Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!ames!sgi!shinobu!odin!wanda!bowen
From: bowen@wanda.SGI.COM (Jerre Bowen)
Newsgroups: comp.sys.sgi
Subject: Re: exit signal
Message-ID: <1990Dec21.012822.502@odin.corp.sgi.com>
Date: 21 Dec 90 01:28:22 GMT
References: <13238@milton.u.washington.edu> <9012160645.AA10200@karron.med.nyu.edu> <1990Dec18.153942.15916@odin.corp.sgi.com>
Sender: news@odin.corp.sgi.com (Net News)
Reply-To: bowen@wanda.SGI.COM (Jerre Bowen)
Organization: Silicon Graphics Inc.
Lines: 45


In article <13238@milton.u.washington.edu>,
merritt@milton.u.washington.edu (Ethan Merritt) writes:
> In article <1990Dec18.153942.15916@odin.corp.sgi.com>
jmb@patton.wpd.sgi.com (Jim Barton) writes:
> >In article <9012160645.AA10200@karron.med.nyu.edu>, Dan
> >Karron@UCBVAX.BERKELEY.EDU writes:
> > 
> >> How can I set a signal handler for a software exit() call ? I am
> >> looking to catch programs that exit via an exit(value) call, or fall
> >> off the end of the main procedure. I do not have control of the user code.
> >
> >Your description is a little unclear of what you want - I don't know
> >what you mean by not having control of the user code. Guessing, the
> >easiest way to do this is to use dbx. Ask dbx to stop on a call to exit,
> >or even trap the exit system call. dbx will also catch any signals
> >posted to the program, so you can always catch the program when it is
> >going south.
> 
> 	I'm not entirely sure what Karron wants in his original posting, but
> this reminds me of something which caused me much grief when I started
porting
> code to IRIX.  Nowhere that I can find in the various libraries is an
> equivalent to the ANSI C atexit() routine, which allows you to establish
> a line of exit handlers which get to clean up before the program exits, no
> matter where that exit is triggered from.  As Karron mentions, SIGTERM looks
> like it might be of some help, but in fact it isn't (too many exit condidions
> do not set SIGTERM).
> 
> 	Any chance of getting atexit() in the next release?
> 
> 					    Ethan A Merritt
>
-------------------------------------------------------------------------------
> Dept of Biological Structure                H510 Health Sciences
> University of Washington SM-20              (206)543-8865
> Seattle, WA 98195                           merritt@u.washington.edu
>
-------------------------------------------------------------------------------

	atexit() is available now as a separate product, released (I believe)
a month or so ago.  It is a routine in the ANSI libc; the package also includes
an ANSI-compliant compiler.  Check it out.

		Jerre Bowen (bowen@sgi.com)