RE: how to create unsigned hexadecimal initial value

["Richard Jackson" <richardjackson@xxxxxxxxxxxxxxxxxx>]

Bob:

So, the problem was in trying to initialize an unsigned value with a signed
value and it isn't a bug when that doesn't work.

Feeling my way somewhat cautiously here, then the code is working as
designed but the error message didn't completely explain the problem that
the PRM/translator was having with the initialization value.  What do you
think of that statement?

For what it's worth, until someone else pointed up the H" situation, I was
seeing X".  My comment that the behavior was a bug was based on seeing the
X" string.  It probably wouldn't have made any difference - I probably would
have assumed that X" and H" had the same semantics - but that specification
difference is not intuitively different.  Wheels within wheels :)

Thanks for helping with this.  Do you have a machine readable MI manual?

Richard Jackson
mailto:richardjackson@richardjackson.net
http://www.richardjacksonltd.com
Voice: 1 (303) 808-8058
Fax:   1 (303) 663-4325

-----Original Message-----
From: owner-mi400@midrange.com [mailto:owner-mi400@midrange.com]On
Behalf Of rjd@us.ibm.com
Sent: Thursday, August 24, 2000 7:36 AM
To: MI400@midrange.com
Subject: Re: how to create unsigned hexadecimal initial value


On 08/24/2000 at 12:32:29 AM, "Peter Dow" <pdow@earthlink.net> wrote:

     What's the difference between H"80000000" in Gene's initial post and
X'80000000' that everyone else used? Is there some significance to using H
and " instead of X and '? Are they supposed to be equivalent?


H'80000000' is a numeric literal.  More specifically, it has the attributes
of a signed integer.

X'80000000' is a character literal.  It has the attributes of a character
type.

Given the following declaration of FLAG,

  DCL DD FLAG AUTO BIN(4) UNSGND;

we've observed several behaviors which can be explained in terms of the
integer vs character attributes of the literal values:

DCL DD FLAG AUTO BIN(4) UNSGND INIT(H"80000000");
  -- fails at compile time because the assembler recognizes that the
initial value is negative (recall that H-literals are signed).

DCL DD FLAG BIN(4) UNSGND AUTO INIT(X'80000000');
  -- works because the initialization of automatic data is implemented
using "copy bytes"-like instructions, which ignore numeric atrributes of
the data.

CPYNV FLAG, H"80000000";
  -- fails at runtime with binary overflow (mch1210).  The exception occurs
when the signed integer value (-2147483648) is converted to unsigned in
preparation for the store to FLAG.

CPYNV FLAG, X'80000000';
  -- fails at compile time because operand 2 has character attributes
instead of numeric attributes

CPYBLA FLAG, X'80000000';
  -- works because CPYBLA ignores the numeric attributes of the receiver

CPYNV FLAG, 2147483648;
  -- works because an unsigned literal is specified

DCL DD FLAG AUTO BIN(4) UNSGND INIT(2147483648);
  -- works because an unsigned literal is specified


Bob Donovan
+---
| This is the MI Programmers Mailing List!
| To submit a new message, send your mail to MI400@midrange.com.
| To subscribe to this list send email to MI400-SUB@midrange.com.
| To unsubscribe from this list send email to MI400-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: dr2@cssas400.com
+---

+---
| This is the MI Programmers Mailing List!
| To submit a new message, send your mail to MI400@midrange.com.
| To subscribe to this list send email to MI400-SUB@midrange.com.
| To unsubscribe from this list send email to MI400-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: dr2@cssas400.com
+---