I believe this has been fixed in the more recent version of the library. 
 
Robert Ramey
 
Avi Bahra wrote:
> I am using boost 1.39 on suse linux with gcc 4.2.1 and using a debug
> build.
> In order to rationalize the long compile times, I was attempting to
> split a file
> into multiple translation units. When I ran my tests, they all
> passed, but then
> crashed with a memory fault.
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xb7cd26d0 (LWP 21897)]
> 0xb7d40a08 in strcmp () from /lib/libc.so.6
> (gdb) bt
> #0  0xb7d40a08 in strcmp () from /lib/libc.so.6
> #1  0x08e19924 in
> boost::serialization::extended_type_info::key_unregister ()
> #2  0x08e19ace in
> boost::serialization::extended_type_info::~extended_type_info ()
> #3  0x08c41afc in ~extended_type_info_typeid (this=0x92179cc) at
> /var/tmp/ma0/boost/boost_1_39_0/boost/serialization/extended_type_info_typeid.hpp:80
> #4  0x08c41bbe in ~singleton_wrapper (this=0x92179cc) at
> /var/tmp/ma0/boost/boost_1_39_0/boost/serialization/singleton.hpp:104
> #5  0x08c209e0 in __tcf_25 () at
> /var/tmp/ma0/boost/boost_1_39_0/boost/serialization/singleton.hpp:121
> #6  0xb7d00599 in exit () from /lib/libc.so.6
> #7  0xb7ce9fe8 in __libc_start_main () from /lib/libc.so.6
> #8  0x08a833f1 in _start ()
>
> It appears that boost serialisation type registration has problems
> when it is split into multiple files. Also it appears
> that most of serialisation test all appear in a single file, hence
> these kinds of issues can easily be missed.
> Is there any work around to this problem ?
>
>   Best regards,
> Ta,
>    Avi
>
>
>
>
>
> _______________________________________________
> Boost-users mailing list
> Boost-users@lists.boost.org
> http://lists.boost.org/mailman/listinfo.cgi/boost-users