LabWindows/CVI app terminates under Win XP when file select popup is open, but everything works fine on Win 98

[David G]

Thanks for the suggestions.  However I don't think the runtime lib is  installed because I specifically tested this, and my app will not run if in a dir by itself, only if in it's own build dir that has the lib files there.

Anyway as I mentioned the issue is resolved for me, everything works perfectly with V5.5 so I am not going to waste time trying to debug V7.1.

And sure I agree with your comments that XP has it's flaws, but that is why companies pay NI $600+ for an upgrade is to handle that stuff.  If NI expects me to debug XP for them then like I said we should get a refund on this supposed "upgrade".

Am I the only one that thinks it's ridiculous that their V7.1 doesn't work on a plain vanilla XP install, but their 1999 compiler (V5.5) works just fine?

Thanks,
David

[David G]

By the way mvr,

Out of curiosity why did you upgrade to 7.1?  Was there something it offered you that the previous version did not?

Thanks,
David

[mvr]

My first involvement with 7.0/7.1 came when working on systems that would be delivered to a customer.  They were not attached to an internal network so they got their own copy of cvi.  Some of the debugger features in 7.1 are very slick.  The ability to look at arrays as images, step back through code, and designate the next point of execution is very useful.  The fp file generation tool has saved me considerable time.  There are some features that I miss when going back to a 6.0  environment, but even cvi6 was a pretty good compiler so if you don't use the newer features, it not like you are stuck in the stone age.  It is unfortunate you have not had the same experience with cvi7/xp as I have.  I had found it to be very stable. 

You may still be able to get someone from this forum who works at NI to take a look at the issue.  The key is finding a way for them to replicate the problem.

[awd]

I have come across a similar problem.  During development I got a "Process Terminartion Error" saying that CVI could not terminate the debuggee process and that I may not be able to rebuild it and that OS may be unstable etc.  I then started getting Fatal run time errors with Unknown source position, when the FileSelect popup was used when pointing to the Desktop.  To cut a long story short, it seems my code is not the problem as the same crash occurs with CVI sample programs.  Try the following...

Load the CVI\samples\fileio\arrayfile.prj, plot the graph then save to a file "data.dat" ON THE DESKTOP.  With the program still running, plot again, and save again to the same file on the desktop.  Again, with the program still running, plot the data and try to save to the same file on the desktop.  The Fileselect popup appears and will stay there inedfinitely UNTIL you hover the mouse over the "data.dat" file WITHOUT even clicking.  All is well very briefly until the little Windows popup tries to appear that tells you what kind of file it is, and at that point it crashes.  This seems only to happen with files on the Desktop and it does NOT happen when saving files to the desktop with non-NI programs.  This happens consistently for me with different machines running XP Pro service pack 1 and service pack 2.  I am using CVI 6.0 and the problem occurs when in the IDE and with compiled and installed apps.

If anyone from NI is watching the thread I too would appreciate a solution this problem, and I have seen other posts describing similar problems.

[awd]

Further to my last message:

I have tried installing compiled versions of my program and the arrayfile program on XP machines without CVI on them, and the programs work fine on there.  I have tried repairing the installation of CVI on one machine, and removing the installtion on another and both these machines still persist with problems.  I am a bit puzzled...if the run time engine has been corrupted (as suggested elsewhere) then this would surely still manifest itself when a program is compiled and installed on another machine - but it doesn't, so long as CVI is not present on the other machine.  Perhaps CVI has damaged something in Windows (and this seems to be where the problem occurs - hence the Fatal error messages with unknown source position when in the IDE) on my machines that have CVI on them (or have had in the past)?

Any suggestions other than reinstalling XP?

[Wendy L]

Hello David,

Sorry for the delay in my response.  I have been OOO for the past week.  I would be happy to try to do whatever I can to try to resolve this problem for you, or at least get to the source of the problem.  Having said that, I would suggest going to the ni.com/support website and select Request Support >> E-Mail NI.  This will create a service request on the issue.  Also include the specs of your machine (Win XP, SP1, SP2, etc) and the executable you are running.  Also, if you go to the System32 directory on your Windows XP machine, right-click the cvirte.dll file, select Properties, select the Version Tab, what is the exact file version you see listed there?  Also, in this email that you send on ni.com/support could you attach the CVI 5 and CVI 7.1 built applications so we can do some more investigation on a few WinXP machines?

