(j3.2006) exception message on termination

[Robert Corbett]

On 06/12/13 18:23, Van Snyder wrote:
> Bill Long wrote:
I was thinking much simpler - the programmer just inserts the CALL before the 
STOP statement. An industrious programmer could, of course, write such a 
subroutine using the existing facilities.
>
> I think Bill means "before every STOP statement."  If the program stops
> for another reason, say deallocating a disassociated pointer, or
> computing the logarithm of zero, it's a bit hard to insert the call.
Unless I missed something (which is possible), the standard does not require the 
status of the exception flags to be written in those cases.

The standard allows the external unit identified by the constant ERROR_UNIT to 
be closed and reopened.  If the external unit identified by the constant 
ERROR_UNIT is disconnected or is connected to an unformatted file, is a 
processor still required to make the status of the exception flags available?  
Should it cause an I/O error condition?

Robert Corbett