PI 1.8RC5 crash with Repaired HSV Separation Script

[Christoph Puetz]

Dear PI Community,

I became interested in this link
http://pixinsight.com/forum/index.php?topic=5149.msg37404#msg37404
and tried Repaired HSV Separation Script today.
I also found a reproducable PI Crash under WIN7 Professional x64,
obviously caused by some Pixelmath expression.
Please see attached screenshots.

Maybe this crash is related to this bug-report (same reason?):
http://pixinsight.com/forum/index.php?topic=5422.msg37254#msg37254

I wonder why the faulty modules is a windows dll "kernelbase.dll" ?

Then I rescaled the image to 50% size and the error occured exactly as above again,
at the same line in the script (deconvoltion, then pixel math, then rescale, *crash*)

Can someone also reproduce this ?

Christoph

[Jan]

I have experienced exactly the same problem, also on W7 64.

[Gary]

I have the same problem. Gets to the same point of rescaling as seen on your screen capture in the consol window then crashes.

[Bob Andersson]

Hi folks,

I wrote the script. The PixelMath expression prior to the crash is "vwid + max(0, repairWindow.mainView.id - data.WhiteClips/4)". The purpose is to add a little extra signal to those bright pixels where non-linearity of the sensor is becoming an issue. If a mask isn't in use then the expression is created at line 614 of the script and executed at line 625.

So far so obvious but I can't see why that should cause the script to crash. It doesn't using the current official release (1.7) under Windows 7 x64. That said the next line in the script is "repairWindow.forceClose();" at line 627. Is it possible that this crash is related to this statement? The reason I ask is because of the current issue 1.8 RC5 crashes when closing image Windows 7. I'm wondering if my script has inadvertently fallen foul of this PI bug but if some expert scripter out there can see an issue with my script code then say the word and I'll get on it straight away.

Bob.

[Christoph Puetz]

Hi Bob, I think that the script itself is not the real cause of the problem.
I had crashes even with other scripts under Windows (-> Bug report section).

But it is helpful for Juan (our PI Master) for debugging, as he said in another post, that he
rewrote a lot of code inside PI.

So we should wait for Juan to investigate this - he surely has a lot of work to do
and therefore we should give him some time. It is also possible, that this behaviour can be
corrected with some other patches that will come soon.

Thanks again for providing your script - this is really great work !!

[Juan Conejero]

Hi,

This is the same problem that we have had with the DarkStructureEnhance script:

http://pixinsight.com/forum/index.php?topic=5401.msg37421#msg37421

This bug has been correctly identified and will be fixed next week, probably with an RC6 version.

In the meanwhile, the workaround in this case is very easy: in line # 627, replace forceClose() by close(). I have attached a modified script that works without problems on Windows 7 x64 with PI 1.8.0 RC5.

Please note that this bug has nothing to do with the crashes that some users are experiencing when closing windows, which we have not been able to reproduce.