This should now be easily doable, and will simplify a lot the topic interfacing.
Note we wouldn't be able to easily remove a publisher, while an interfacing sub is also setup, but this isolates the issue to a more specific use case.
Implementing dual update loop (one for all state, one for diffs only) both running, would make that cleanup simpler, and more solid for multimaster (no need to rely on get_num_connections, instead we wait for the full state loop to run and cleanup leftovers...)
This should now be easily doable, and will simplify a lot the topic interfacing.
Note we wouldn't be able to easily remove a publisher, while an interfacing sub is also setup, but this isolates the issue to a more specific use case.
Implementing dual update loop (one for all state, one for diffs only) both running, would make that cleanup simpler, and more solid for multimaster (no need to rely on get_num_connections, instead we wait for the full state loop to run and cleanup leftovers...)