IA64, GCC 4.3, redefinition of 'struct boost::serialization::implementation_level<std::vector<long int, std::allocator<long int> > >'

[Stefan Eckardt]

Hi,

I get the following error message compiling with gcc 4.3.0 on ia64:

/usr/local/include/boost/serialization/vector.hpp:185: error: redefinition of 'struct boost::serialization::implementation_level<std::vector<long int, std::allocator<long int> > >'
/usr/local/include/boost/serialization/vector.hpp:185: error: previous definition of 'struct boost::serialization::implementation_level<std::vector<long int, std::allocator<long int> > >'
/usr/local/include/boost/serialization/vector.hpp:185: error: redefinition of 'struct boost::serialization::implementation_level<std::vector<long unsigned int, std::allocator<long unsigned int> > >'
/usr/local/include/boost/serialization/vector.hpp:185: error: previous definition of 'struct boost::serialization::implementation_level<std::vector<long unsigned int, std::allocator<long unsigned int> > >'

This error can be reproduced with this small example

#include <boost/serialization/vector.hpp>
int main()
{
	return 0;
}

Removing the lines

    BOOST_SERIALIZATION_COLLECTION_TRAITS_HELPER(signed long, C)     \
    BOOST_SERIALIZATION_COLLECTION_TRAITS_HELPER(unsigned long, C)   \

from boost/serialization/collection_traits.hpp solves this problem for me.

Thanks for considering this issue.

Stefan

[Robert Ramey]

Hmmmmmm - I don't think removing the above lines is a great idea. This would make your archives which use vectors unusable for programs built with a different compiler.

This needs to be investigated some more. Is it possible that on this compiler "long int" is a synonym for some other type such that is also in the list? e.g int64_t ?

As I said - needs more consideration.

Robert Ramey

[anonymous]

Replying to ramey:

You could test it by moving the declaration of line 185 to the front. Then the "bad" declarations should cause the error.

[Robert Ramey]

I think this has been addressed by fixing another problem. We'll keep an eye on it.

Robert Ramey

[jlp@…]

Replying to ramey:

I think this has been addressed by fixing another problem. We'll keep an eye on it.

I have this same problem with GCC 4.3.1 on AMD64 on Linux and I would like to ask you which one was the other fix so I can test it here.

[stefaan.deroeck@…]

But it seems this was already fixed in the 1.36 release, my apologies. I should have been more thorough.

[f.a.oliehoek@…]

Hi,

I am a bit confused here. What exactly was fixed "I think this has been addressed by fixing another problem" ? Is there a simple way to get it working on my current debian/lenny distro on AMD64? Should I follow the advise of Stefan Eckardt?

(versions that are currently installed: ii libboost-serialization1.34.1 1.34.1-11 serialization library for C++ ii g++ 4:4.3.2-2 The GNU C++ compiler )

Or will version 1.34/35 simply not work? And is the only solution to upgrade to 1.36? (something I'd rather not do, as I've had bad experiences with upgrading to new boost versions)

Thanks, Frans

[fa.oliehoek@…]

p.s.: It seems that the problem hit me since the gcc upgrade:

[2008-10-06] gcc-4.3 4.3.2-1 MIGRATED to testing (Britney) ​http://packages.qa.debian.org/g/gcc-4.3.html

Thanks!

[gary.mirams@…]

See this website for a quick fix: ​http://www.nabble.com/-usr-include-boost-serialization-vector.hpp:126:-error:-td20801412.html

Replying to fa.oliehoek@uva.nl:

p.s.: It seems that the problem hit me since the gcc upgrade:

[2008-10-06] gcc-4.3 4.3.2-1 MIGRATED to testing (Britney) ​http://packages.qa.debian.org/g/gcc-4.3.html

Thanks!