You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.
Support for codecvt is provided by the standard library rather than being strictly tied to a compiler implementation. By default, GCC will use libstdc++ as its standard library implementation. On OS X you can actually change that using the compiler flag -stdlib=libc++ as described on this stackoverflow page.
You indicated however that you're on Debian Linux so you don't have this luxury. You are also correct that, as of this writing, libstdc++ does not have full support for codecvt. Particularly, it is missing codecvt_utf8, which is the only thing nanodbc uses from that header. It may be possible to refactor the code to avoid use of this utility, for example using boost.
Note that the code path which makes use of codecvt is only entered in two cases:
When a wide string data column is fetched as a std::string.
With unicode support enabled for nanodbc, when a string data column is fetched as std::wstring.
If your code isn't doing either of these things, as a workaround you could replace the codecvt_utf8 calls with code to throw a runtime exception indicating that these operations are not currently supported on Linux.