PDA Cluster Limitations

[Biff_Whiffle]

I am evaluating the PDA module for Pocket PC 2003. Within my application is a typedef cluster containing singles, doubles, U8's, U16's, U32's, strings, and several arrays of numeric types. It doesn't contain any clusters, just scalars and arrays. It has probably 40 elements. The cluster is one element of the larger cluster of information being circulated within the while loop through the shift register.

This typedef's elements are populated within one case inside the while loop of a queued state machine, then passed through the shift register to the next iteration. When some of the cluster elements are unbundled in the next iteration, the strings are empty, the numerics are zero and the arrays are empty. It's as if all the data is lost going through the shift register, or is never bundled correctly in the first place. The other elements of the larger cluster are unaffected. Most of them are simple scalar, path or array types.

Any clue as to why this is happening?

[tst]

I haven't tried using typedefs in the PDA module, but I know it can't display clusters, so I'm wondering how it would handle a typedef'ed cluster. Try removing the rest of your code and running just the populate case and you will be able to tell. I have used clusters in the module, so if this doesn't work, there might be a problem with using typedefs. Try looking at the documentation.
Other things you can try as troubleshooting is running it with the cluster not as a typedef and checking the data before feeding it into the SR. Personally, I believe the most likely culprit to be the typedef.
Besides all that, I have found the PDA module to be "not fully baked". It has many annoying bugs and behaviors. I have no information on this, but hopefully version 8, which will probably come out in a few months, will be much better.

[Biff_Whiffle]

Thanks.

I agree with your assessment of the PDA module. I hope the next version is less agravating!

[tst]

I still suggest that you narrow the problem down to something and if it is a bug, report it to NI. At the very least, they can fix it for the next version. I have found several annoying "small" bugs, like event structures not working when checking for events on different controls, charts not updating when they're not visible and things not being hidden when they should be. If it really is a bug, try making a small example demonstrating and send it to NI.

[Biff_Whiffle]

Here is a simple vi to illustrate the problem.

[tst]

Here's what I came up with: The shift register and the state machine are irrelevant to the problem. I think the problem actually comes from trying to unbundle a cluster that's inside another cluster. I tried to demonstrate this myself, but for some reason LV kept crashing when I tried to save the VI. So here's what you should try (Make sure you don't have anything important open).
Start a new VI. Create the same cluster from scratch. Wire it into a "bundle" and wire the result into an "unbundle by name". Select the data to be unbundled. Place a while loop or an event structure after this, so you can see the results. Now, I'm assuming you didn't see the values. Try using unbundle to extract the cluster and then use unbundle by name on the cluster. Does it work now?
If this went as I described (like I said, my LV kept crashing, so I couldn't check), then the problem seems to be in unbundling a cluster inside a cluster by name and is therefore a bug. Email NI with the link to this thread to make sure they see it. The solution would be to unbundle twice - first to get the cluster and then to get the values.
BTW, I use pocketPC, not palm, but I don't think this makes a difference here, since we both saw the problem.

[Biff_Whiffle]

Good catch. I've been able to reproduce your results. It might be a good workaround; I'll see if it works in all cases.

[Philip_Courtois]

Hi 'tst' and Biff,

'tst': I think this behavior has already been reported to our R&D department, but our server is currently down so I can't check.

Biff: Let me know if you didn't send us an email regarding this issue and I will then make sure that behavior has been reported so it can be fixed for future releases of the PDA Module.

Thanks to both of you!

[THailey3]

This was reported to R&D (# 3EEEC41R) for further investigation.