EXCEPTION_ACCESS_VIOLATION- A fatal error has been detected by the Java Runtime Environment

[Rafael_S_]

Good afternoon,

I'm developing with Intel RealSense + Java and everything is running well, I can see image and some landmarks, but after a while the EXCEPTION_ACCESS_VIOLATION is thrown.

Someone can help me? I've attached the file with the error rescription given by eclipse/java.

OS: Windows 8.1 x64
Language: Java
IDE: Eclipse
RSSDK: R3

Thanks in advance.

[Bradley_F_]

I had similar low-level crashes when working in C#.  In my case, it was an unchecked bounds issue when working with images.  Basically, the thin layer on-top of the C or C++ code isn't making everything very safe and you're getting a low-level C or C++ error from what you're asking the native library to do, through the thin wrapper layer.

In my case, I was getting an image from the SDK, and then asking the SDK to copy the raw data.  I made the mistake of getting the image dimensions once, at the start of the session, and just used those dimensions for every image I got back.  But really, the SDK was changing state on its own, and suddenly started returning images of a different size without me making a change (the 3DScan API does that on its own).  Which was unintuitive.  When that happened, neither the C or C++ layer, nor the thin wrapper layer checked the bounds and as a result, it just diligently tried to do what I asked, and threw a low level Access Violation when it tried to access parts of an image that were not there.  Which completely bombed out my IDE and was pretty much unable to be debugged in a normal manner.  I ended up debugging the IDE itself using GDB IIRC, which let me trace into the C or C++ layer enough that something clicked and I figured out what was at least throwing the violation enough to intuit the rest.

In another instance, I was getting "pure virtual function call" crashes because I was accidentally trying to release or dispose images twice, because two different levels of the C or C++ API were both trying to release the same data.  Again, neither the thin layer, nor the C or C++ API were being particularly "nice" about it.

I hope that experience helps somewhat.  I know it's not the concise and direct response you'd prefer.

[Rafael_S_]

Thanks for sharing, that's a good heads up. I'll try to see if it's images problem because I did the same about the sizes.

Regards