Thanks again in advance.

[Wendy L]

Hello awd,

After following the procedure you mentioned in your previous post, I was unable to reproduce the "Process Termination Error" on my machine using the arrayfile shipping example.  So when you create a distribution kit of one application then install it on another machine, it is going to also install the runtime engine of the version of CVI you compiled your application in.  Now when you run the application on the new machine, it is going to run off the latest version of the runtime engine on that machine.  If you have a more recent version of the runtime engine on that machine, it will use that version instead of the one shipped with the distribution kit.  Having said that, for the computers that are seeing the problem, which version of the runtime engine do you have on the machine (go to the System32 directory on your Windows XP machine, right-click the cvirte.dll file, select Properties, select the Version Tab)?  If it is not the latest runtime engine (8.0...) then I would suggest downloading the latest version from the following site.  After doing so, do you still the problem?

If so, to clarify, are you only able to reproduce the problem on machines that have the IDE installed? 

As in, if you create a distribution kit and deploy you application (including the CVI runtime engine) to a machine that does not have CVI or the CVI runtime engine on it, will it work? 

On the other hand, if you deploy to a machine that has CVI installed (and therefore, the runtime engine is installed), does it always crash?

And for the last scenario, if you deploy to a machine that already has the CVI runtime engine installed, but not the CVI environment, does it crash?

Thanks

[David G]

Wendy,

What awd said is absolutely correct.  The file select popup crashes when you are looking at a file on the Desktop and you hover the mouse over a file.  The first time you do it's fine, but the 2nd or 3rd time you open the File Select popup, it then crashes when you hover over a filename and the little yellow tip box was going to come up with details about the file.

I downloaded V 8.0 of the runtime dll as you suggested.  It did not make any difference.

I don't think there is any point in me emailing my app to you.  If you could not reproduce this bug there with the example that awd suggested, then you are not going be to able to reproduce it with my app either.

So I hope you guys can find a few more XP machines to try it on there.  I'm sure me and awd's XP machines are not the only ones that this will happen on.  I have XP Pro w/SP2 by the way.

Looking forward to the next version of your runtime dll that will fix this.... 

Some more notes for your team:
- App compiled on CVI V5.5 runs fine even with V7.1.1 rte dll in system32 folder.
- App compiled on CVI V5.5 has the same problem now when new V8.0 rte dll in system32 folder.  In other words, it appears V8.0 is worse than V7.1, because with 7.1 rte dll the app compiled on CVI 5.5 would run OK...
- Now after having tried several versions of the rte dll it seems things will not work for the app compiled from CVI 5.5.

So, I guess just ignore the above notes.  The important thing is for your SW team to try the example awd suggested and keep trying different XP machines until you are able to duplicate it.  Until you do this, we are not going to get anywhere.  Note that your problem might be that your IT dept has all your XP machines configured the same way.  You may need to try some XP machines with different configurations.  Maybe one of your guys has an XP machine at home or something...

Thanks,
David

[awd]

Wendy,

I have now been able to establish that the Arrayfile distribution kit fails as described previously on both machines that currently have the IDE installed.  It fails on 1 of 2 machines that previously had the IDE installed. It fails on 2 of 4 machines that have never seen NI software before.  All are running XP Pro, some SP1, some SP2 but there is no pattern to that.  The machine that previously had the IDE, but runs arrayfile ok is different in that it is a dual Xeon workstation.  The two others that did not fail are almost virgin XP systems.  Those that fail are less than 2 years old and from different manufacturers. Those that fail (and the workstation) run McAffee antivirus, but turning off on-access scanning does not solve the problem.  All machines had cvirte 6.0.4.105, but installing 8.0.0.273 had no effect.

As an aside, despite having installed 8.0.0.273 on the development m/c, CVI still generates distribution kits with 6.0.4.105, but I don't know where it is getting the DLL from!

I don't suppose this helps narrow down the problem.

[David G]

I got an email saying there was a new post, but don't see one, though interestingly the last post was shown as by awd but it's not showing up in the thread... I guess the NI msg board may not be working so well...

Anyway, Wendy, have you had any luck in reproducing the priblem??

Thanks,
David