When a code has painted itself into a corner, or discovered an unexpected catastrophic error, it is useful to the developer if the code provides a back trace. Although it’s probably a source of confusion for the user, at least the user can send it to the developer. One can get a back trace in a debugger such as gdb, but not after the code has stopped . Otherwise, one must know where the problem was detected or reported, to plant a break in the debugger. Most of us have some kind of CRASH_BURN routine that causes a floating-point exception or rewinds an invalid unit or… . Univac FORTRAN V used “RETURN 0” to cause a back trace.
It would be helpful if there were a standard method for the code to ask for a back trace. The obvious place would be on the ERROR STOP statement. If there’s reluctance to provide a standard method, at least add a note that processors ought to provide a compiler option to cause a backtrace when an ERROR STOP statement is executed.