Ability to wire property/invoke nodes interlinked

[BrotherTom]

It would be nice if property/invoke nodes supported "interlinked wiring" among eachother. Something like this:

At the moment I move the wires manually behind the nodes. This saves some space and (in my opinion) contributes to a well-arranged block diagram. But do not select "Clean up wire" afterwards!

[CMal]

Since you're using the same method multiple times on the same object, wouldn't it make more sense to just use a for loop?

[RnDMonkey]

CMal, I think your advice is good for this specific case, but misses the general nature of the request. At its core, the question is basically "can property/invoke nodes' write entries pass their inputs out the far side, unmodified?"

I don't think having a hidden pass-through feature of the propert node entries would be bad, but I think it does raise the question of how you mutate between reads and writes. I don't know enough about LabVIEW guts to know if that's hard, or maybe it's easier to just modify the LabVIEW EXE to treat the far side of the entry as invisible to wires? That seems like it might be a risk for other non-associated wires trying to use the space, but there are always opportunities for error when wiring nodes are deep inside a graphic. It's one reason I don't like the fact that the actual wiring point is in the middle of the node and not where the little arrow is on the edge. 🙂

[AristosQueue (NI)]

This would be a very bad thing. It would create serialization among things that do not need to be serialized generally. In the specific case of control refnums, sure, the property nodes are serialzied because they're all in the UI thread, but for many other types of property nodes, that's just not true. We should never be creating dataflow dependencies for pure value functions. There are some old functions in vi.lib that do this for things like "duplicate path", but we avoid that today. Pass-through terminals for non-reference values are just a bad idea -- all you do is defeat the natural parallelism of LabVIEW.

[RnDMonkey]

Stephen,

You point about serialization is well taken. Would it be possible to make it merely a graphical change that allows the wire to come out the other side rather than steer around the property node boundaries? I don't have a vested interest in this, really, just want to see the idea played out to its conclusion.

I completely concede that having an pass-through that is an actual output terminal is not a good idea. In this specific case it wouldn't make a difference, but as a general rule it would not be so.

[AristosQueue (NI)]

No, I don't think so. If it comes out of the node, it's a connection. To me, that's misleading for the diagram. I think that going around is *good*, not *bad*. And I think that this strung together case that is in the idea is very rare -- generally it is either separate refnums or it is inside a loop. In the separate refnums case, then routing around is a good thing to show no data ordering dependence. If it is inside a loop, there aren't multiple copies of the node.

[Darren]

Any idea that has not received any kudos within a year after posting will be automatically declined.