Unbundle by name nodes incorrectly re-link when defining type is modified.

[McKenzie]

Under LabVIEW 6.1 I am seeing the following problem: If the type of a bundle or unbundle by name node is changed, for example by modifying a typedef, the node sometimes creates erroneous links to a sub-element with the same type. This only happens when the cluster contains two or more sublusters of the same type.

To duplicate the problem, you can do the following:

Create a cluster called 'Subcluster.' Put two numerics in it, 'Numeric' and 'Numeric 2.'

Create another cluster, called 'Cluster.' Put two copies of Subcluster in it, calling them 'Subcluster,' and 'Subcluster 2.'

Now go to the diagram and create a constant of the cluster. Wire this constant to an unbundle by name node, and i
ndividually select out the four Numeric elements. The four outputs on the unbundle by name should now read 'Subcluster.Numeric,' 'Subcluster.Numeric 2,' 'Subcluster 2.Numeric,' and 'Subcluster 2.Numeric 2.'

Now, go to the constant, and delete Subcluster 2. As you do so, watch what happens to the Subcluster 2 outputs of the unbundle by name node. Instead of becoming invalid, they magically transform from 'Subcluster 2' to 'Subcluster.'

This appears to be a reversion of a LV 4.0 bug documented in the knowledge base. (see doc. no 0NT8T17I)

Is there a workaround for this, or more detailed information about exactly what circumstances provoke it?

regards,
McKenzie

[Shan]

I followed your procedure and got your results in 6.1. In version 7, the unbundle appears to behave correctly (i.e. the names of the deleted cluster numerics become invalid).

Doesn't help you much in 6.1, but . . .

Tim

[McKenzie]

Thanks, Tim.

I appreciate you performing that check. At least NI must be aware of it if they fixed it for version 7, and that's a Good Thing. 😉

I'd still be interested in knowing if there is any kind of work around or a methodical way to identify when this happens. Our scenario is that sometimes we change a typedef in a fairly large code base, and this can result in hard to spot re-wirings of bundle and unbundle nodes. This kind of behavior can lead to bugs which are very hard to track down.
Anyway, thanks again.

regards,
McKenzie