(j3.2006) Fortran 2008 "ALL STOP" and exit status code

[Bill Long]

Hi Tobias,


Tobias Burnus wrote:
> Hello,
>
>  From the Fortran 2008 draft standard:
>
> "Execution of a STOP statement initiates normal termination of 
> execution. Execution of an ALL STOP statement initiates error 
> termination of execution."
>
> NOTE 8.28: "If the stop-code is of type character or does not appear, or 
> if an END PROGRAM statement
> is executed, it is recommended that the value zero be supplied as the 
> process exit status, if the
> processor supports that concept."
>
> As "ALL STOP" initializes an "error termination" should one not then 
> recommend to return a non-zero status for stop codes of the type character?
>   

Maybe the word "error" is confusing here.  The intention here is that 
users would execute ALL STOP if they want to kill off all the images 
because continuation is pointless.  This could be because of bad input 
data, algorithm convergence failure, or getting a failed status back 
from an I/O statement or memory allocation.  These cases may or may not 
correspond to what one would expect to cause a non-zero exit status.  We 
should discuss your suggestion, but the issue may not be as simple as 
you indicate.  (I'd also note that your comment pertains to a Note, 
which is not part of the normative text of the standard.)

>
> Additionally, it is not fully clear to me whether STOP and ALL STOP stop 
> a single image or all images. "When an image is terminated by a STOP or 
> ALL STOP statement" implies that it only stops a single image; however, 
> "ALL STOP" implies by its name that all images are stopped.
>   
STOP halts the execution of just that image. ALL STOP halts the 
execution of its image and then causes the other images to halt 
execution as well.  You are getting the correct inference from the name 
ALL STOP.  The semantics here rely on the meanings of "normal 
termination" and "error termination" which are covered in 2.4.5 
"Execution sequence", paragraphs 5 and 6.  Pages 33-34 in 08-007.  In 
particular, see the text at the top of page 34.

Cheers,
Bill


> Tobias
> _______________________________________________
> J3 mailing list
> J3 at j3-fortran.org
> http://j3-fortran.org/mailman/listinfo/j3
>   

-- 
Bill Long                                   longb at cray.com
Fortran Technical Support    &              voice: 651-605-9024
Bioinformatics Software Development         fax:   651-605-9142
Cray Inc., 1340 Mendota Heights Rd., Mendota Heights, MN, 55120