Again.
It has to do with the fact that you cannot shift a vertex of a plane that’s able to be drawn into a scene at a low enough cost to run for practical purposes in the way that a rip curl would need it to be shifted.
Youd need tons of tris on your water mesh, and while dragging them up/down is trivial (what waves normally do) introducing x/y motion into the vertex mathematically becomes problematic / makes things look really janky.
Best you could hope for is to reach a near vertical plane as the leading portion of the curl.
Unfrotunately there is no way that the plane would be vertical when you sweep a wave across the mesh - you would always have a tilted plane between 2 verts.
That’s essentially way no geometry nodes or mathematical solutions for this will probably ever exist.
Sure, given enough mesh density and a controlled scenario you could make it somewhat work currently.
But when you use 30k tris on a single 2m tile of water there isn’t much else you can do with the rest of the scene…
Hence why an alembic cache simulation is probably better anyway?