Hi there, There is a C++ library for which I'm working on an FFI wrapper. I've written a translation unit which defines several functions with an extern "C" API. Each of the functions does a little bit of C++ to operate the third party library, but the point is that I now have a C API that I can wrap with the FFI. The next thing I did was to try and get Cabal to compile this translation unit for me. I eventually hacked together a Setup.hs that I barely understand that calls gcc to compile the .cpp file and then calls `ar` to, um, well do whatever it is that ar does with .o files, make some sort of archive, or something. Next, I need to get my hsc file (in which I define some imported functions and some data types with Storable instances for shovelling data through the FFI) to compile. This file uses the hsc2hs #include macro to include the .h file of my translation unit. In turn, that .h includes some of the C++ header files from the third party, and eventually an STL header: vector. At this point, the compiler, gcc, complains that it can't find vector. The gcc command that hsc2hs (I think?) is creating does not include -lstdc++, and nethier does it use g++, either of which would, I think, result in a successful compilation of the hsc2hs .c file. In my cabal file I have: extra-libraries: stdc++ ghc-options: -pgml g++ -Wall But these don't seem to fix anything. Perhaps they don't have any effect on how hsc2hs operates? There's a specific question here which goes something like: how do I get hsc2hs to compile code that uses STL headers? But there's also a more general question as to whether my whole approach of creating an extern "C" wrapper is actually going to work? Any answers would be very appreciated. Richard