents/Ks/PC_Software/Test$ g++ -Wall -Wextra -pedantic -Os -march=native -std=c++11 -I/mnt/c/boost/boost_multiprecision_standalone/include -I/mnt/c/boost/modular_boost/boost/libs/math/include -I/mnt/c/boost/modular_boost/boost/libs/config/include ./test.cpp -o test.exe
./test.cpp:10:1: warning: multi-line comment [-Wcomment]
   10 | //TGT_INCLUDES  = -IC:/boost/modular_boost/boost/libs/multiprecision/include        \
      | ^
In file included from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/endian.hpp:9,
                 from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int.hpp:14,
                 from /mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_bin_float.hpp:11,
                 from ./test.cpp:60:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp: In function ‘constexpr T boost::multiprecision::backends::detail::type_max()’:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:25: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                         ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:29:65: note: in expansion of macro ‘INT128_MAX’
   29 |    std::is_same<T, boost::multiprecision::int128_type>::value ? INT128_MAX :
      |                                                                 ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:47: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                                               ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:29:65: note: in expansion of macro ‘INT128_MAX’
   29 |    std::is_same<T, boost::multiprecision::int128_type>::value ? INT128_MAX :
      |                                                                 ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:88:41: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   88 | #    define UINT128_MAX ((2 * (unsigned __int128) INT128_MAX) + 1)
      |                                         ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:30:66: note: in expansion of macro ‘UINT128_MAX’
   30 |    std::is_same<T, boost::multiprecision::uint128_type>::value ? UINT128_MAX :
      |                                                                  ^~~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:25: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                         ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:88:51: note: in expansion of macro ‘INT128_MAX’
   88 | #    define UINT128_MAX ((2 * (unsigned __int128) INT128_MAX) + 1)
      |                                                   ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:30:66: note: in expansion of macro ‘UINT128_MAX’
   30 |    std::is_same<T, boost::multiprecision::uint128_type>::value ? UINT128_MAX :
      |                                                                  ^~~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:47: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                                               ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:88:51: note: in expansion of macro ‘INT128_MAX’
   88 | #    define UINT128_MAX ((2 * (unsigned __int128) INT128_MAX) + 1)
      |                                                   ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:30:66: note: in expansion of macro ‘UINT128_MAX’
   30 |    std::is_same<T, boost::multiprecision::uint128_type>::value ? UINT128_MAX :
      |                                                                  ^~~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp: In function ‘constexpr T boost::multiprecision::backends::detail::type_min()’:
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:25: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                         ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:85:26: note: in expansion of macro ‘INT128_MAX’
   85 | #    define INT128_MIN (-INT128_MAX - 1)
      |                          ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:40:65: note: in expansion of macro ‘INT128_MIN’
   40 |    std::is_same<T, boost::multiprecision::int128_type>::value ? INT128_MIN :
      |                                                                 ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:82:47: warning: ISO C++ does not support ‘__int128’ for ‘type name’ [-Wpedantic]
   82 | #    define INT128_MAX (__int128) (((unsigned __int128) 1 << ((__SIZEOF_INT128__ * __CHAR_BIT__) - 1)) - 1)
      |                                               ^~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/detail/standalone_config.hpp:85:26: note: in expansion of macro ‘INT128_MAX’
   85 | #    define INT128_MIN (-INT128_MAX - 1)
      |                          ^~~~~~~~~~
/mnt/c/boost/boost_multiprecision_standalone/include/boost/multiprecision/cpp_int/checked.hpp:40:65: note: in expansion of macro ‘INT128_MIN’
   40 |    std::is_same<T, boost::multiprecision::int128_type>::value ? INT128_MIN :
      |                                                                 ^~~~~~~~~~

ckormanyos · 2021-12-13T08:05:27Z

Hi folks, this is moving along well.

There are several issues I'd like to highlight and get feedback on.

Are we testing standalone? I remember some comment(s) from @jzmaddock but did not understand the exact depth or meaning thereof. So @mborland do you think we have explicit standalone tests in the branch and/or on develop?
When embedding cpp_dec_float versus cpp_bin_float, I find that bin-float adds about 200 additional kilobytes of program code almost entirely needed for string-streaming (via std::stringstream). Since my to total benchmark size (101 decimal digit cube root) is less than 100 kilobyte program code, adding 200 kilobyte for string-streaming is exorbitant. The heavy weight comes from reading string limbs and converting them to cpp_int. I'd suggest we try to replace std::stringstream-ing with lightweight std::stoul() or similar as we did in cpp_dec_float.
Now that standalone can basically, ... stand alone, it would be really helpful to ramp up the warning level. i have good experiences with adding GCC's -Wconversion and -Wsign-conversion. This results in about 25 warnings in dec-float and fewer than 50 warnings in bin-float. I would like to pursue resolution of these warnings since both i as well as a couple folks I know use such warnings and perpetually need to disable them for Boost. Are we agreed should I repair them? Or should I let them go?
I will finish searching for -pedantic ISO-C++ non-conformace resulting from __int128. I will report thest presently.

