[Lazarus] debugger error stack

[Martin Frb]

Ok that looks like 2 issues.

> I opened the debugmanager and put in the breakpoint...but my program
> compiled and ran!
The problem you have might be a race condition. Highly sensitive to any 
change in timing. Unfortunately any kind of debug/log will change timing.

Have you tried first building, then F9? Because if it is timing, then 
maybe that will work too. Of course it still leaves the question where 
the issue is caused. But that may be very hard to find. I am running out 
of ideas what to try.


The sigsegv might have been in gtk (not sure if the stack can be 
trusted). probably not the debugger but something else.
> The Call Stack window was unresponsive
Stange. You are sure it was the stack window of the outer IDE? It is 
easy to get them mixed up, and if the inner IDE was paused in the 
debugger of the outer, then it would be unresponsive.

Or do you mean, it refused to show more lines of the trace? That happen 
if the error is in libraries or kernel. (due to incorrect calls)

However, if there are issues / crashes in the IDE, then while debugging 
they may be displayed as debugger error. This is because the debugger 
code use ProcessMessages, and if the IDE fails in that, then that fails 
back to the debugger.

If this is gtk related (rather than gdb/debugger / not sure if the stack 
can be trusted...) and there is a chance that is the case, then that is 
out of my depth. I dont maintain the gtk related code, and have hardly 
ever looked at it.



  "cant find bounds"
This happens if you press F8/F7 or step-out while paused in the OS / gtk 
libs. the debugger does not know where those functions start or stop. So 
that is normal.


On 05/03/2015 18:15, JuuS wrote:
> Hi Martin,
>
> I'm going to do some more here and see what I can find, but I did a
> trace of sorts...
>
> I opened the debugmanager and put in the breakpoint...but my program
> compiled and ran!
>
> But, on closing the slave lazarus window though I got
> an error from the master lazarus...screenshots attached. The Call Stack
> window was unresponsive so I can't give any more information.
>
> Maybe this tells something? If so let me know, but I am going to look
> into this a bit more myself over the next few days.
>
>
>
> Julius
>