Hmmm, a huge chunk of code, hundreds of lines say, replicated a time or two  in 
the same source, with subtle minor changes and compile-time conditioned with 
easily unnoticed uncommented lonely AIF's.  Maybe that's worse, but I'm easily 
chaffed by head fakes.  :-)

-----Original Message-----
From: IBM Mainframe Assembler List [mailto:[email protected]] On 
Behalf Of Steve Comstock
Sent: Wednesday, December 04, 2013 12:32 PM
To: [email protected]
Subject: Re: Base-less programming

On 12/4/2013 1:16 PM, Tony Thigpen wrote:
> I try to balance the use of B *+2+4 style branches against the
> cluttering up by using a bunch of labels. It's never for more than 2
> instructions.


> Nothing worse than looking at long series of tests followed by bit
> setting instructions where every other instruction has a label.

Really? Nothing? A startling lack of imagination there.

:-)


-Steve Comstock



>
> Tony Thigpen
>
> -----Original Message -----
>   From: Chris Craddock
>   Sent: 12/04/2013 02:58 PM
>>> Date: Wed, 4 Dec 2013 13:37:31 -0600
>>> From: [email protected]
>>> Subject: Re: Base-less programming
>>> To: [email protected]
>>>
>> <snip>> >
>>>> Am I reading the book right?
>>>>
>>>
>>> No. Although the assembled instructions have the displacement in
>>> half words, your source code should still use the original "offset".
>>> HLASM itself will halve the value. And it will complain bitterly if
>>> the offset is not even. So you just replace the B??? with J??? and
>>> leave the operand itself alone.
>>
>> Well.. yes, but being pedantic; how about just using a label?!? I
>> cringe whenever I see carefully crafted branch statements with instruction 
>> lengths.
>> That's trivially broken by any down-stream change, whereas a label as
>> a jump or branch target will never be wrong - no matter how much the
>> code changes in between. Make your own and the next poor fool's job easier.
>> Just saying....
>> CC
>>
>

Reply via email to