Improving the LabVIEW Help: Working with Data Types

[LRohre]

Have you ever had trouble figuring out how to work with the waveform data type, the dynamic data type, or some of the other more complex data types in LabVIEW? As a tech writer on the LabVIEW team, I'd like to improve our documentation about working with data. What would you like to see? What would be helpful? Have you ever given up on a particular data type because it didn't work? Have you ever created a replacement data type because you preferred not to use a LabVIEW data type?

[Ben]

I'd like see the format of each.

The advantages and disadvantages

Bench-marks moving large data set ( 1M plus) around using each type.

Resoultion of time stamps.

Eaxamples on how to use each type with graphs and charts and how to exploit each or disable attributes.

Storing and retrieving.

Dangers of data definitions changes with LV version etc.

Ben

[Matthew_Williams]

To expand on an item from Ben's list, the format on disk of the various data types,
as well as the format on disk of clusters and arrays would be very useful.

Structure of variants.

Techniques to calculate memory footprints for various structures, including
clusters and arrays.

Matt

[andre.buurman@carya]

Hello,

For me the digital waveform was very difficult to understand, specially compared to the analog waveform.

The concept of the Signal data type also remains some what of a mistery, what are the advantages

Regards,

André

[johnsold]

I try to avoid the dynamic data type, at least in part because it does not seem to be well documented or the documentation is not easy to find. (Context help is no help on this one.) Further, it seems that the functions to convert to other datatypse are not located in any obvious places on the palettes. Polymorphic behavior is quite counterintuitive. A recent posting on the Forum had a VI with a dynamic datatype connected to a comparison function (like >0). The output, rather than being a boolean was still dynamic data. It apparently worked like a boolean, but the wire obsured more than informed. Similarly coercion to display a dynamic type on a scalar indicator (DBL) does not seem to be well defined. What, exactly, will be displayed?

Is the dyanamic datatype actually good for anything?

I do not find the waveform type particularly useful, but at least it seems to be clearly defined and the get waveform components and build waveform primitives are easily located. Very often I disassemble the waveform, operate on the array, and then reassemble it. I might be happier if some of the functions which only accept or generate waveform datatypes were available with array and dt inputs or outputs.

I started with LV 1.2, so I may be biased against some of the new stuff. If an advantageous use case and the means by which to use the "feature" can be demonstrated, I will look at it.

Lynn

[falkpl]

I second the details for variants.  I am only recently using them since it took me many years and gerat help from this post to get a good grasp on variants to get comfortable vith them. 

[parthabe]

In my opinion, the Event Structure help is of no help.

When I hover over the mouse over any of the data of the Event Data Node, I expect its description pops-up after I press Ctrl=H; but it wont...

If this is given it ll be of immense help for everyone, esp beginners, by avoiding to go into the detailed help each n every time.

This remains as I said above even in 8.5. I m using 7.1.

NI must look into this before LV 9.0, atleast...

[Kevin_Price]

I've been programming LabVIEW since around 3.1 / 4.0.  Over the years and versions, I've found it necessary to approach a lot of the new datatypes and algorithms with some healthy skepticism because I don't think we're usually given a "fair and balanced" overview.  Some of the things that are highly promoted that bring "ease of use" for beginners (and I *do* understand the importance of that when you're growing your LabVIEW user base) can carry significant performance penalties that are mentioned in more of a whisper, if at all.  I'd just like a more full disclosure of the tradeoffs -- the good, the bad, and the ugly.

Waveforms -- very rarely use them, unless required for analysis functions.  I have a vague notion (between a belief and a memory) that they used to carry a significant performance penalty compared to arrays, but that the gap is now much smaller.  Still, I'm most comfortable with my old habit of using data arrays.  Some of this came about because a lot of my work uses counters where I may have a variable "dt" value or where my AI is clocked by a counter and the "dt" isn't internally generated.

Dynamic signals -- have never used them.  Have no clear idea what good they're supposed to be, and they seem to be tied in primarily with Express VI's, which I have also studiously avoided.

Variants -- Use them only slightly.  Performance issues are a question mark.  Exception: have learned that the implementation of Variant properties allows their use as an efficient way to create associates for string lookup tables. 

Digital Waveform -- have typically avoided it except when graphing digital data during debug.  Haven't found any compelling reasons to use it.  I do like the notion of a compress/expand capability for sparse digital data, but haven't exercised it enough to trust the implementation. 

Recap:  I don't need more simple "how-to's" in the help.  I need a LOT more "why bother's" that include both pros and cons. 

-Kevin P.