I thought again about your mention of “Timers waiting for such Threads to reach NotRunning States”. Are you polling the Thread state until State turns Not running ? Not sure if it would solve the error, but would it not be lighter to simply trigger a timer from within the thread from the very end of the Run code, so it basically fires when the code has finished executing ?
Maybe but it’s still an issue to be solved 
Never said otherwise 
Hopefully the Xojo team will find what is wrong in the project you attached to the bug report.
But i will try if this may be a workaround in the meanwhile. Thank yoi Michel.
It’s just this bad feeling that something is wrong with the new Xojo, something that may hit my users because i did not see it in my tests.
I did not know that a Thread is alowed to access a Timer which is an Object of a Window. Thank you for your advise, because this makes things a bit easier (and often faster). 
And your advised change makes a difference. But now the App is stopping at other “random” positions with StackOverflowError’s.
It’s an object, not a control. So it has no UI. Hence no issue being instantiated, called or manipulated within a thread. Very convenient IMO.
[quote=222119:@Sascha S]I did not know that a Thread is alowed to access a Timer which is an Object of a Window. Thank you for your advise, because this makes things a bit easier (and often faster).
And your advised change makes a difference. But now the App is stopping at other “random” positions with StackOverflowError’s.[/quote]
Glad it helped.
Since the issue happens when you access SQL, I wonder also if the SOE would not be coming from the fact that the 64 bit code is too fast for the mySQLCommunityServer. That would account for the seemingly random errors, because collisions happen depending on the speed of data transfer. Now how would one time the requests to avoid collisions is another matter I am not versed enough in to risk any suggestion.
Since the issue does not present in OSX, I would tend to put the blame on the Windows framework, and am sure Xojo will find where it happens.
It all happens in 32Bit.
Sorry. Then it cannot be synchronization.
Could it be that a user is twice or more times logged in and connected to the Mysql server? I have seen several times now that my compiled webapp crashes when both connections try to get data simultanious. (just a shot in the dark)
No, this is impossible (for various reasons).
And it all worked flawlessly in all Xojo Version in the last 2.5 Years. Something changed in Windows Builds which is causing these issues now.
But that no one else encountered these issues, still makes me wonder if i make something wrong…