Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site philabs.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!gatech!ut-sally!seismo!cmcl2!philabs!brb
From: brb@philabs.UUCP (Brian Bertan)
Newsgroups: net.micro.pc,net.micro
Subject: Two inquiries : Object file transfer, and Cross "C" development.
Message-ID: <530@philabs.UUCP>
Date: Sat, 7-Dec-85 03:08:20 EST
Article-I.D.: philabs.530
Posted: Sat Dec  7 03:08:20 1985
Date-Received: Mon, 9-Dec-85 03:49:38 EST
Distribution: net
Organization: Philips Labs, Briarcliff Manor, NY
Lines: 38
Xref: watmath net.micro.pc:6131 net.micro:12997


I wish to transfer object files from an IBM-PC to a VAX.  I will discuss the
motivation shortly, but first a question:

  How can I reliably transfer MS-DOS object files to a VAX?

I have attempted the transfer through KERMIT Ver2.28.  The file transfer was 
successful, but when I transferred the object file back to the PC, I found
a comparison between the original and the transferred files failed.  The
files were of different length.  Also, the object file with the round trip
to the VAX would not link ( the original linked without error ).  This leads
me to believe that object file transfer under Kermit is not very reliable.

Does anyone have any comments or suggestions?

Motivation:

I plan to send the file from the VAX to an Intel Series III MDS running ISIS.
The object code is generated by the Mark Williams C compiler.  Mark Williams
has to object formats: proprietary, and MS-DOS.  The MS-DOS should be com-
patible with the ISIS format (IS THIS CORRECT?).  The application involves
no system calls, and no libraries are needed.  The target processor is an 
8086.  We have software which transfers from PC to VAX, and VAX to ISIS, but
no PC to ISIS.  Hence the transfer roundabout method is neccessary.

I know this seems to be an unusual method of development, but my only
alternative is to run PLM/86, which does not support 32 bit signed arith-
metic.  Disscussions with Intel and Mark Williams technical staff led me to
believe this scheme would work.  The Intel C compiler (which is written by
Mark Williams) is very expensive.  We already have the MS-DOS compiler in
house, with an updated version on the way.

Any help would be greatly appreciated.

                                                  Brian Bertan
						  Philips Labs
						  Briarcliff Manor, NY
{allegra | decvax | seismo}!philabs!brb           (914) 945-6309