All programs run under the control of a IMS master task which handles all sorts
of errors and abends in a coordinated way and decides what action is appropriate
(retry, abort etc.). We do not want ANY software running below this master task
doing its own error handling. Programming languages are PL/1, ASSEMBLER and C,
but the distinction doesn't matter, because abends from all three languages are
caught by the same error handler. TSO is not part of the game.
We had some discussions with a software vendor who wanted to do signal handling
to catch some divisions by zero which he had not checked correctly, that is, he
did not test the divisor, but instead wanted to catch the
division-by-zero-signal. We did not allow him to do this, which lead to some
> On Mar 27, 2010, at 09:45, Bernd Oppolzer wrote:
>> At least at our site signal handling in C is not allowed.
> So if the programmer at the terminal presses the INT key,
> the program simply terminates with an uncaught signal?
> That's _so_ Twentieth Century! Are Assembler programs subject
> to the same rule?
> Similar reasoning would preclude TSO/ISPF's handling of the
> ATTN key. ...
>> Even third party software vendors have to follow these rules,
>> or else they are out of business.
> ... Third party, but not IBM itself? And what about the operator
> MODIFY and STOP commands? Do those cause signals?
> -- G.