LabVIEW 8.0.1 Runtime Crash w/ Application.exe - Urgent!

[Hannamon]

Hi Kenny;

Thanks for the continued input; no, one of the first things I did was to mass compile the entire project (for other IMAQ issues with v8.2, I am sticking with v8.0.1 for now), but this had no effect.  Also, I cannot say with absolute certainty that the problem is actually the subpanels, but all the available evidence seems to be pointing in this direction.

I am using some proprietary code, so I unfortunately am not willing to post it, however, I may be willing to send you a copy directly if you think it will help.  My next step, at this point, is to build further complexity into my GUI to eliminate any unnecesary subpanels (i.e. copy the SubVI interfaces and add them to the GUI and communicate by means of queues), and see how that works, since when I add subpanels and extended directory structures, I can make WinXP crash also.

Let me know.  Thanks!

DJH

[Hannamon]

Here's something else I noticed with the subpanel displays in my GUI: as mentioned, I have subpanels contained within tab displays.  For example, I have an "Image Display" tab which contains subpanels on two tabs, an "Image Tools" tab which contains subpanels on three tabs, and an "Image Acquisition" tab which has subpanels on two tabs.  For each tab control containing multiple subpanels, if I set one subpanel setting to "Synchronous", all remaining subpanels on different tabs within that same single tab control adopts the same properties, while subpanel properties on different tab controls remain unchanged.  Why is this?  I would have expected each subpanel display to be completely independant of each other?  Is this due to my tab setup?  And is it possible this is tied in with my Win2K crashes?

DJH

[Kenny_K]

One thing that I came across is that someitmes when you add fiels to be loaded dynamically/set as include only if referenced, the front panel gets removed.  This is in the source file settings.

I tried to create the subpanel with a tab, within a tab and tried it on a 2k machine and it still worked, I based it off the example that you created.

Kenny

[Hannamon]

For any dynamically loaded files, I always select the "Dynamic VI" setting in the builder.  I found out early on that if I used the "Include only if referenced" that it was removed.  This is not the case here.

And insofar as my provided example, yes, it does work in both WinXP and Win2K.  Of course, this is a far less complex instance of my program I am working with, but the subpanel usage is the same.  In other words, if I am using the subpanels correctly in my example, I am therefore using it correctly in my program.

As mentioned earlier, if you think it would help, I would be willing to send you a copy of my code to see if you are able to figure anything out that I cannot, I just do not want to post it on this forum.  Let me know.  Thanks!

DJH

[Kenny_K]

sure, if you want to email me at kenny.kreitzer@bradleycorp.com, I will take a look at it as soon as I can.

I agree that if you can do it in an example, it should be done corrctly in the main program.  But it nver seems to go that way.

Kenny

Message Edited by Kenny K on 12-04-2006 08:36 AM

[Hannamon]

Hello All;

Still no solution for the Win2K crashes.  I've pretty much gone through as much as the forums (items relating to Runtime crashes) as I can, and where-ever there was even a remote similarity, it has been investigated.  Here are some of the items investigated and their reults:

- In LabVIEW 8.0/8.0.1, when selecting a VI to go to a particular destination directory during application build, it only sends the selected dynamic VI to be placed there.  LabVIEW 8.2 seems to place any subservient associated VI in this directory as well.  This being the case, it may have been possible that there was some sort of conflict going on here.  I redid some code to allow me to place everything within the DIAAS.exe (i.e. Application.exe) file during runtime.  Result: Win2K still crashes with WinXP fine (i.e. no change).

- There were threads regarding the List Folder Contents vi being buggy or simply not working in LabVIEW 8.2.  Since my program makes use of this to identify certain Add-On VIs in separate folders (during runtime and development system), I replaced all of this with an INI file listing of all applicable framegrabber drivers and add-on client VIs for this applicable release, thereby eliminating the search-and-add-to-list functionality of the List Folder Contents VI.  Result: no change.

- There were still more threads regarding problems during initialisation of the program during runtime, particularly when it came to the use of shift registers.  Since my program makes use of shift registers, I looked into this.  My program uses shift registers that are initialised in a separate case statement that is run immediately upon startup via the use of a "First Call" function.  I rewrote this to eliminate the need for the First Call and I now initalise the shirt registers using a constant on the leftmost side of the register.  Result: no change.  By the way, this also includes going through the code (painstakingly) and making sure that all representations (i.e. I32 --> I32) were being converted properly.

- Still another thought was that if I used subpanels imbedded within tab displays, that there might be a conflict there (this brought on since setting a delimiter such as "Synchronous Display" for a single subpanel within the tab display caused all subpanels contained within the tab display to take on the same feature.  So, code was rewritten to limit a single subpanel per tab display.  Result: no change.

So far, there have been no favourable results in this endeavour.  Others (thanks, Kenny!) have also looked at my source code without finding anything that would cause Win2K to crash while WinXP is fine.  If I do not use subpanels at all and simply use multiple windows the program is fine.  Also, with subpanels, the program is fine during development mode.  Upgrading to LabVIEW 8.2 does not help.

At this point, should this simply be attributed to a subpanel bug?  If so, how is this brought to NI's attention?  Or is there another option I am overlooking?  Thanks!

DJH

[Kenny_K]

You can formally ask NI for help at http://sine.ni.com/apps/we/niae_asc.main

[Hannamon]

FOUND IT!!!!!!!!

Turns out, the problem had absolutely NOTHING to do with anything I thought it was.  Subpanel usage was fine; tab control usage was fine; dynamic VIs were fine.  The problem was when I was polling the VI Is Frontmost property.  For some reason, this was enough to cause Win2K to crash while WinXP remained stable.  I have attached an example (not the actual code) of what I was trying to achieve.

The idea for my code, as I've mentioned before, is that for actual usage, there is a single runtime GUI, where all the other VIs are displayed by means of subpanels and tab controls.  However, the code is also set up to be run as individual windows, mainly for debugging.  I had run into a problem previously where graph displays were not being updated with new ROIs or images, mainly because they were in the background.  Bringing them frontmost also did not necessarily update them, either.  So, I wrote in some code to detect when the VI was frontmost, so that it could automatically update itself.

Turns out, that for individual windows, this seems to be fine.  I had no problems.  However, when this same polling code is invariably included in the single-GUI version, apparantly BIG problem!  It was originally thought that since the VIs were in the subpanel, they would never actually be Frontmost, and so would never activate the circuit.  This is true, but this code was also included in the Main GUI also, for similar reasons.  Therein was the mistake.

So anyway, now my code has been corrected, and it is off to the next bug in my code.  Thanks to all who offered input.

DJH

[Kenny_K]

Congrats on figuring it out

Kenny