On Feb 5, 2008 2:52 PM, Joshua Jensen <[EMAIL PROTECTED]> wrote: > I'll give a simple example, if I'm understanding the request. For the PS3, > you link SPU code in the PPU executable. SPU code is just .c/.cpp files, > possibly with libraries linked in. In any case, there are two separate > toolchains needed, one for PPU code and one for SPU code.
Yes, that's a good example. In my case it's an in-house toolchain for code which runs on a PCI device: a host program loads code onto the device. It's not the embedding that's the key thing, it's simply being able to build two libraries in one invokation of make in a dependency-aware fashion. It might sound a bit perverse, but it's necessary since the host code needs to know locations of certain symbols in the device's code. If I were invoking a second CMake, would I be able to make the host CMake aware of the second as a dependency? I have a feeling this could get messy. I guess an alternative approach would be to implement a set of custom macros analogous to ADD_LIBRARY et al, which are aware of another toolchain. John _______________________________________________ CMake mailing list [email protected] http://www.cmake.org/mailman/listinfo/cmake
