I’m developing a library using google/angle as WGL, GLES implementation, and glad as GLES loader. I was wondering whether I could rely on the fact that all my windows would be using the same instance of angle libEGL.dll. More precisely, according to glfw/egl_context.c at 3.3.8 · glfw/glfw · GitHub glfw would only load a single instance of libEGL.dll (and also libGLESv2.dll, which we can see at glfw/egl_context.c at 3.3.8 · glfw/glfw · GitHub )
The reason I’m asking is, that if it is something I could always rely on even in future versions, then I would be able to use a single glad context for managing every window, thus simplifying my code.
Maybe that should be obvious, but I thought that it might not be the case, and wanted to clarify this. Also, is there any mention of that behaviour in the docs?
The EGL libraries are only loaded once, but I would not rely on using a single glad instance for all your windows, as each window has it’s own GL context and additionally on multi GPU systems (common on laptops for example) it is possible for windows created on one display to have a different GPU and different capabilities to windows on created on another display.
It’s possible that Google’s ANGLE abstracts that away for you so you only see one device, but you will still have one GL context per window and I would still recommend using one Glad context per window as per Glad’s documentation and examples. If you write your window creation code to also create/free the Glad contexts this should lead to the same amount of code (but slightly more memory) as for one global Glad context.