Re: J4 - presentation/discussion on "Future of the COBOL Standard"
- From: Clark F Morris <cfmpublic@xxxxxxxxxxxxxxx>
- Date: Tue, 11 Mar 2008 19:08:11 -0300
On Tue, 11 Mar 2008 02:08:27 -0000, tim <TimJ@xxxxxxxxxxxx> wrote:
On Mon, 10 Mar 2008 09:20:46 -0600, Howard Brazee wrote:
On Mon, 10 Mar 2008 07:16:52 -0000, tim <TimJ@xxxxxxxxxxxx> wrote:
COBOL is not a good candidate for running on JVMs and .NET due to
technical problems ie use of redefine and pointers (which are in effect
arbitrary redefines), also packed decimal which is not well supported off
the mainframe. You can make it work, but it will run slowly.
Why does being able to use packed decimal make CoBOL a poor candidate?
If it doesn't fit in your environment, there's no requirement that it
has to be used.
If the program you are compiling says packed-decimal the standard says it
must be stored in packed decimal format. Of course you can provide
optimization flags that allow this rule to be overridden in the name
of speed. But with redefines and pointers/linkage it gets very difficult to
maintain correctness in the generated code if you change the format from
packed decimal to binary for example.
Packed decimal is mostly about the same speed as binary on IBM mainframes,
but it's a lot slower than binary on most Unix/PC CPUs. Packed decimal is
a lot faster than display format on mainframes for arithmetic, due to lack
of microcode support for most arithmetic operations performed on display
format data.
The only true decimal arithmetic on the IBM 360/370/390/z systems is
packed decimal and (IEE754 something on the latest z). It is more
efficient than converting to binary, doing the arithmetic and
converting back. It also is more efficient than what I think is
available for decimal arithmetic on the Intel chips. It is not nearly
as efficient as binary on the z series (or predecessors) if you are
not trying to get decimal rounding and behavior in division and other
places where it matters. Incidentally COBOL is NOT one of the
languages for which support for the new floating point decimal
arithmetic is announce on the z series.
Clark Morris
Clark Morris
.
Tim
- Follow-Ups:
- References:
- Prev by Date: Ping: Bill Klein on proposed cobol standards
- Next by Date: Re: J4 - presentation/discussion on "Future of the COBOL Standard"
- Previous by thread: Re: J4 - presentation/discussion on "Future of the COBOL Standard"
- Next by thread: Re: J4 - presentation/discussion on "Future of the COBOL Standard"
- Index(es):
Relevant Pages
|
