Introducing LabVIEW 2009

[RavensFan]

---

Ben wrote:

A moving average?

Say you want to averge the las t20 readings, the value that bubbles out of that terminal would be subtratced from the running total and the new value added in.

Ben

---

---

Dan_u wrote:

1. This was possible before with shift registers (just expand them), but afaik not with feedback nodes.

It can be useful for filters (even though you usually need several previous values) or for instance in simulation, to simulate a delay of 4 cycles use a z^-4.

Daniel

---

Quick forum comment.  Ben, your post said it was viewed 0 times.  Dan's immediately afterwards was viewed 7 times.  How is that possible?  I know you commented on a 0 view issue before.

Are you using feedback nodes in your post so that we are seeing the View count from 7 or 8 iterations ago? 

I was aware that shift registers could be expanded, although I can't remember a situation where I needed to actually do that.  Most of the examples I had seen of expanded shift registers were for small moving averages such as looking at only 3 or 4 of the past iterations.  (It would be really impractical to create a loop with the SR expanded to hundreds of nodes)  But they usually used all 3 or 4 nodes in the same loop and summed them together.  So you needed i-4, i-3, i-2, i-1 all in the same iteration.  It does sound like the this delay feature of the new feedback node does sort of resemble expanded SR feature that has been around for a while.  But they really wouldn't quite be the same.  Shift registers could give you all data of the last several iterations at once, but the negative is that you can only practically do it for a relatively small number of iterations.  The new Feedback node could give you a piece of data from a long time ago (hundreds or thousands of iterations ago),  but there is no practicaly way to get the data from all of the iterations in between.

I think I see how Ben's example of the  moving average would work.  You would basically use the feeback node with a 20 iteration delay so that you can set up a series of iterations and extract that i-20 value.  You would also need another normal feedback node that is set for 1 to store the sum.  Take the data from the sum FB, subtract the data coming from the i-20 FB, add the new piece of data to sum and pop it into the i-20 FB

In reality, I think if an averaging scheme got too complicated, I would just package it up in an action engine.

Message Edited by Ravens Fan on 08-27-2009 11:17 AM

[RavensFan]

---

Ben wrote:

I'm trying to learn the new icon editor. I suspect it is based on some software that I have never touched (it sure ain't paint) so it is taking me some time to learn. Drawing a line with the right-mouse button (for the background color) is no easy feat when you are left-handed.

But I have a question!

What is this button supposed to do and how do I use it?

Curious,

Ben

---

It says it Moves.  As far as I can tell, when you use it, it gives you a hand that lets you drag that entire layer around.  I don't know how useful it is.  Under the old Icon Layer, I would move things, (such as text around a pixel or two here or there to center things up.  But since I had a black border, I would only want to move the text, so I would just lasso the text and shift that round.  With layers, now you can have different things on different layers, so I guess I would keep my border on one layer and only move the text layer around using that tool.

[tst]

One important note about the move button is that it does NOT always move the current layer. If you click on an object which is found in another layer, it will automatically grab and move that object. This is very convenient if you're aware of it, but can be quite confusing if you're not.

Ben, I would suggest you try PJM's refactored editor. It has some advantages over the shipping version.

[Ben]

---

tst wrote:

One important note about the move button is that it does NOT always move the current layer. If you click on an object which is found in another layer, it will automatically grab and move that object. This is very convenient if you're aware of it, but can be quite confusing if you're not.

Ben, I would suggest you try PJM's refactored editor. It has some advantages over the shipping version.

---

Oh yes.

I kept clicking on blank area that I thought were within a selected area. There has to be a color in the pixel that is clicked for that to work.

Re: custom versions of anything

I need to be able to walk into a bare room isolated from the outside world and code with what comes out of the NI box. Getting used to any other luxury will only hadicap me when I am on site. Saying things like "Well back at my office I would/have ..." will do no good. So when you head off to your next customer visit with you memory stick* full of custom stuff, increment you blessing count becuase the rest of have to make do with "stone knives and bear skins".

Ben

* Pens and paper only. Evn those are searched both coming in and going out.

** When having my car inspected at one my frequent customers sites, the gurads did there normal "No guns or weapons" where I normally reply "...and a Swiss Army knife does not count" when I realized I was driving my wifes car and the trunk was full of her swords she uses for Kung Fu! The guards made an exception for me that day.

[Neil.Pate]

I am just starting to play around with 2009, and maybe its just me going crazy, but the UI seems a little more responsive than previous versions. i.e. right-clicks open context menus quicker, etc.

Is it just me?

If it is a change under the hood then extra kudos to NI.

caveat: I am tinkering around with a tiny project by comparison to what I am normally used to, so it could just be this...

[mfletcher]

---

a100man wrote:

Seriously where is the documentation? There is NONE! I tried the parallel for loop and one of the loops that the wizard Identified as a "green light" had a bunch of shift registers, and the data was corrupted by the parallel process. No run arrows were broken or any indication that this was not a viable implementation. Just bad data. So my question is this:

Where can I find the documentation on how this feature is supposed to work? What happens when you use a shift register? I need more than the one paragraph help screen that doesn't explain much!  

---

I realize I'm joining this thread late, but I am curious about your problem with the parallel for loop. LabVIEW can parallelize loops with shift registers in two situations: 1) when it can prove that the iterations read/write disjoint portions of the data in the shift register, or 2) when the shift registers are used to compute a reduction (see the "Parallel For Loop Reduction" example). If there are shift registers and neither of those situations apply, LabVIEW will break the run arrow. (LabVIEW will produce warnings for other things that may or may not be safe, so be sure to enable warnings.)

If the wizard gave you a green light for those loops but they produced incorrect results, I would like to investigate this. Can you post the VIs?

As for the documentation question, I'll forward this on to the appropriate people.

[jchrisj]

I read your post yesterday after I discovered that my data processing code is now broken in LabVIEW 2009 because of the MathScript issue. I haven't heard from my sales engineer yet so I don't have an answer to my critical question - is the MathScript RT module available for LabVIEW under Linux?

thanks,

Chris

[Dennis_Knutson]

Have you seen this.

Message Edited by Dennis Knutson on 01-21-2010 01:45 PM

[jchrisj]

No, I had not found that listing, and thanks, it answered my question.

Chris

[GrantM]

Hello Chris,

The MathScript RT Module is available on Linux.  Note, however, that the MathScript RT Module provides MathScript support for both the desktop and the RT platform.  Since LabVIEW for Linux does not support deployment to RT, the MathScript RT Module for Linux will only support the desktop platform.

Grant M.
Senior Software Engineer | LabVIEW MathScript | National Instruments