Completely agree that routing could improve in many ways on many points. I even started typing this idea as just that, but I decided to keep it simple to test a theory on the idea exchange:
Minimalist ideas get more kudos and get them quicker. Plus they have a better chance of acceptance by R&D...
If that iceberg of potential in routing was tapped I think all the debate around the cleanup tool would jump forward quickly. I generally put nodes about where I want them or at least grouped in layout patterns I intend to keep. The wires are the problem and if they were intelligent from the start the cleanup would be much easier. For fun, select a jumbled area in some code, deselect all nodes, hit cleanup and watch it squirm!
NI seems to be adding features every new release, which suffer from exactly from this type of irritating limitation. I am thinking of wire labels, subdiagram labels, etc... and I bet the same will happen with the new version.
I am not against new features, but eventually, handling the problems with them, as a user, turns out to consume more time than they are supposed to save (dropping a free label, in the two cases above). Not mentioning getting NI's attention on the bugs...
Those are GRAPHICAL issues, which you would think that developers of a GRAPHICAL programming language would be selected for being able to grasp and triple-test during development, but apparently that is not the case.
The issue you raise is a typical example. How you are supposed to wire the two remaining elements without wire crossing is beyond me. But yet, NI developers didn't think of the problem. And I bet that automatic cleaning would result in something unsatisfactory most of the time as well...
I think the NI strategy has secretly been to build things like this to a critical mass until users breakdown and accept the cleanup tool as their master.
I guess most of you know that you can change the wiring by hitting the space button before placing the wire (see graph below). Yet, I thought that it might be useful to mention in this discussion. Surely there are limitations to this method. E.g. if you try to do the same when connecting the 2 in the graph, it totally messes up the wire by inserting 4 bends. Also, the result seems to be dependent on the direction you draw the wire in (left to right or right to left)...
I saw this post and thought "OMG, I can't wait to kudos this, I'd kudos it 15 times if I could". Then I kept reading and saw Frank's post and my thought changed to "OMG I can't believe I've never seen this trick, that's amazing". Then I of course went to LabVIEW to test the trick (I partially didn't believe it at first) and realized that this only sometimes works and my thought changed to "OMG this could have been so awesome...":
Moral of the story: I severely kudos making LabVIEW smart enough to do this for me. I also severely kudos fixing the space trick to make it work always. I also severely kudos any documentation that describes these super secret tricks to the unknowing user.
The space trick works better if you combine it with one more trick - after you start a wiring operation, you can left click on the BD to tack the wire to specific points. You can use that together with the space trick to get LV to do what you want.
In LabVIEW 2021, wires are routed in a way that leaves room for wiring additional terminals.