Messages

1

Feature requests / Re: SFML_SYSTEM_ANDROID is a platform, but we should test on features

« on: May 20, 2022, 10:22:20 pm »

Yes, exactly.

I know it might be a little bit slow to load single files, but it would provide a way to load data from an external point of view.
Meaning that you don't have to package everything in the apk file, but will be able to build for example, a "launcher" application that will start some games from any location on the SD card.

2

Feature requests / SFML_SYSTEM_ANDROID is a platform, but we should test on features

« on: April 23, 2022, 01:08:23 pm »

Hi everyone,

Lately I've been playing around with SFML 2.5.1 on Android (I'm currently using it at the core of a ruby based engine, where SFML is used in a dynamic library (so) file).
Things work pretty well, but for modularity reasons (the engine should be able to load different games with various resources), I need to externalize loaded resources from any directory location of the Android device.

Meaning that, I want to specify an explicit filesystem location while using the existing internal SFML loaders functions `loadXXXFromFile`, and not having to explicitely load everything in memory on the ruby side then pass it to SFML using `loadFromMemory`.

I would suggest adding a CMake constant `SFML_SYSTEM_USE_FS` instead of directly testing on the platform `SFML_SYSTEM_ANDROID`.
Of course, the constant would not be defined by default on Android, but would be on other platforms.

Example : Line 120 of 'src/SFML/Graphics/Image.cpp' would be something like

bool Image::loadFromFile(const std::string& filename)
{
    #ifdef SFML_SYSTEM_USE_FS
        return priv::ImageLoader::getInstance().loadImageFromFile(filename, m_pixels, m_size);
    #else
        priv::ResourceStream stream(filename);
        return loadFromStream(stream);
    #endif
 }
 

And the same goes for `Font.cpp`.

What do you think about this request ?
If the community agree with it, I'm able to push the changes in a Pull Request on the official repository.

End note :
Currently, I'm patching SFML 2.5.1 to suppress the if preprocessor directive.