I have just noticed that Catch and Complete nodes do not show a config error if they are not connected to any nodes. This happened because I deleted the node that was being monitored by a Catch node and then recreated it, this meant that errors were no longer caught but there was no indication that the Catch node was no longer connected. It took me a little time before I realised what was going on, particularly as I had named the catch node in a way that indicated which node it was monitoring. A classic case of the comments in code not matching what the code is actually doing.
Logically, if no node is associated with the config node, the latter should itself trigger the error. Currently, errors emitted by a config node are not catchable by the catch node. The PR above is, of course, one part of the solution.
For me there issue here is that a config node is generating an exception. Should config nodes have that much logic that they could possible generate exceptions? I thought config nodes only store configuration which is then used by the actual workspace nodes to create connections or set values - which then might cause exceptions.
