"Peter Smithson" <[log in to unmask]> wrote:
> In article <[log in to unmask]>, [log in to unmask]
> says...
>
> > I strongly discourage the use of the COBOL intrinsics (the old "CK"
> > procedures).  Do not use them unless you have a really good reason.
>
> Why?

Several reasons:

1.  The "CK" procedures were a temporary solution created back in the 1970s
to allow the use of KSAM/3000 from the old HP COBOL/3000 product, which did
not implement COBOL's "Indexed I-O" module.  That was before HP COBOL II
existed.  COBOL/3000 itself has been obsolete for many years.

2.  HP discourages the use of the "CK" procedures.  They are still
documented in Appendix A of the "Using KSAM XL" manual, but note this
wording:  "COBOL compilers (COBOL 68 and earlier) required special
intrinsics to access keyed files.  The following intrinsics are provided
only for the maintenance of COBOL 68 or earlier COBOL programs using KSAM
structures.  Note.  Do not use these intrinsics for new programming.
Current COBOL file access modules provide KSAM file access."

3.  Access to KSAM using the standard COBOL statements provided by HP COBOL
II/iX has been tested very thoroughly.  Before each release of MPE/iX, and
before the release of each power patch, the compiler and run time-libraries
are tested against the COBOL Compiler Validation Suite, as well as a
regression test suite maintained by HP.

4.  The COBOL II/iX compiler is still supported.  When there are changes in
MPE/iX that might affect COBOL, for example the recent introduction of
KSAM64, the compiler is tested against those changes and modified as
necessary.  (Yes, COBOL II/iX was updated to provide support for KSAM64
files.)

5.  If you avoid using extensions to the standard (like allowing duplicates
on the prime record key), accessing KSAM files using the standard COBOL
statements is highly portable.  The "CK" procedures are not.

6.  I have this general philosophy.  If there are two ways of doing
something, and one method has its syntax and semantics thoroughly specified
by an ANSI or ISO standard, and is highly portable, whereas the other method
is home-grown, less thoroughly specified, and non-portable, you should have
a good reason to choose the nonstandard approach.  In the case of the "CK"
procedures, a good reason would be if you have lots of ancient COBOL
programs that already use them, and you have no intention of porting those
programs to another platform.  Other than that case, I'm hard pressed to
think of a decent reason to use them.

Walter

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *