Newsgroups: comp.parallel.pvm
From: heimes@spva.ph.ic.ac.uk (Frank Heimes)
Subject: Re: PVM problem or network problem?
Organization: Imperial College of Science, Technology, and Medicine, London, England
Date: Tue, 22 Nov 94 14:28:30 GMT
Message-ID: <1994Nov22.142831.4358@cc.ic.ac.uk>

In article <jacobsCzG0nF.Los@netcom.com>, jacobs@netcom.com (Allan Jacobs) writes:
> wang@cteq11.msu.edu (Xiaoning Wang) writes:
> 
> >I installed pvm version 3 in my ~/pvm3 directory on cteq11 machine, everything
> >seems fine. but when I run pvm from cteq11 and try to add host, say cteq14, it
> >says : cteq14 Can't start pvmd. Then I tried run pvm from cteq14 and then add
> >host cteq11, it works fine.
> 
> I got a similar error today when I switched from PVM3.2.x to PVM3.3.5.  There
> is a bug somewhere in PVM3.3.5.  After some trial and error, I found
> a workaround.  Simply moving the assignment of PVM_ROOT in my .cshrc
> files to somewhere very near the top makes the problem go away.

Hi all,

I have a possible explanation for this circumstance. The .cshrc must not produce
any output in non-interactive sessions (like scripts or rsh that don't start an
xterm). So most .cshrc contain lines like

     if ( ! ( { tty -s } && $?prompt )) then
       exit
     endif

As variable and environment settings are needed for interactive sessions as well
as scripts, pvm's .cshrc.stub should be inserted ABOVE these lines, which abort
.cshrc in non-interactive sessions.

adios,
       Frank Heimes

