For legacy projects still using OpenGL, a common approach is to first convert OpenGL calls to Vulkan via solutions such as Zink and then land on Metal through MoltenVK, forming a multi-stage translation chain; this chain introduces extra CPU overhead and potential compatibility deviations, and debugging difficulty rises markedly, so it is better suited as a transitional measure than a long-term architecture.