daldegam/eos-portable-archive

eos-portable-archive's Issues

Error loading file in boost 1.69.0

I can't find portable_binary_(i)oarchive in boost/archive.
so, I tried to copy those files in eos to boost/archive.
Then, it is different from tutorial, I can't use boost::archive::portable_binary_oarchive but eos::portable_oarchive could.

    template<class Archive>
    void Map::serialize(Archive &ar, const unsigned int version)
    {
        // don't save mutex
        ar & mspMapPoints;
        ar & mvpKeyFrameOrigins;
        ar & mspKeyFrames;
        ar & mvpReferenceMapPoints;
        ar & mnMaxKFid & mnBigChangeIdx;
        ar & mapScaleRatio;
    }
    template void Map::serialize(eos::portable_oarchive&, const unsigned int);
    template void Map::serialize(eos::portable_iarchive&, const unsigned int);

Next, when I compile code, it will show the warning:
The use of BOOST_ENDIAN and BOOST_BYTE_ORDER is deprecated. Please include <boost/predef/other/endian.h> and use BOOST_ENDIAN_BYTE instead
Finally, It saved the file but I can't load it.

Help me find where wrong, thanks.

boost 1.69 support

Hi,
In the boost 1.69 release the boost::spirit library has decided to:

Drop own FP routines in favor of boost::math PR#392 #13531
(https://www.boost.org/doc/libs/1_69_0/libs/spirit/doc/x3/html/spirit_x3/changelog/spirit_3_0_3.html) .

This, of course, creates a bit of an issue. Is it possible to please fix it for boost 1.69+ ?

Thank you.

error: no type named ‘bits’ in ‘traits’ {aka ‘struct boost::math::detail::fp_traits_native<double>’}

A little background: I'm updating boost 1.67 to 1.74. I also updated eos-portable-archive from a previous version that didn't support boost >= 1.69 to the most recent one.

So, in portable_oarchive.hpp, we have these asserts around line 393 in the save overload that saves floating point numbers:

		template <typename T>
		typename boost::enable_if<boost::is_floating_point<T> >::type
		save(const T & t, dummy<3> = 0)
		{
			typedef typename fp::detail::fp_traits<T>::type traits;

			/** <snip> **/

			typename traits::bits bits;
			BOOST_STATIC_ASSERT(sizeof(bits) == sizeof(T));
			BOOST_STATIC_ASSERT(std::numeric_limits<T>::is_iec559);

Before updating, everything worked fine. After updating, I'm hit with an error that says the bits member of traits in the snippet above is not a member of boost::math::detail::fp_traits_native<double>:

/dependencies_directory/eos_portable_archive/eos/portable_oarchive.hpp:393:26: error: no type named ‘bits’ in ‘traits’ {aka ‘struct boost::math::detail::fp_traits_native<double>’}
    typename traits::bits bits;

Right before the asserts, there's a comment that says that, if I got to this point in the code, one of the following was likely the problem:

1. I'm serializing a long double which is not portable (not the case)
2. I'm serializing a double but have no 64 bit integer (not the case)
3. My machine is using an unknown floating point format (unlikely? It's just a 64 bit CentOS 6)

These comments assume I even get to the asserts below anyway, which I don't, so I believe this is something else? Also, these checks were in place in the previous version of boost + eos, so I don't think it has anything to do with my architecture.

Anyway, I went and checked which traits struct was this and then I get this in boost/math/special_functions/detail/fp_traits.hpp:

template<class T> struct fp_traits_native
{
    typedef native_tag method;
};

It's just a tag type. For some reason, the native traits struct is being selected, but it clearly isn't expected here.

In the header section of portable_oarchive.hpp, I see this:

// endian and fpclassify
#if BOOST_VERSION < 103600
#include <boost/integer/endian.hpp>
#include <boost/math/fpclassify.hpp>
#elif BOOST_VERSION < 104800
#include <boost/spirit/home/support/detail/integer/endian.hpp>
#include <boost/spirit/home/support/detail/math/fpclassify.hpp>
#elif BOOST_VERSION >= 106900
#define BOOST_MATH_DISABLE_STD_FPCLASSIFY
#include <boost/math/special_functions/fpclassify.hpp>
#include <boost/endian/conversion.hpp>
#else
#include <boost/spirit/home/support/detail/endian/endian.hpp>
#include <boost/spirit/home/support/detail/math/fpclassify.hpp>
#endif

If I go over to fpclassify.hpp:

template<class T> struct fp_traits
{
    typedef BOOST_DEDUCED_TYPENAME size_to_precision<sizeof(T), ::boost::is_floating_point<T>::value>::type precision;
#if defined(BOOST_MATH_USE_STD_FPCLASSIFY) && !defined(BOOST_MATH_DISABLE_STD_FPCLASSIFY)
    typedef typename select_native<T>::type type;
#else
    typedef fp_traits_non_native<T, precision> type;
#endif
    typedef fp_traits_non_native<T, precision> sign_change_type;
};

Since BOOST_MATH_DISABLE_STD_FPCLASSIFY is defined, shouldn't it choose the fp_traits_non_native traits struct? Why is it choosing the native one? I'm not sure what's going on.

And this is as far as I've got. I'm lost at what is happening here and I'm not familiar with this code.

I hope this helps you investigate.

Where is portable_binary_(i)oarchive

In your tutorial, these scripts include boost::archive::portable_binary_(i)oarchive.
However, I don't find portable_binary_(i)oarchive in boost/archive.
Should I copy scripts that in eos folder to boost/archive?
But I tried to copy them and include them, like tutorial :

#include <boost/cstdint.hpp>
#include <boost/archive/portable_oarchive.hpp>
#include <boost/archive/portable_iarchive.hpp>

then I still can't find
boost::archive::portable_binary_(i)oarchive

sorry. please delete this issue