[Info-vax] VMS port to x86

[Paul Sture]

On Wed, 30 May 2012 04:46:54 -0700, Neil Rieck wrote:

> This is true of VMS-BASIC if you only consider the fact that it has
> built-in ISAM support. I can't remember if it was COBOL-74 or COBOL-85
> (or perhaps the industry discussions between them), but you could only
> call it "a standard COBOL implementation" if it had built-in ISAM
> support. So DEC followed IBM's lead and placed full ISAM support into
> VMS. (in the RSX-11/RT-11 days you needed to buy a license for the
> indexing piece of RMS). Once a full ISAM API was there, it quickly
> leaked into other products like VMS-BASIC.

VAX-COBOL circa 1982/1983 had standards related restriction on how you 
defined ISAM files.  One example was that no two keys could start at the 
same offset in a record (a pain if you wanted to access files created in 
another language without that restriction).

IIRC its predecessor was COBOL-11 and was COBOL-74 based.  VAX-COBOL 
introduced features based on what was expected to be included in 
COBOL-85.  This definitely made life easier for us.  COBOL-74 appeared to 
me to be hindered by the capabilities of lesser implementations, with the 
lowest common denominator approach enabling portability.

Yes, I just mentioned portability and COBOL in the same sentence. 
Portability is not really a new idea :-)

> Indexing was also done (to a lesser extent) in the UNIX world (at least
> the BSD world) for COBOL, but COBOL was never as popular as C; and ISAM
> on UNIX was never as popular as relational (this is where Oracle takes
> off). But lack of standard ISAM modules in UNIX can only be the
> explanation for why many C programers today are unable to wrap their
> brains around record processing (most always want to open a file with
> "fopen()" then produce stream i/o)

The Compaq COBOL documentation clearly defines the differences between VMS 
and Tru64 flavours of the compiler (an interesting read if you are into 
that kind of stuff).

> There are other aspects of VMS-BASIC which are truly unique to DEC/CPQ/
> HP. WHEN-ERROR blocks and MATRIX commands are the first which spring to
> mind.
> 
> As others have already pointed out, the main work in porting the so-
> called "DEC languages" will be done once in the GEM back end.
> 
> NSR
> 
> p.s. according to this link:
> http://h71000.www7.hp.com/openvms/30th/t_past_text.html#1980 VMS-BASIC
> first appears in 1980 and is called "BASIC Multikey ISAM".
>



-- 
Paul Sture