Actually this seems to be an X IO error. X should never die on whatever the application does. See also https://bugzilla.redhat.com/show_bug.cgi?id=920548 And no, it's not the number of precedents because that is just one, the range. It may be related to the virtual offset where the highlighted range would be drawn, but this is speculated so far, I didn't manage yet to debug that properly. It doesn't even die within Calc but apparently in the Gtk main loop yield.