Anything that involves raytracing would need to change from having a finite camera position in world space, to having an infinite one with parallel rays. There are some ways to make the math work out the same, but in general... nope.
Like, consider that you never actually need to scale anything in a perspective view, because you literally can't tell the difference (all the rays remain the same). In an orthographic view, scaling is the only way to actually zoom in/out. Oops, any code that assumed your view is unit-scale will break.
Perspective vs orthographic literally affects every single one of those fancy features you like.