[comp.lang.fortran] IEEE exception handling

fsset@bach.lerc.nasa.gov (Scott E. Townsend) (11/29/90)

[ sorry if this has been hashed around before, I haven't needed to deal with
  FORTRAN or numeric exceptions for quite a while ]

I don't want to start any religious wars, but I would like to know if there is
a standard way of configuring IEEE exception handling in FORTRAN.  I've looked
at Silicon Graphics 'handle_sigfpes' and Sun's sigfpe/ieee_handler stuff.
These are similar in function, but quite different at the user level.

Does manipulating the exception semantics really need to look so different?
I realize that the various flavors of UNIX tend to have different capabilities
regarding handling things like SIGBUS, but I had hoped the the numeric stuff
could be a bit more portable than I've seen so far.

Any enlightement would be appreciated.

--
------------------------------------------------------------------------
Scott Townsend               |   Phone: 216-433-8101
NASA Lewis Research Center   |   Mail Stop: 5-11
Cleveland, Ohio  44135       |   Email: fsset@bach.lerc.nasa.gov
------------------------------------------------------------------------

khb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) (11/29/90)

In article <1990Nov28.224835.17336@eagle.lerc.nasa.gov> fsset@bach.lerc.nasa.gov (Scott E. Townsend) writes:

...
   Does manipulating the exception semantics really need to look so different?
   I realize that the various flavors of UNIX tend to have different
   capabilities regarding handling things like SIGBUS, but I had hoped
   the the numeric stuff could be a bit more portable than I've seen so far.

It could have been, but

	*)  IEEE 754 chose to ignore language binding issues
	    (they had enough problems getting done anyway;> )

	*)  language groups (such as X3J3) have resisted putting in
	    ieee 7|854 stuff out of deference to those nifty Cray's, 
	    IBM 3090's, VAXen, etc. 

	*)  etc.

Thus everyone has had to do the best they can. This is very
unfortunate, it inhibits clever use of the facilities of 7|854, and
adds to the difficulty in porting software, etc.

--
----------------------------------------------------------------
Keith H. Bierman    kbierman@Eng.Sun.COM | khb@chiba.Eng.Sun.COM
SMI 2550 Garcia 12-33			 | (415 336 2648)   
    Mountain View, CA 94043