SharedMemory::addException() is not atomic

[duncan3dc]

If two forks are adding an exception at the same time, it's possible that they both read the current memory segment, then whichever thread writes second overwrites the first threads addition.

This isn't a major issue as at least an exception will always be reported when one has occurred, but if multiple occur we may miss the details of one (this is probably the cause of the occasional test failure we see on Travis

[danielspk]

Hi, do you have an example of this error?

I would like to analyze it! I am interested in using your project.

[duncan3dc]

What type of example are you looking for? Is the one in the current test suite not suitable?

[danielspk]

Yes, sorry I didn't see it