Thoughts... Suggestions... ?

Cc: @mborland and @jzmaddock and @NAThompson

mborland · 2021-12-13T19:20:11Z

@ckormanyos I am working on re-writing test_arithmetic to contextually disable any boost usage. That will resolve the standalone testing problem I think.

ckormanyos · 2021-12-14T11:06:28Z

working on re-writing test_arithmetic to contextually disable any boost usage. That will resolve the standalone testing problem I think.

Great! Thanks @mborland. As soon as this is/will be running, I think we have a really good standpoint on standalone.

Optionally... I will take a look at remaining __int128 warnings, and check out any other elevated conversion warnings. i will also investigate the situation with string-streaming.

Again, this stuff is mostly optional and not needed for standalone (with the exception of the __int128 warnings on -pedantic that we did identify as a goal). So I will get back to you if any changes are required on that detail.

The higher conversin warnings and string-streaming investigations are not required but I will try to find a chance to look into them anyway.

ckormanyos · 2021-12-15T18:44:34Z

Hi @mborland I did a quick check of bin-float, dec-float, gmp-float and mpfr-float.

Many thanks. I find in this quick check that the all formerly remaining __int128 warnings on -pedantic are now gone.

[ci skip]

ckormanyos · 2021-12-18T12:17:48Z

Hi @mborland we might want to merge from develop again. We observe that @jzmaddock is excellently handling issue-after-issue.

When I merged develop over to my standalone_chris branch, I found a merge conflict or two in misc.hpp. In addition to all the changes from mborland:standalone, I did, however, also have a small number of changes handling conversion warnings in cpp_dec_float. So that might have complicated my particular merge.

It might be easier to merge to mborland:standalone_fixes directly from develop first --- or optionally from my branch remotely. Either way, the few changes I had/have for -Wconversion and -Wsign-conversion I would also be happy to reproduce any time if needed.

[ci skip]

…one_fixes [ci skip]

mborland · 2021-12-19T17:13:09Z

@jzmaddock Good call. Ran the new gmp_float to long double conversion through valgrind to validate the change.

==101192== Memcheck, a memory error detector
==101192== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==101192== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
==101192== Command: ./log_adptr --leak-check=full
==101192== 
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and h
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and a
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and a
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and c
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and c
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and s
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and s
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and t
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and i
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and i
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and j
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and l
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and l
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and m
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and x
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and x
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and y
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and f
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and f
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and d
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and d
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and e
Testing mixed arithmetic with type: N5boost14multiprecision6numberINS0_8backends14logged_adaptorINS2_9gmp_floatILj0EEEEELNS0_26expression_template_optionE1EEE and e
No errors detected.
==101192== 
==101192== HEAP SUMMARY:
==101192==     in use at exit: 0 bytes in 0 blocks
==101192==   total heap usage: 13,335 allocs, 13,335 frees, 607,350 bytes allocated
==101192== 
==101192== All heap blocks were freed -- no leaks are possible
==101192== 
==101192== For lists of detected and suppressed errors, rerun with: -s
==101192== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

Will fix the gmp_int conversions.

[ci skip]

mborland · 2021-12-21T09:44:33Z

@ckormanyos Do you see any large or outstanding issues that we should address here? CI is back to green with all these changes.

ckormanyos · 2021-12-21T10:43:33Z

This is great @mborland !

Do you see any large or outstanding issues that we should address here? CI is back to green with all these changes.

Functionally, all is OK.

One thing you might want, ... or freely choose to disregard...

I just did commit (via ci skip in my standalone_chris branch) a bunch of changes to handle conversion warnings in cpp_bin_float. Some of the casts I selected were quite difficult to choose due to all the massively templated code. I also added missing default cases to switches. If you would like to take a look at the casts and added default cases missing from switches, we could consider integrating my changes.

It's your choice if you'd like to look at and potentially accept these corrections to conversoin warnings or not. If not, I am good to go at this point with the excellent state we have reached.

If we decide to accept my corrections to conversoin warnings in Multiprecision, there is an associated little list of changes needed in Math that I can also present.

Cc: @jzmaddock

ckormanyos · 2021-12-21T14:58:26Z

CI is back to green with all these changes.

my corrections to conversoin warnings

Ummm... Alternatively, we could merge to develop immediately. If I would want to to work on elevated warning levels on certain compilers, I should be happy to do that in future branches.

This would be more consistent with the topic of this PR which is going standalone.

@mborland and @jzmaddock thoughts...?

