Exit Sage ungracefully upon SIGQUIT

[jdemeyer]

When Sage is somehow totally stuck, one can still quit Sage using SIGQUIT (CTRL-\ on the keyboard). It would be useful if this would give a nice traceback. In other words: treat SIGQUIT like SIGSEGV, except that it should always exit Sage, never raise an exception.

Depends on #13908
Depends on #13996

CC: @nexttime @pjbruin

Component: c_lib

Author: Jeroen Demeyer

Reviewer: Peter Bruin

Merged: sage-5.13.beta2

Issue created by migration from https://trac.sagemath.org/ticket/14029

[jdemeyer]

Changed dependencies from #13908 to #13908, #13996

[jdemeyer]

Author: Jeroen Demeyer

[jdemeyer]

Attachment: 14029_SIGQUIT.patch.gz

[pjbruin]

comment:7

This looks like a useful feature to me. The patch looks good and behaves as expected, and all doctests pass.

The fact that sage-CSI is run does mean that if GDB is available, it now takes about 10 seconds before you are back at the shell. Pressing ^\ a second time does not immediately interrupt sage-CSI. I'm not sure why; is the ^\ somehow intercepted by GDB? Pressing ^C during these 10 seconds does generate a Python KeyboardInterrupt in sage-CSI, but does not immediately get you back to the shell either.

Anyway, the issues in the above paragraph are only slightly annoying and should probably be addressed elsewhere than in this ticket (if at all), so no reason against a positive review.

[pjbruin]

Reviewer: Peter Bruin

[jdemeyer]

comment:8

Replying to @pjbruin:

The fact that sage-CSI is run does mean that if GDB is available, it now takes about 10 seconds before you are back at the shell. Pressing ^\ a second time does not immediately interrupt sage-CSI. I'm not sure why; is the ^\ somehow intercepted by GDB? Pressing ^C during these 10 seconds does generate a Python KeyboardInterrupt in sage-CSI, but does not immediately get you back to the shell either.

That's right indeed. It seems that, once GDB gets hold of a process, there is no easy way to get out.

[jdemeyer]

Merged: sage-5.13.beta2