Sounds to me that others might find this useful. How about making a little write up, and maybe a small test/demo, zipping it up and uploading to the serialization section of the boost vault? Robert Ramey. Todd Greer wrote:

Hi,

I'm using the Serialization library, but need to skip unrecognized data when reading in an archive. This is for forward compatability purposes; I need for older versions of my software to consume files produced by newer ones, which may have added elements.

As this functionality is not currently in the library, I have created a new archive type (as I think Robert Ramey suggested at some point). It inherits from xml_iarchive_impl and overrides load_override to identify when it has just read an opening tag with an unexpected name. It then skips whatever XML it finds within the unexpected element, reads the closing tag, and returns to normal processing. I use it with the unaltered xml_oarchive, and the pair satisfy my requirements.

I'm including it below in the hopes that someone will find it useful. I've tested it on MSVC 7.1 (Visual Studio .NET 2003) with a reasonably complex XML file with both nested and non-nested unexpected elements. If you'd like, I can post some tests as well (I just need to strip out some project-specific details).

Please let me know if you see any places where it could be improved, or where I've completely missed something; this is my first use of the serialization library, and my understanding is far from comprehensive.

Many thanks to all who've contributed to the Serialization library, Todd Greer Senior Software Developer, Affinegy LLC

#include #include #include <algorithm>

const char whitespace[] = " \t\n\r";

//Skip whitespace, then return true if we're looking at '

assert(ch == '<'); ch = is.get(); bool const end_tag = ch == '/'; is.putback(ch); is.putback('<'); return end_tag; }

template<class Archive> class skip_xml_iarchive_impl: public boost::archive::xml_iarchive_impl<Archive> { #if BOOST_WORKAROUND(BOOST_MSVC, <= 1300) //see xml_iarchive_impl public: #elif defined(BOOST_MSVC) friend boost::archive::detail::interface_iarchive<Archive>; protected: #else friend class boost::archive::detail::interface_iarchive<Archive>; protected: #endif

skip_xml_iarchive_impl(std::istream & is, unsigned int flags = 0): xml_iarchive_impl<Archive>(is, flags) {}

void skip_all_elements() { std::string ignore; if(this->This()->gimpl->parse_string(this->This()->get_is(), ignore) && ignore.npos != ignore.find_first_not_of(whitespace)) return;

//We're either looking at a start tag or an end tag. while(!at_close_tag(this->This()->get_is())) {//It's a start tag. if(!this->This()->gimpl->parse_start_tag(this->This()->get_is()))

boost::throw_exception(archive_exception(archive_exception::stream_error ));

++depth; std::string const name = this->This()->gimpl->rv.object_name; skip_all_elements(); load_end(name.c_str()); } }

template<class T> void load_override( #ifndef BOOST_NO_FUNCTION_TEMPLATE_ORDERING const #endif boost::serialization::nvp<T> & t, int ){ if(!t.name()) return xml_iarchive_impl<Archive>::load_override(t, 0); load_start(t.name()); // don't check start tag at highest level const std::string name_found = this->This()->gimpl->rv.object_name; bool const skip = depth > 1 && 0 != (this->get_flags() & no_xml_tag_checking) && name_found != t.name(); if(skip) skip_all_elements(); else boost::archive::load(* this->This(), t.value()); load_end(skip ? name_found.c_str() : t.name()); if(skip) //We still haven't loaded t... load_override(t, 0); //...so load t. }

template<class T> void load_override(T & t, BOOST_PFTO int i) { xml_iarchive_impl<Archive>::load_override(t, i); } };

class skip_xml_iarchive: public skip_xml_iarchive_impl { public: skip_xml_iarchive(std::istream & is, unsigned int flags = 0) : skip_xml_iarchive_impl(is, flags) {} };