One small follow up:
 
After some persistence and some additional suggestions from other members
of this list I was able to eliminate the 'CIERR 900' error.  The trick was
to always have a reader and a writer against the message file when using
the 'input' command. So:
 
 ....
!  continue
!  input _text <*msg  >*msg
!  echo @ !hptimef : ![rtrim(_text)] :
 ...
 
did the trick.
 
Again thanks to ALL for your help
 
Regards
 
Paul H. Christidis
 
 
______________________________ Reply Separator _________________________________
Subject: Re: Command re-direction to message file
Author:  Larry Byler <[log in to unmask]> at CCGATE
Date:    6/19/1996 1:40 PM
 
 
Paul H. Christidis ([log in to unmask]) wrote:
: Thanks to all who replied to my question.  All of the replies suggested the
: use of 'GMULTI' on the file equation that references the message file.
 
: I had already tried that setting but it was causing the following error in
: the batch process:
 
: End of file on input. (CIERR 900).
 
: The 'workaround' that I've implemented is to ignore the 'cierror 900' and
: loop back and reread the message file.   I thought that the scheme should
: be 'cleaner'(no errors 900, or rereading the file) but it does accomplish
: the desired task.
 
: Thanks again.
 
: Regards
 
: Paul H. Christidis
 
I also was forced to use Paul's method (ignore the CIERR 900) in a test
suite I wrote four years ago.  I wanted to use message files as the
synchronization mechanism between multiple job streams, but kept running
into this problem.  There *was* an SR filed against this (1650151396)
back in 1991, but it was closed as a User Misunderstanding :-(.  I also
checked the offending code in Command I/O Redirection and found what
appeared to be an effective solution (and documented that in the SR),
but that's as far as it got.
 
-Larry "glad I don't have to do that again" Byler-