Path: utzoo!attcan!uunet!lll-winken!ames!pasteur!ucbvax!UICVM.UIC.EDU!U18189
From: U18189@UICVM.UIC.EDU (Michael Sperberg-McQueen)
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: Help?  (EBCDIC/ASCII translation)
Message-ID: <8901162016.aa28233@Louie.UDEL.EDU>
Date: 17 Jan 89 00:51:42 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 54

Perhaps someone in this discussion can help me, at least by directing
me to somewhere where I can get further information.

I am using IBM TCP/IP software on a PS/2 to telnet in to an IBM 3081
running VM/SP over an Ethernet.  To the VM system I appear as a 3270.

The data clearly is passing over an eight-bit connection, because all
the 190 printable code points of EBCDIC display as characters on my
screen, not just the 94 characters of the old national EBCDICs.  But the
translation is unlike any I have seen documented anywhere; above all, it
does not correspond, as far as I can tell, to the extended code pages
for EBCDIC (code page 37 or code page 500) or ASCII (ISO 8859-1).  We
would like to provide ISO 8859-1 / EBCDIC CP 37 support for our users,
so I would like to modify / customize / correct the translation.

After some searching with DEBUG, I found an EBCDIC-ASCII translation
table within TELNET.EXE (two, actually, which differ in some ways);
by zapping the table with DEBUG I was able to get the desired translation
between PS/2 extended ASCII (Code Page 850) and the U.S. Country Extended
Code Page for EBCDIC (I believe this is Code Page 37, but I could be
wrong).  This (a) confirms that the PC software does the E-A conversion,
and (b) seems a kind of awkward way of modifying the translate table.
The same table also seems to affect ASCII-EBCDIC mapping (the caret
sign changed its ASCII-to-EBCDIC mapping when I changed the table).

My questions:

1 Is there any kind of support now in place for customizing the
ASCII-EBCDIC translation?  (I'm thinking of utilities like CONVXLAT EXEC
on the VM side, or of the assembler macros for the Yale ASCII IUP for
the Series/1.)  Zapping the code with DEBUG is all right for me, but
it's not something I feel good about recommending to users.  Or even
to the guy who is supposed to support the PC-based Telnet software.

2 Can anyone tell me what code pages were used to define the
EBCDIC-ASCII translation for the "new" characters (ASCII characters
higher than hex 80, EBCDIC characters undefined in the old 94-character
code pages)?  (That is, what is this translation table *doing*?)

3 Will future releases of the product support Code Page 37 or Code Page
500 (or any other of the Country Extended Code Pages)?  We finally have
a chance to see ASCII-EBCDIC conversion become reversible and involve no
information loss -- it would be a real shame not to take advantage of
the opportunity.  At our site we finally managed, within the last year,
to support the extended character sets on our Series/1s and 7171s.
But we also need to support them on Ethernet connections.

Any help you can give would be greatly appreciated.  As I'm not a
subscriber to this list, please reply directly to U18189@UICVM
or U18189@UICVM.UIC.EDU -- many thanks.

-Michael Sperberg-McQueen
 Systems
 University of Illinois at Chicago