home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!sun-barr!news2me.EBay.Sun.COM!exodus.Eng.Sun.COM!appserv.Eng.Sun.COM!appserv!khb
- From: khb@chiba.Eng.Sun.COM (Keith Bierman-khb@chiba.eng.sun.com::SunPro)
- Newsgroups: comp.sys.sun.misc
- Subject: Re: Quick arithmetic exception warning question
- Date: 21 Jan 93 12:15:41
- Organization: SunPro
- Lines: 36
- Distribution: world
- Message-ID: <KHB.93Jan21121541@chiba.Eng.Sun.COM>
- References: <jeffb.727565349@aquifer.geology.uiuc.edu>
- NNTP-Posting-Host: chiba
- In-reply-to: jeffb@aquifer.geology.uiuc.edu's message of 20 Jan 93 21:29:09 GMT
-
-
- In article <jeffb.727565349@aquifer.geology.uiuc.edu> jeffb@aquifer.geology.uiuc.edu (Jeffrey Biesiadecki) writes:
-
- ... we have, but didn't find the answer. ieee_flags() works to clear..
- before exit points, but it would be much better to just turn off the trapping
-
- There is no trapping occuring. It is common to confuse the
- *retrospective* ("foo has occured") with trapping ("foo is occuring, now
- what do I do?"). You may wish to reread Chapter 5 of the Numerical
- Computation Guide.
-
- As you suggest, turning off the flags will eliminate the message (this
- is, in fact, mentioned on page 99 of the NCG as the socially approved
- way to turn off the message).
-
- Another approach is to replace the ieee_retrospective routine with one
- of your own; e.g.
-
- subroutine ieee_retrospective()
- return
- end
-
- will do nicely.
-
- However, whichever approach you select, you are losing potentially
- useful information and gaining *no* computational performance (you do
- save a couple of printf's at the end of your run, of course).
-
- cheers
-
- --
- ----------------------------------------------------------------
- Keith H. Bierman keith.bierman@Sun.COM| khb@chiba.Eng.Sun.COM
- SunPro 2550 Garcia MTV 12-40 | (415 336 2648) fax 964 0946
- Mountain View, CA 94043 <speaking for myself, not Sun*> Copyright 1993
-
-