Hello all : I'm having great difficulty with boost::filesystem::path (again) and wondering if anyone else has seen this behavior. I'm using vc7.1 (msvc 2003) testing with both static and dynamic link boost::filesystem libraries. The behavior is as follows : 1. The boost regression tests all pass ( ie, in BOOST_ROOT/status running "bjam --dump-tests" and sorting through the output ). 2. If I use fs::path in my code with an absolute windows path ( ie c:/temp ) it _always_ throws. As an example, when I include the following code from the fs test file path_test.cpp in my own code it fails : fs::path p = fs::path( "c:/", fs::native ); assert( p.relative_path().string() == "" ); assert( p.branch_path().string() == "c:" ); assert( p.leaf() == "/" ); assert( p.root_name() == "c:" ); assert( p.root_directory() == "/" ); assert( p.root_path().string() == "c:/" ); assert( p.has_root_path() ); assert( p.has_root_name() ); assert( p.has_root_directory() ); assert( !p.has_relative_path() ); assert( p.has_leaf() ); assert( p.has_branch_path() ); assert( p.is_complete() ); 3. The exceptions are being chucked by the checker code; if I use boost::filesystem::no_check it doesn't fail. The checker complains that "c:" is invalid but clearly it's not ! 4. Before anyone mentions it, the rtls are consistent across all libraries including boost. Perhaps that's the problem though? Do people use boost successfully with /MD and /MDd? Has anyone else seen similar behavior or have any suggestions about what may be happening? As a long shot is this something to do with wchar_t settings on msvc ( /Zc:wchar_t )?? If it's not too much trouble (and I'm assuming it doesn't break the license ) can someone send me the debug build of boost::fs lib (either static or import/dll)? I'll happily pay you back in good kharma! Many thanks in advance; Steve a.