Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!sun-barr!apple!amdahl!karl
From: karl@uts.amdahl.com (Karl Rowley)
Newsgroups: comp.unix.cray
Subject: Re: Unformatted data transfer..
Message-ID: <c2GQ02Jod5fv01@amdahl.uts.amdahl.com>
Date: 18 Oct 90 18:04:33 GMT
References: <9171@helios.TAMU.EDU>
Reply-To: karl@amdahl.uts.amdahl.com (Karl Rowley)
Organization: Amdahl Corporation, Sunnyvale CA
Lines: 26

In article <9171@helios.TAMU.EDU> suray@cs.tamu.edu (Subhankar Ray) writes:
>
>Here is a test program, where I was generating a file on YMP, named 
>ibm( good for IBM) and then I moved it to IBM (under 
>MVS/os) using   binary ftp. But when I was trying to make an 
>unformatted read on IBM, it was
>giving end of record ( or end of file depending on the recfm and lrecl 
>modified by IEBGENER). Am I missing something ?
>

An unformatted file on the Cray contains other information besides the data.
Each record begins with a header that contains the length of the record.  This
is necessary to enforce the semantics of Fortran I/O.  I'll bet that the structure
of an unformatted file on your IBM machine is different from  that on the Cray, and
that this is the source of your problem.

If you can locate documentation of the structure of unformatted files on the two
machines, you could write a utility to do the translation.  Or, you could write
the data out in 'pure' binary form on the Cray by calling some C code, assuming 
that there is a way to then read the data on the IBM machine.

					Karl Rowley
					Amdahl Corp.
					Santa Clara, CA
					karl@uts.amdahl.com
					...!ames!amdahl!karl