The late-model assembler is vastly superior to previous editions, but there's
always room for improvement.
In this particular case of MVC, the assembler, if it were so designed, could
assume the length of the source in
MVC Target(),=C'A literal'
This would be no different than
MVC Target(L'Source),Source
. . .
Source DC C'A literal'
except that the first is less cumbersome to code.
The first case currently causes an assembly error, so an "upgrade" is possible
without affecting existing programs.
[MVC 0(,R8),=C'A literal" does not cause an error, so this specification
could not changed without some other indications.]
The danger in either case is that there's no length checking between the Target
and Source. The code, in either case, can "run off the end of the page." Use
of labeled fields does not buy such protection.
Now perhaps the assembler could make some attempt to check the lengths (there's
an improvement), but avoiding literals does not alleviate that danger.
As applications move away from assembly language, one of the reasons cited is
the (perceived, at least) difficulty in writing assembly-language programs.
As good as the assembler is, this anxiety could be reduced if the assembler
were more "user friendly."
There is room for improvement.
Richard Zierdt
-----Original Message-----
From: IBM Mainframe Assembler List [mailto:[email protected]] On
Behalf Of [email protected]
Sent: Wednesday, May 16, 2012 12:37 PM
To: [email protected]
Subject: Re: OT? Assembler "enhancements"?
On Wed May 16 16:26:58 2012 [email protected] wrote:
> But most of all would be some way to imply the length of the second
> operand on SS format instructions. I cant begin to imagine how much
> time I have spent counting the length of some second operand and
> having to manually code its length in the first operand. This is time
> consuming and error prone and should be done by the Assembler.
Well the reason it's not is to stop people from trying to stuff 10 pounds of
data in a 5 pound bag. The length, if not specified explictly, has to be the
length of the receiving field. And this has its problems too when you run off
the end of the page or move garbage because the receiving field is shorter than
the sending field. Bottom line, you really should be s pecifying lengths
explicitly and it's very easy to do. And you shouldn't be using literals in
code.
What about something like this (sorry for any line wrap):
PRINTLINE DS CL133
SOME-FIELD DS 8
.
.
MVI PRINTLINE,X'40' SET FIRST CHAR BLANK
MVC PRINTLINE+1(L'PRINTLINE-1),PRINTLINE BLANK REMAINDER OF
PRINTLINE
MVC PRINTLINE(L'SOME-FIELD),SOME-FIELD MOVE THE RIGHT LENGTH!
As far as the original post goes, call me easy to please, but I have been
coding in assembler since MFT (and this is not bragging on this list because I
know guys older and more experienced than I hang out here) and I really don't
have anything that bothers me enough to file and RFE on the assembler. If
anything it has more than enough good stuff and very little bad.
