We had discussed doing this earlier, but it seems we missed it.
@AndrejSafundzic Start and End nodes are something we do because it makes the algorithm easier. But these dummy nodes leak into our actual cluster as well i.e. cluster.nodes, flow, node_processes; everything contains references to these.
We should not need to expose this at all! Anything outside bootstrap_flow() needs zero knowledge about a dummy start and end node. It makes our graph look very restrictive visually speaking. Also from a interface design of view, it doesn't make much sense.
It should be a super small change, please remove this as soon as possible.
We had discussed doing this earlier, but it seems we missed it.
@AndrejSafundzic Start and End nodes are something we do because it makes the algorithm easier. But these dummy nodes leak into our actual cluster as well i.e. cluster.nodes, flow, node_processes; everything contains references to these.
We should not need to expose this at all! Anything outside bootstrap_flow() needs zero knowledge about a dummy start and end node. It makes our graph look very restrictive visually speaking. Also from a interface design of view, it doesn't make much sense.
It should be a super small change, please remove this as soon as possible.