Fatal Internal Error fpsane.cpp line 442 - can't find the insane object, even with Heep Peek window

[csalwen]

According to NI support:

From my testing, it appears that the only time I don't
experience the crash as you described is if I open/run/save the vi in
LabVIEW 2009sp1 under Windows 7.  Various other versions of LabVIEW that I
have tried on both my Windows 7 machine as well as the XP install all crash
with the error you have reported using the steps provided in the initial email.

He later updated me:

From what I can tell, R&D was
able to successfully identify the issue that was causing the crash you were
experiencing.  I would like to thank you for your help in providing a means
by which this problem could be easily reproduced as this definitely aided
in the process.  The issue is slated to be fixed in a future release (and
the release containing the resolution to this problem should document the
fix in the release notes). 

I am using Labview 9sp1 on a Windows XP machine.  I'd think if you called support they should have some sort of notes, started with my call and the bug report and maybe they could tell you more.  I don't know if they'd like more examples of the problem.

[DavidZ.]

To omas:

do you have an example vi or snippet of that "connect a void array to the X input of a General Histrogram V" ?

To csalwen:

thanks for posting your NI support answer. In the meantime i continued to work in LV8.5 and didn't get the error there, but of course I'd like to be able to work on a newer LV-version, too.

[DavidZ.]

Finally back to LV2010, I took the time with around 30 forced LV-crashes to delete all grease away from my fpsane-error-creating-VI to isolate the problem (as using the Heap Peek window didn't help) and reduced the VI from almost 700KB and subvis to one single 25KB-big VI (attached), also replacing all previously customized controls with standard ones. 

By the way, the error occurred in LV2010 and LV2010SP1 under WindowsXP and WindowsVista, but no problem in LV8.5.1, I couldn't test combinations with LV2009 or Windows7.

Then it seems I have a similar problem as described in omas' post: Empty array to Input X of General Histogram.vi going to XY graph.

So far I didn't find a better solution than to feed an array with one value to the General Histogram.vi, in case an empty array is "approaching" the General Histogram.vi. (vi also attached)

[csalwen]

Your example vi's are the same as far as I can see, but it I understood what you mean and it does appear that you've found the problem.  I can tell you that the problem is in LV2009 as well.

Thank you for tracking this down.  I had figured out that there was some weird behavior when there was an empty array input to the histogram, but I hadn't realized that it was causing the crashing problem.  Too bad support didn't mention it.

I had stopped using the General Histogram and instead I had made a simpler subvi that calls the dll that is at the heart of the General Histogram.  I've motified that to not call the dll if there is an empty input.  I actually saw some pretty weird behaviour while I was making that fix.  I was trying to put a case structure around the call library fuction node.  When I did this with the call library fuction node  inside the case and the indicator array I was writing the histogram output to outside the case structure, the wire from the call library fuction node would turn into a void wire.  I ended up putting the Histogram indicator in the case with the dll call and putting a local in the case for an empty array input.

[omas]

Hi Davidz

I have made a example of what caused the problem for me.
It takes a couple of runs to get it to crash.

Regards.

[DavidZ.]

I forgot to mention a very important action that provokes the crash:

The vi (vi attached) crashes directly and reproducibly when you let it run first (with Histogram Source=Empty Array) and then try to save the vi. 

to omas:

thanks for your sample vi. When I try the same with your vi, I mean to let it run, then save, the crash also happens directly.

to csalwen:

sorry, yes I put 2 times the same vi with different names. Here now attached the corrected vi creating the fpsane.cpp error. By the way, could you post your workaround sample with that subvi calling the dll that is at the heart of the General Histogram for better understanding? thanks! 

[BPerlman]

Benthescientist, thank you.   You helped me with a completely unrelated problem!  You mentioned copying the block diagram and pasting to a new VI and I thought, "hey, that's a great idea, why didn't I think of that" and tried it on my problem VI (with disappearing scrollbars).

It helped me.  Thank you!

B.

[csalwen]

Here you go.  These examples are LV2009sp1.  I removed all the flexibility in the General Histogram since I knew I only wanted to use it in one way.  The histogram was in a loop since I had some very large data sets so I separated the Bins and the Axis calculations and put them outside.  This lent itself to fixing the call by doing a check for an empty array and returning a histogram of zeros and a -1 as an error code when the input was an empty array.  The -1 was pretty random.  The dll call returns some other negative value when the input is empty, but it also returns a histogram with some large, seemingly random numbers in some of the first few elements of the histogram array.

[jochembax]

Hey,

I got the same error in my program.

My program also has to save and load a lot, and has a remote connection with the web publishing tool.

My program works fine until i try to acces the program with a different computer.  If i go to the website (build by the web publishing tool"), the program shuts down and gives this error. 

When i try to acces the website on the same computer where i run the program, there is no problem and it works fine.

Has anywone a answer for this problem?

Jochem Bax

[csalwen]

Are you using the histogram?  Have you talked to NI technical support.  My problem was with the histogram, so it sounds like a different case.