I don’t quite understand what encoding the SDL functions (specifically SDL2 ) work with.

For example:

SDL_Surface *SDL_LoadBMP(const char *); const char *SDL_GetError();

Is it utf-8 ?

8-bit encoding in system locale?

Latin-1 ?

  • The idea should be UTF-8. You can check and tell us. :) - HolyBlackCat
  • I would like to know for sure) - user294535

2 answers 2

Judging by the migration guide, one of the key changes was unicode support. All lines for text input events go to UTF-8. In addition, SDL_CreateWindow explicitly indicates what UTF-8 is expecting. Most likely the correct answer is UTF-8.

    Why beat around the bush if we have the source ?

    SDL_LoadBMP is a macro that passes a file to the SDL_RWFromFile function without changes.

    In the function SDL_RWFromFile there is any different platform dependent code:

    • for Android, it is used whenever possible fopen, and if not, then Android_JNI_FileOpen , which uses the NewStringUTF function, which accepts some modified UTF-8 ;

    • for Windows, windows_file_open is used, which using the WIN_UTF8ToString macro (in which SDL_iconv_string) gets the Windows UTF-16LE string from the input UTF-8 encoded string;

    • for Apple systems, inside SDL_OpenFPFromBundleOrFallback , fopen is used, but there is no exact information about it, but the file system uses UTF-8;

    • for Windows RT, fopen_s is used, the exact information about the encoding of which I also did not find, but if it is similar to the fopen function, then the ANSI format is probably used;

    • for all other systems, file is transferred to fopen without changes, and the encoding depends on the platform and system settings, although in modern Unix-like operating systems this is most often UTF-8.

    Thus, if I correctly interpreted the source code and documentation, the answer is obtained - it depends on the platform and build parameters .

    It is possible that strictly other UTF-8 is used in other functions of the SDL that are not related to the file system, but apparently, you can pick up problems from just SDL_LoadBMP.

    By the way, the SDL_RWFromFile documentation states that UTF-8 is used. It turns out, lying?

    • How all is difficult. I've been thinking how to cope with the zoo encodings, and the problem turns out to be more fundamental. So it makes sense to use only the Latin everywhere? After all, the characters ascii and utf-8 with codes 0-127 are the same. - user294535
    • "It turns out, lying?" most likely just shortcomings on the principle of "good enough". - Petr Abdulin