Since forever, X an XV imagesink has been caching the active pool. The gain of caching the active pool is very small compare to how much negotiation flexibility it removed. When x(v)imagesink propose an already in use pool, the receiver won't be able to change it's configuration unless it is stopped, pipeline drained and started again. It is also harmful in the case an filter moves from passthrough to not passthrough, as offering the active pool will lead to two owners. When the original owner is done with the old pool, it will stop it, which will cause the second owner to eventually receive FLUSING on acquire call.

In this architecture, there is no limitation that would enforce this caching. Hence I suggest we remove it.