Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!mips!apple!mgchow
From: mgchow@Apple.COM (Mike Chow)
Newsgroups: comp.unix.aux
Subject: Re: Minor problems with ELM and A/UX
Message-ID: <51509@apple.Apple.COM>
Date: 14 Apr 91 20:17:36 GMT
References: <1991Apr11.131729.3514@panix.uucp> <1991Apr12.034901.7685@servalan.uucp> <23930@memqa.uucp>
Organization: Apple Computer Inc., Cupertino, CA
Lines: 27

In article <23930@memqa.uucp> qfhca81@memqa.uucp (Henry Melton) writes:
>> 
>> Just as a wild guess, perhaps ELM is getting confused about signal handling
>> because of the SysV semantics being enabled by default.  Try sticking a 
>> call to set42sig() somewhere in the main() function and see if that helps.
>> I've found that's helped other programs that die strangely of "Alarm Clock",
>> and other programs as well -- as I recall, Epoch didn't work worth a damn
>> until I put the set42sig() in...
>> --
>> Richard Todd	rmtodd@uokmax.ecn.uoknor.edu  rmtodd@chinet.chi.il.us
>> 	rmtodd@servalan.uucp
>> "Elvis has left Bettendorf!"
>
>Great info!  set42sig let me compile oneko.  Keep these tips coming.
>
>

I THINK the "-ZB" flag to the A/UX C compiler will implicitly set up BSD
signal semantics and BSD defines, so one might want to try this flag when
compiling BSD sources for A/UX.

My general rule of thumb is to use set42sig() when porting BSD or SunOS 
derived code to A/UX.  Telltale signs are sigvec stuff, use of select(),
and BSD signal types (eg, SIGWINCH).

Mike Chow
mgchow@apple.com