TestStand execution stops sometimes

[Scott_Richardson]

Ofer -

I am sorry, TestStand 3.5 did allow for the font change as I mentioned above, but the difference between 3.5 and 4.0 is that 4.0 added some code to address a font scaling issue in 3.5, and this font scalling issue introduced a leak of DC handles.

[bruno_p]

Scott,

Thanks for your help.

Many sequence files have been developped (by about 12 different developpers). I cannot modify all sequence files.

And the bug is not predictable so I cannot just modify one sequence file or try your workaround on a test machine.

I'm waiting for an official patch.

But I would like to do some tests as soon as possible.

I think that I found the code section where the font size is set (in the msgbox cvi project) and where the device context is used.

Can I correct the bug if I just add a call to the ReleaseDC() function at the end of the code block which calls the GetDC() function ?

 Bruno

[Scott_Richardson]

Bruno -

Yes, that is what is needed to fix the issue, something like this:

//need to convert font size to pixels
HDC hDC = 0;
int scale, sizeInPixels;
nullChk(hDC = GetDC(NULL));
scale = GetDeviceCaps(hDC, LOGPIXELSY);
ReleaseDC(NULL, hDC);

[Rand_H]

Hello Scott & Co,

Like some other people I have this problem in TS 3.5 and I am a little bit confused about bug presence in TS 3.5 or not. Is your solution only for TS 4.0 bug?

Colud you confirm if R&D people is focused only in this 4.0 font leak error?

Rgds,

Hector.

[Scott_Richardson]

Hector -

The font related leak only occurs in 4.0 and later, so the current fix is to recompile the source as indicated above. The issue that some have seen in 3.5 has not yet been diagnosed because we have only hit the 3.5 issue once during testing. It is unclear whether this 3.5 issue also occurs in later releases. We are still trying to investigate this further.

One question is, what is the version of the cvirte.dll file located in the windows system32 directory on systems that have a problem with TestStand 3.5?

[Rand_H]

Hello Scott,

Our version of cvirte.dll is 8.0.0.273

Is it bad one?

Rgds,

Hector

[Scott_Richardson]

Ofer, Bruno, and Hector -

I wanted to give you an update. It took some time to debug the hang issue, but I found an issue in some code that the TestStand step type uses from the CVI RTE. When we can fix this issue, it will likely require you to install a newer version of the CVI RTE. Until then the only way to prevent this occasional hang from occurring is to not require the step to internally use a separate thread and an IE control to display content. So this limits you to displaying only BMP images, and not PNG, JPEG and HTML pages, which require the IE control. I am still investigating other options and will let you know more as I go.

[Rand_H]

Hello Scott,

I'm very happy to see you are still investigating. We will try to change HTMLs to BMPs to analyze if it solves or not these hangs in our machines.

Thanks so much,

Hector.

[bruno_p]

Scott,

After some experiments, I already adviced the developers to use bmp images (and I adviced it in a previous post).

I'm very happy that you confirm this workaround and that you found the bug.

I had some other unexpected behaviours (but not popup freeze) that seem to disappear with your previous bug correction.

Currently I have not any more complaint of the users.

I will check that and confirm you that I have no more bug.

Thank you very much for your great investigation.

Bruno

[ofer_o]

Hello Scott,

Thanks a lot for all your efforts. I'm keeping on with jpeg files because tha problem rarely occurs, and I'm waiting for the solution.

Thank's a lot

Ofer