On Mar 8, 2011, at 08:37, john gilmore wrote:

> ...  The HLASM can assemble code for AMODE(64) execution, but it can use only 
> 31-bit arithmetic in the course of doing so.  Would that things were 
> otherwise!
>
Fully aware that this would require enormous changes to code
and data structures and is unlikely to happen (barring a
requrement imposed by PL/S), yet I wonder:  If HLASM were
magically be transformed to use 63-bit evaluation throughout,
and truncate to 31 bits where context requires, what
incompatibilities would arise?  Of course some constructs
which now report "overflow" would quietly assemble.  I'll
ignore this.  And I can exhibit one bizarre case where
base-displacement resolution succeeds (producing arguably
incorrect results) and would fail with 63-bit evaluation.

> About this topic, I do not share Tony Harminc's aversion to the use of 
> assembly-time set symbols.
>
Is this traditional, antedating the SETC instruction, or is
there some basis in that lookahead processing for SETC differs
from that for EQU?

> There is a locally famous anecdote about Quine here in Cambridge.  He is said 
> to have waited to see if a speaker used 'data' in the singular or plural.  
> Hearing it used in the singular, he decamped, secure in the knowledge that 
> the speaker would have nothing interesting to say.
>
> Analogously, when I see a piece of assembly language that makes no use of the 
> macro language, I decamp, secure in the knowledge that it will be of 1) of 
> little interest and 2) gratuitously long-winded and detail-ridden.
>
But that was a different thread, best left to languish.

-- gil

Reply via email to