Cc: @NAThompson

jzmaddock · 2021-12-21T15:58:15Z

include/boost/multiprecision/cpp_dec_float.hpp

-   ss >> d;
-
-   return d;
+   return std::strtod(str(std::numeric_limits<double>::digits10 + (2 + 1), std::ios_base::scientific).c_str(), nullptr);


This is a pain in the butt, but can you check that this doesn't re-awaken this issue: #167. The problem is that the global locale can sometimes interpret strings in mighty weird ways :(

include/boost/multiprecision/gmp.hpp

include/boost/multiprecision/tommath.hpp

jzmaddock · 2021-12-21T17:10:49Z

Ummm... Alternatively, we could merge to develop immediately. If I would want to to work on elevated warning levels on certain compilers, I should be happy to do that in future branches.

+1, otherwise the diffs get too big!

ckormanyos · 2021-12-21T18:29:15Z

merge to develop immediately

+1, otherwise the diffs get too big!

Done. Merged to develop. Review comments to be handled in a future PR.

jzmaddock · 2021-12-21T19:30:53Z

Review comments to be handled in a future PR.

OK, but some of those look like proper breakages to me.

ckormanyos · 2021-12-22T08:05:03Z

Review comments to be handled in a future PR.

OK, but some of those look like proper breakages to me.

Uggghhh. Sorry that was not the intent and I had not realized that. I guess my intensity in the drive toward standalone got the better of me. I'd need help on some of those details.

@mborland and @jzmaddock should we move forward with next PR or should I revert the changes?

Sorry for any disruption I caused.

jzmaddock · 2021-12-22T11:52:44Z

@mborland and @jzmaddock should we move forward with next PR or should I revert the changes?

I think I would vote for reverting.

Sorry for any disruption I caused.

No worries!

mborland · 2021-12-22T13:41:51Z

@ckormanyos I go on holiday starting tonight so I concur with reverting the merge.

This reverts commit b034196, reversing changes made to 75b17b9.

ckormanyos · 2021-12-22T19:12:16Z

vote for reverting

concur with reverting the merge

OK. Agreed. With one episode of back and forth, I have reverted to 75b17b9

This includes John's fix of #407 but reverts my recent commit/push from Matt's standalone_fixes branch. Due to the one episode of back-and-forth, I inadvertently fired off three CI runs.

I was a bit shaky on the revert syntax, but i'm pretty sure I hit the right commit now. Could you please check @mborland and @jzmaddock ?

Thank you.

jzmaddock · 2021-12-22T19:31:38Z

Thanks Chris, I think you got it: https://github.com/boostorg/multiprecision/compare/75b17b943a6b9d02d5a2be8c439b3a3267dcba79..7a563f36f4868f56bcb7adf5ee4b4a972605870b

ckormanyos · 2021-12-23T07:13:46Z

I think you got it

OK, Cool --- back on track.

Thanks John. Thanks Matt. Let's pick this up maybe after the holidays where we left off...

…one_fixes"" This reverts commit 04b327b.

ckormanyos and others added 4 commits December 10, 2021 08:19

[ci skip] Merge branch 'develop' into standalone_chris

dd3d493

Avoid passing argument to type_min/max

8dfda64

cpp_dec_float review fixes

e7c473c

[ci skip]

[ci skip] Try find noexcept warning(s)

501406a

Fix iso C++ warnings for using __int128

7389f23

[ci skip]

Merge remote-tracking branch 'boostorg/standalone_chris' into standal…

6ee1691

…one_fixes [ci skip]

More __int128 warning cleanup

b3a88e3

[ci skip]

[ci skip] Sync with mborland:standalone_fixes

20151ac

Re-write (U)INT128_MAX Macros

2e912e6

mborland and others added 7 commits December 16, 2021 17:45

Revamp test_arithmetic to support standalone mode

8f7089b

[ci skip]

Fix standalone conversions for tommath int

40740b7

Replace some instances of std::stringstream in cpp_dec_float

0458d13

[ci skip]

[ci skip] Merge develop and mborland:standalone

44aef25

Fix order of defines in standalone tests

dabe54a

[ci skip] Remove obvious conversion warn dec-float

7e240df

[ci skip] Merge 'develop' to standalone_chris

5e23cad

mborland added 4 commits December 18, 2021 14:26

Implement gmp_int to long double conversion

a35cc58

Add gmp_float to long double conversion

69467eb

[ci skip]

Merge remote-tracking branch 'boostorg/develop' into standalone_fixes

8652473

[ci skip]

Merge remote-tracking branch 'boostorg/standalone_chris' into standal…

d11a32b

…one_fixes [ci skip]

mborland added 5 commits December 19, 2021 19:33

Fix memory leak in gmp_int to long double conversion

6f554ff

[ci skip]

