I discovered this rather odd KNIME AP behaviour which is on the face of it caused by a poorly wired loop, but I wonder if it is indicative of some other hidden issue.
The workflow presented here is available on the hub so, if interested, you can try it out for yourself
There is annotation in the workflow to describe the situation in each case. Attempt to execute each loop by simply executing the Variable Loop End node, and see what happens.
Under some circumstances, a Component that is not part of the loop blocks processing even though there is nothing incorrect about the Component. In all but the final case, there is a branch in the loop that is not attached to the Loop End, but executing the Loop End causes this branch to attempt execution. So the loop is not “well wired” but it generally works. However, a component ahead of the branch can, for some reason, not execute correctly.
Watch the videos, to see what I mean (there’s no audio but hopefully they are self-explanatory):
Obviously the flows shown here are contrived and this shouldn’t generally cause an issue if the flow is properly wired, but it is odd that (as far as I can see) the component causes problems here that does not occur for a regular node, and that sometimes the component does not cause a problem.