Peter Relson wrote:
<begin snippet>
If you do need to save/restore, then there are several choices of which
BAKR...PR is one, STMH...LMH is another, and having your caller provide a
144-byte savearea and using STMG...LMG is another
</end snippet>
Of these three the third, a 144-byte savearea and STMG/LMG is in my view the
one to use IN NEW CODE, i.e., in new routines being written ab initio that call
others. AMODE(31) is not yet dead, but it is moribund, and laying traps for
yourself or your successors in new code is bootless.
Predictably then, I do not find the argument from sloth for using the BAKR/PR
pair persuasive in general, although there are of course some situations in
which it is convenient or even, as a practical matter, necessary to do so.
That said, this thread was not the best place to say it. I have said before,
but not it would seem been heard to say, that the topics of doubleword
arithmetic and AMODE(64) are disjoint in fact if not in the minds of the
ignorant. Doubleword floating-point arithmetic was/is in fact available in
AMODE(24) routines. COBOL does doubleword integer arithmetic in compiled
AMODE(31) code. Etc., etc.
Why this confusion? I have looked, and I have found nothing in any IBM
document that panders to it.
John Gilmore Ashland, MA 01721-1817 USA
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
