This doesn't fix the root cause of #223, and there may still be an issue with how errors are being handled (I'm wondering if this would leave a sync primitive in a permanently unsigned state, so maybe we need to tear down more of the swapchain in this case than we do currently.) But this seems like a reasonable improvement for now.
This doesn't fix the root cause of #223, and there may still be an issue with how errors are being handled (I'm wondering if this would leave a sync primitive in a permanently unsigned state, so maybe we need to tear down more of the swapchain in this case than we do currently.) But this seems like a reasonable improvement for now.