On 2013-11-08 14:15, Peter Rosin wrote: > On 2013-11-08 12:18, Panicz Maciej Godek wrote: >> 2013/11/8 Peter Rosin <p...@lysator.liu.se <mailto:p...@lysator.liu.se>> >> >> The SDL library, for some obscure reason, has its own special take on >> that and >> prescribes that you should keep using main() even if you are doing a GUI >> app. >> I think the SDLmain library contains the real WinMain@16 entry point and >> that >> entry point in turn calls the application main function. Or I should >> perhaps >> say SDL_main (see that -Dmain=SDL_main define above). Some part of this >> fragile SDL crap fails. I don't know what. >> >> Perhaps the SDL_main library was compiled to expect an ordinary main >> entry >> point instead of the GUI WinMain@16 version? >> >> Just to be clear, I'm not an SDL user. This is just my understanding of >> this. >> The above description might very well be flawed in some way, but SDL >> initialization is peculiar. >> >> >> I can later try to add some of the options from the libtool invocation >> generated >> by autoconf to my invocation of gcc to see which particular option causes the >> failure and then let you know. I think that your description of the way SDL >> does >> things on mingw is sound (and I think that the goal is to ensure >> portability, as >> unix programmers have no idea what the WinMain is) > > Hmmm, I have this hunch that the -nostdlib option that libtool adds to the > g++ invocation beats -mwindows, just like it beats -pthread. But I don't > know that for a fact...
No, wait. You're not building a library, so -nostdlib isn't relevant... Sorry for the wasted bandwidth. Cheers, Peter _______________________________________________ https://lists.gnu.org/mailman/listinfo/libtool