This documentation will explain how to quickly port FLTK to a new platform using the Pico driver system. For now, only the sample SDL Pico driver on OS X compiles and barely runs. cmake -G Xcode -d OPTION_APPLE_SDL=ON ... tl;dr - the recent commit should be transparent to all developers on other platforms. On OS X, the CMake setup add the option OPTION_APPLE_SDL=ON that will run FLTK on top of SDL, which in turn runs on top of the new "Pico" driver set. The greatest help I can get form the core developers is to continue to refactor the platform specific functionalities into the drivers. --- OK, long version. I know that I am repeating myself and I don't expect those of you who "got it" weeks ago to read this again. Writing this down is also for me to avoid losing track ;-) Goal 1: find all the stuff that must still go into drivers Goal 2: have a base driver for all future porting efforts Goal 3: make porting fun with early gratification to the coder Goal 4: extensively document a sample port (SDL), and how to improve it Goal 5: use SDL as a base library, thereby porting FLTK to iOS and Android This is the start of a new driver, named "Pico". "Pico" is here to implement what I called the minimal driver. "Pico" implements (eventually) a complete set of drivers. The drivers will be limited in functionality, but they will be good enough to allow basic UI's with most FLTK widget types. If "Pico" compiles without any "USE_PICO" or similar defines, we have reached goal 1. "Pico" will implement all driver functionalities "in good faith", meaning, in a way that FLTK runs without crashing. Only very very basic functions are not implemented. A driver that derives form "Pico" needs only to implement those missing functions, thereby reaching goals 2 and 3. As far as I can tell, those will be: - open a single fixed size window - setting a pixel in a color inside that window - polling and waiting for PUSH and RELEASE events an their position By implementing these three simple functions in the SDL driver, "test/hello" and quite a bunch of other tests will run (yes, they will be slow, but the will work). This will give the person who is porting FLTK to their platform a very early confirmation that their build setup is working and a very early gratification, reaching goal 3. Obviously, SDL is much more powerful, and by the nature of the virtual functions in the driver system, we can incrementally add functionality to the SDL driver, and document this incremental nature, reaching goal 4. If we do all this right, we have goal 5. If SDL is too big or complex for, say, Android, we can simply start a new native Android driver set by implementing the three functions mentioned above, and then go from there. - Matthias