if you don't like tracking - you can turn it off for the types you want. to add your own characters between archives try something like this: { ofstream os("filename); { boost::archive::text_archive oa(os); oa << .... } // close archive - leaves stream open os << '\n'; // or whatever { boost::archive::text_archive oa(os); // another archive in the same stream { boost::archive::text_archive oa(os); oa << .... } // close archive - leaves stream open os << '\n'; // or whatever } // closes output stream Robert Ramey François Mauger wrote:
Hi
I have large amount of data to store/load from files. I use boost::serialization library to do it. It has very nice features.
I use version 1.33:
libboost-serialization-dev 1.33.1-9ubuntu3.1
Checking different strategies to play with my data and i/o archives, I met the following problem: if I save several text_oarchives within the same output file (a trick to break side-effects of memory tracking), then the deserialization failed for there is no separator between successive text archives.
I have to explicitely add a 'std::endl' in the output stream to make it run.
This pb does not appear with xml archives for the </tag> at the end is unambiguous to parse the end of each archive. I did not check for binary archives, but I guess there will be no pb.
For me a mandatory 'white' character should be added as the last byte in a text output archive (when destructor is invoked?). This will make more coherent (symmetric!) in comparison with xml/binary ars. This is only a suggestion: I cannot imagine all the side-effects such a strategy could imply.
A sample demo file is attached.
Thanks for your attention.
frc
_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users