[Info-vax] VMS on Raspberry Pi 5
Martin Gregorie
martin at mydomain.invalid
Tue Nov 14 18:58:44 EST 2023
On Tue, 14 Nov 2023 15:24:59 -0500, Arne Vajhøj wrote:
X> $ purge foobar.txt
>
Yes, I remember those periodic purges.
I did like the customisable editor too and, IIRD, the ability to attach
customisations to file types. If there was an open source version of that
editor I might even be using it still.
>> I was just a database developer on the VAX system, so had no
>> knowledge of much about it apart from what I needed to write COBOL and
>> interface that with DEC's relational database system.
>
>
> VMS + Cobol + Rdb was a common combo - it is till used today.
>
Worked well too - one of the better COBOL implementations
.
> Did you use embedded SQL or Rdb module feature?
>
The Rdb module.
The one real thumbs-down feature of the project, and the one that sank it,
was the bloody awful James Martin Associates system design method and its
almost equally poor diagramming rules. I mean, how can a usable Data
Structure Diagram ever be built when the overall system design rules
forbid you from adding any data items to entities but still let you insert
relationships between entities? Straight away this means you have:
(a) no way of verifying that the relationship is needed
(b) no way of validating end-to-end data flows through the system DSD.
IOW there's no way that you can verify the DSD, let alone get it reviewed
and signed off.
...and meanwhile we, the database design team, were never able to build
built a schema because the system designers never did hand us a usable DSD
to work from.
That surprised me because I'd bought a copy of James Martin's "Design of
Man-Computer Dialogs" - IMO one of the best books on application system
design I've ever seen. I bought my copy about ten years prior to that
fiasco and used it extensively when a small team of us were building BBC
Radio Three's music planning system: This was a complex system that had to
track and record every stage of building and broadcasting a concert from
the producer's original idea, checking that it wasn't too similar to any
other concert that would be broadcast around the target date, tracking and
booking all the performers and then, after the broadcast, making sure they
were paid as well as handling the repeat fees paid to performers if any
parts of a concert are rebroadcast later.
I doubt that this system, known as Orpheus, would have gone together half
so well without the ideas in that book,
For the record, it was built on an ICL 2966 mainframe, written in COBOL
and using IDMSX, a Codasyl database, as its data store. A BBC internal
team, with me as an external member, had completed that that about ten
years before the fiasco on the VAX.
IIRC there were even JMA consultants on the team that failed complete the
financial system on the VAX, but despite their presence, We never did get
a completed data structure to work from before the project collapsed.
--
Martin | martin at
Gregorie | gregorie dot org
More information about the Info-vax
mailing list