[Info-vax] Learning VMS application programming

Bob Gezelter gezelter at rlgsc.com
Sun Sep 7 14:56:18 EDT 2014


On Sunday, September 7, 2014 2:42:40 PM UTC-4, Bugs Bunny wrote:
> On 2014-09-07, Paul Sture <nospam at sture.ch> wrote:
> >
> > Believe it or not in the early 90s I read an article about a COBOL
> > compiler written in COBOL, for PCs IIRC.  They saw it as a means of
> > concentrating on that skill and making sure it worked under duress.
> > I have no idea now what it was called or whether it was successful.
> 
> It was probably Realia or a previous or post incarnation of it. IIRC it was
> written by the same guy that nearly single-handedly wrote SPITBOL and the
> GNAT Ada compiler. He's done a lot of other magic too. All of it is scary
> good.
> 
> >> Even back when I was using FORTRAN for system monitoring applications
> >> in my school days, I knew enough even then to realise I would _never_
> >> have dreamed of writing kernel mode code using it.
> 
> Yes and today despite Fortran being better than ever it still doesn't
> qualify for systems programming. And I don't believe it ever will.

"Bugs",

That "guy" would have been Robert Dewar. Robert was the Chair of the NYU's Computer Science department for some of my undergraduate years (he was also supervising my group when I implemented the PDP-11 code generator for LITTLE, which was designated a "machine independent systems programming language".

The term "systems programming" is somewhat ambiguous. Hoff's distinction of "inner-mode" is more explicit. "Systems programming" is less precise. In the case of LITTLE, the idea was not to write "inner-mode" code, but a portable way of implementing compilers and run-time libraries, a goal which we effectively realized.

During that period (late 1970's), we wrote a good deal of systems software in LITTLE, although all of it ran in user mode. We wrote compilers, run-time libraries, text formatters, remote job entry engines, and operating system simulators (and to answer the obvious question, the LITTLE compiler and run-time libraries were written in LITTLE, with a very small amount of host-dependent assembler for low-level interfaces).

- Bob Gezelter, http://www.rlgsc.com



More information about the Info-vax mailing list