- Status Closed
- Percent Complete
- Task Type Bug Report
- Category QCAD (main)
-
Assigned To
Andrew - Operating System Ubuntu Linux
- Severity Medium
- Priority Very Low
- Reported Version 3 RC4
- Due in Version Undecided
-
Due Date
Undecided
- Votes
- Private
Attached to Project: QCAD Bugtracker
Opened by Stefan - 30.03.2012
Last edited by Andrew - 06.04.2012
Opened by Stefan - 30.03.2012
Last edited by Andrew - 06.04.2012
FS#530 - sporadic unwanted close of QCad
Using the move and scale tools, QCad sometimes (reason unknown so far) crashes in RC4 (never seen that in RC2).
Closed by Andrew
06.04.2012 14:54
Reason for closing: Fixed
Additional comments about closing:
06.04.2012 14:54
Reason for closing: Fixed
Additional comments about closing:
Most likely fixed by disabling debugger
(RC5)
on another machine (also running XUbuntu), copy+paste seems to work, but pressing ESC after successful insertion will instantly crash QCad RC4
same with line tool: sometimes pressing ESC multiple times will instantly crash RC4
This is really weird. The behaviour is in no way reproducible and sometimes occurs even on window-change (ctrl+tab). Normally, I would suspect my local RAM, but it occurs on two independent machines and only with RC4 (not with RC2).
Yes, I am aware of these problems. It's caused by ECMAScripts being interrupted by system events.
These system events (focus in, focus out, mouse click, key press, etc) might trigger C++ or ECMAScript code that for example deletes an object the first script was relying on which leads to crashes.
This is most likely to happen when something is being terminated (e.g. after pressing Escape for tool termination or Ctrl-W for closing).
I've discovered today that switching off the Qt debugger (Preferences - Scripting) at the very least improves (if not fixes) the situation here.
I'd be extremely thankful for feedback of people who can confirm / deny that observation on their systems.
Thanks!
40 minutes without this problem after disabling the debugger. I will report after more testing (will take some time).
Thanks for the report. Looks promising :)
It also starts to make some sense:
Scripts should normally not be interrupted by system events. But the debugger is notified whenever a function starts or ends (even if the debugger window is not shown). The debugger might allow for event processing during those notifications which I would consider to be a Qt bug.
I had some time to do some tests, no "unwanted close" with disabled debugger so far. I will report after more tests.
After several hours of testing, there was no unwanted close of QCad with disabled debugger. (should the de-bugger introduce such a bug?)