Conditionally remove boost::math usage from gmp

806d43a

Conditionally remove boost::math usage tommath

16da2ca

[ci skip]

unique_ptr leaks memory. Replace with hand-rolled version..

bf0f684

Remove unneeded asserts that trip up some compiler versions

5d4be65

mborland marked this pull request as ready for review December 21, 2021 09:36

jzmaddock reviewed Dec 21, 2021

View reviewed changes

include/boost/multiprecision/gmp.hpp Show resolved Hide resolved

jzmaddock reviewed Dec 21, 2021

View reviewed changes

include/boost/multiprecision/gmp.hpp Show resolved Hide resolved

jzmaddock reviewed Dec 21, 2021

View reviewed changes

include/boost/multiprecision/tommath.hpp Show resolved Hide resolved

ckormanyos merged commit b034196 into boostorg:develop Dec 21, 2021

ckormanyos added a commit that referenced this pull request Dec 22, 2021

Revert "Merge pull request #403 from mborland/standalone_fixes"

04b327b

This reverts commit b034196, reversing changes made to 75b17b9.

mborland deleted the standalone_fixes branch January 2, 2022 12:24

mborland added a commit to mborland/multiprecision that referenced this pull request Jan 2, 2022

Revert "Revert "Merge pull request boostorg#403 from mborland/standal…

7f0723e

…one_fixes"" This reverts commit 04b327b.

mborland mentioned this pull request Jan 2, 2022

Standalone fixes #409

Merged

mborland added a commit to mborland/multiprecision that referenced this pull request Jan 6, 2022

Revert "Revert "Merge pull request boostorg#403 from mborland/standal…

5095770

…one_fixes"" This reverts commit 04b327b.

ckormanyos mentioned this pull request Mar 24, 2022

Standalone? #364

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Standalone Fixes #403

Standalone Fixes #403

mborland commented Dec 11, 2021

ckormanyos commented Dec 11, 2021

ckormanyos commented Dec 11, 2021 •

edited

Loading

mborland commented Dec 11, 2021

ckormanyos commented Dec 12, 2021

ckormanyos commented Dec 12, 2021 •

edited

Loading

ckormanyos commented Dec 13, 2021 •

edited

Loading

mborland commented Dec 13, 2021

ckormanyos commented Dec 14, 2021

ckormanyos commented Dec 15, 2021

ckormanyos commented Dec 18, 2021 •

edited

Loading

mborland commented Dec 19, 2021

mborland commented Dec 21, 2021

ckormanyos commented Dec 21, 2021 •

edited

Loading

ckormanyos commented Dec 21, 2021 •

edited

Loading

jzmaddock Dec 21, 2021

jzmaddock commented Dec 21, 2021

ckormanyos commented Dec 21, 2021

jzmaddock commented Dec 21, 2021

ckormanyos commented Dec 22, 2021

jzmaddock commented Dec 22, 2021

mborland commented Dec 22, 2021

ckormanyos commented Dec 22, 2021

jzmaddock commented Dec 22, 2021

ckormanyos commented Dec 23, 2021

Standalone Fixes #403

Standalone Fixes #403

Conversation

mborland commented Dec 11, 2021

ckormanyos commented Dec 11, 2021

ckormanyos commented Dec 11, 2021 • edited Loading

mborland commented Dec 11, 2021

ckormanyos commented Dec 12, 2021

ckormanyos commented Dec 12, 2021 • edited Loading

ckormanyos commented Dec 13, 2021 • edited Loading

mborland commented Dec 13, 2021

ckormanyos commented Dec 14, 2021

ckormanyos commented Dec 15, 2021

ckormanyos commented Dec 18, 2021 • edited Loading

mborland commented Dec 19, 2021

mborland commented Dec 21, 2021

ckormanyos commented Dec 21, 2021 • edited Loading

ckormanyos commented Dec 21, 2021 • edited Loading

jzmaddock Dec 21, 2021

Choose a reason for hiding this comment

jzmaddock commented Dec 21, 2021

ckormanyos commented Dec 21, 2021

jzmaddock commented Dec 21, 2021

ckormanyos commented Dec 22, 2021

jzmaddock commented Dec 22, 2021

mborland commented Dec 22, 2021

ckormanyos commented Dec 22, 2021

jzmaddock commented Dec 22, 2021

ckormanyos commented Dec 23, 2021

ckormanyos commented Dec 11, 2021 •

edited

Loading

ckormanyos commented Dec 12, 2021 •

edited

Loading

ckormanyos commented Dec 13, 2021 •

edited

Loading

ckormanyos commented Dec 18, 2021 •

edited

Loading

ckormanyos commented Dec 21, 2021 •

edited

Loading

ckormanyos commented Dec 21, 2021 •

edited

Loading