Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!gatech!hao!murphy
From: murphy@hao.UCAR.EDU (Graham Murphy)
Newsgroups: comp.os.vms
Subject: Re: setjmp and longjmp in VAX-C
Message-ID: <680@hao.UCAR.EDU>
Date: Mon, 25-May-87 12:18:22 EDT
Article-I.D.: hao.680
Posted: Mon May 25 12:18:22 1987
Date-Received: Tue, 26-May-87 01:45:04 EDT
References: <121@acad.UUCP>
Reply-To: murphy@hao.UUCP (Graham Murphy)
Organization: High Altitude Obs./NCAR, Boulder CO
Lines: 31
Keywords: setjmp longjmp VAX-C
Summary: Beware automatic variables

In article <121@acad.UUCP> robert@acad.UUCP (Robert Wenig ext 609) writes:
>
>I have found that setjmp does not work correctly if the module in which
>the call was contained was compiled with OPTIMIZE on. Has anyone else
>encountered similar problems?
>
>
>Robert Wenig
>Autodesk

I assume by "does not work" you have found that some variables
don't retain their expected values when you return to the
routine. This is because the longjmp restores ALL the
registers, which means any variables that have been optimized
into these registers will take on the values they had when
the setjmp call was made. Can be tricky as it is not always
obvious when this will happen (except with hindsight.) 
    In all other respects I have had no problems with 
setjmp/longjmp having used it in many optimized routines.



Graham Murphy
High Altitude Observatory
National Center for Atmospheric Research

-- 
UUCP: {hplabs, seismo, nbires, noao}!hao!murphy
CSNET: murphy@ncar.csnet  ARPA: murphy%ncar@CSNET-RELAY.ARPA
INTERNET: murphy@hao.ucar.edu
NOT-NET: P.O. Box 3000, Boulder, CO 80307-3000.   (303) 497-1514.