Hi,
Though the annotations under labview 2009 have behaved 99% of the time since my upgrade, I have just been able to duplicate what I saw in my last post, this time with data. I have not been able to duplicate the issue on full runs or with a debug routine since a couple weeks ago. But on the latest run, after running for the last few days, I went to reload the tdms file to look in more detail at the data. When I did, I noticed the annotations were way off to the right at the beginning of the run, which strangely enough, do not appear that way in the graph where the data is being acquired. This time I have been able to save some data for you to analyze.
So a little more info, my routine is saving the annotations to the tdms file itself, using "tdms set properties", where the group is "AnnotationArrayElementX", and the property names and property values are associated with "index", "name" etc. When we want to look at the data to post analyze, the standalone routine then reads these annotation groups and re-populates the annotation array, so they will display. To make sure this was working correctly, I copy pasted the annotation array from the vi which wrote the tdms, and the vi which read the tdms. I see the annotation data as being exactly the same, and I have attached vi which contains both arrays. But what is very strange is, notice the label offset X and Y. They are not 0 and 100 as they are being set in my vi. To start off with, X is a very very large integer, and Y is something other than the hard coded 100 by some small amount. And correspondingly, the annotations appear way off to the right, although they do not appear this way in the vi which wrote the annotations (?!)). When the annotations begin to behave later on in the run (yes they do correct themselves), the X offset goes back down to some small number, as you can see in the arrays around element 124. This change to a very small value corresponds to the appearance in the graph, where they seem to straighten out, from being pushed way to the right initially.
So there is some calculation going on with the label offsets behind the scenes, which under certain circumstances, causes a very large integer value to be stored to the X offset, instead of 0 (or a very small number, which I am not sure how that is being generated).
You can run the attached vi, which is a simulation of what I am doing in the full run. It is just placing an anotation every 2000 samples. Take a look at the label offset number in the annotation array after the annotation is placed, and you will see it is some number other than 0, which under certain rare circumstances, is a very large integer.
Info on the attachements:
GraphFileSave.png - This is a screen shot of the graph as it appeared in the vi which wrote the tdms file, containing the annotation data.
GraphFileLoad.png - This is a screen shot of the graph as it appeared in the vi which read the tdms file.
Graph_when th ings got better.png - A screen shot of the graph at the point where suddenly the X offset values in the annotation array went from a very large integer to something very small (other than 0 however), around 16 hours into my test. FYI, the previous run where I posted I was not able to get detailed data, was the opposite. That is, it started out ok, then 13 hours into the run the annotations went way to the right.
Annotations.vi - This contains the arrays which I copy pasted from the vis which wrote and loaded the annotation data, and they are so named with their control labels.
Annotation debug routine for NI.vi - The stripped down version of our standalone vi, all non-applicable code either removed or disabled.
The tdms file itself is 18mb, which I assume I cannot attach here, but perhaps I can compress and try if you would like to look at that as well.
So though the problem is much better, this is hit and miss enough where
it is effecting my data gathering in a very negative way. Can someone please take another look at this? Again I am indeed running LV 2009 now.
Thanks
David Jenkinson
