complex_adaptor compile error with boost-1.77.0 and later

[vmauery]

I recently updated my OS, which included an update to boost (from 1.74 to 1.81). I suddenly found that my little personal calculator program would fail to build. I bisected between boost-1.74 and boost-1.81 and found that boost-1.76 was the last working release.

Setup:

$ g++ -dumpversion
13
$ clang++ -dumpversion
17.0.6

boost (current): 9f83dd55ea67f1bab62c20d29bc5527345aafc55 multiprecision (current): de3aded8632e0ef0f17dcaf274f5699a25139738

Test case:

/*
Originally from
https://www.boost.org/doc/libs/1_84_0/libs/multiprecision/doc/html/boost_multiprecision/tut/complex/cpp_complex.html
but simplified to do fewer operations and modified to use a custom type.

Compile-OK: g++ -DUSE_EXAMPLE_TYPES cpp_complex.cpp -o cpp_complex.cpp
Compile-Fail: g++ cpp_complex.cpp -o cpp_complex.cpp -lgmp -lmpfr

*/

#include <iostream>

#ifdef USE_EXAMPLE_TYPES
#include <boost/multiprecision/cpp_complex.hpp>
#include <complex>
#else // USE_CUSTOM_TYPES
#include <boost/multiprecision/complex_adaptor.hpp>
#include <boost/multiprecision/mpfr.hpp>
#include <boost/multiprecision/number.hpp>
#endif

template <class Complex>
void complex_number_examples()
{
    Complex z1{0, 1};
    std::cout << std::setprecision(
        std::numeric_limits<typename Complex::value_type>::digits10);
    std::cout << std::scientific << std::fixed;
    std::cout << "Print a complex number: " << z1 << std::endl;
    std::cout << "Square it             : " << z1 * z1 << std::endl;
    std::cout << "Real part             : " << z1.real() << " = " << real(z1)
              << std::endl;
    std::cout << "Imaginary part        : " << z1.imag() << " = " << imag(z1)
              << std::endl;
}

int main()
{
#ifdef USE_EXAMPLE_TYPES

    std::cout
        << "First, some operations we usually perform with std::complex:\n";
    complex_number_examples<std::complex<double>>();
    std::cout << "\nNow the same operations performed using quad precision "
                 "complex numbers:\n";
    complex_number_examples<boost::multiprecision::cpp_complex_quad>();

#else // USE_CUSTOM_TYPES

    using float_backend = boost::multiprecision::mpfr_float_backend<0>;
    using complex_backend =
        boost::multiprecision::complex_adaptor<float_backend>;
    using mpc = boost::multiprecision::number<complex_backend,
                                              boost::multiprecision::et_off>;
    std::cout << "\nNow the same operations performed using custom complex "
                 "numbers:\n";
    complex_number_examples<mpc>();

#endif
    return 0;
}

Compile matrix:

----------------------------------------------------------------------------------
| command line               |  Boost version  |  Outcome
----------------------------------------------------------------------------------
| g++ -DUSE_EXAMPLE_TYPES    |  boost-1.76     |  OK
----------------------------------------------------------------------------------
| g++ -DUSE_EXAMPLE_TYPES    |  boost-latest   |  OK
----------------------------------------------------------------------------------
| g++ -lgmp -lmpfr           |  boost-1.76     |  OK
----------------------------------------------------------------------------------
| g++ -lgmp -lmpfr           |  boost-latest   |  Fail
----------------------------------------------------------------------------------

I see the error with both g++ and clang++. The clang++ error is easier to read:

In file included from cpp_complex.cpp:7:
In file included from /usr/local/include/boost/multiprecision/complex_adaptor.hpp:9:
/usr/local/include/boost/multiprecision/number.hpp:1096:23: error: no member named 'thread_default_precision' in 'boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>'
 1096 |       return Backend::thread_default_precision();
      |              ~~~~~~~~~^
/usr/local/include/boost/multiprecision/detail/precision.hpp:198:23: note: in instantiation of member function 'boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>::thread_default_precision' requested here
  198 |       m_old_prec = R::thread_default_precision();
      |                       ^
/usr/local/include/boost/multiprecision/detail/precision.hpp:178:7: note: in instantiation of member function 'boost::multiprecision::detail::scoped_default_precision<boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>>::init' requested here 
  178 |       init(has_uniform_precision() ? R::thread_default_precision() : (std::max)((std::max)(current_precision_of<R>(a), current_precision_of<R>(b)), (std::max)(R::thread_default_precision(), current_precision_of<R>(c))));
      |       ^
/usr/local/include/boost/multiprecision/number.hpp:228:52: note: in instantiation of function template specialization 'boost::multiprecision::detail::scoped_default_precision<boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>>::scoped_default_precision<int, int, boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>>' requested here 
  228 |       detail::scoped_default_precision<self_type>  precision_guard(v1, v2, *this);
      |                                                    ^
cpp_complex.cpp:15:13: note: in instantiation of function template specialization 'boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>::number<int, int>' requested here
   15 |     Complex z1{0, 1};
      |             ^
cpp_complex.cpp:47:5: note: in instantiation of function template specialization 'complex_number_examples<boost::multiprecision::number<boost::multiprecision::backends::complex_adaptor<boost::multiprecision::backends::mpfr_float_backend<0, allocate_dynamic>>, boost::multiprecision::et_off>>' requested here
   47 |     complex_number_examples<mpc>();
      |     ^

[jzmaddock]

That smells like a bug (and a big gap in our testing if so), will investigate...

[jzmaddock]

Confirmed, should be fixed in referenced PR.

Note that your test case still fails due to setting the stream precision to INT_MAX or something similar - numeric_limits is not functional for complex types, and variable precision types have no meaningful value for numeric_limits<>::digits10.

[ckormanyos]

Great find. I'd like to see if the new coverage process can pick up the new tests. Awesome!