NT daddy turns his hand to Xbox

Anyone remember RSX-11M?

Fairly early on I was writing device drivers for RSX-11M for medical imaging and lab automation. The RSX distro came with the complete source files to the OS - you had to recompile the whole thing to add a driver (!). Browsing through the OS sources and looking at the change logs was incredible - Cutler wrote practically the whole thing himself, in a very short period of time. And the code quality and real time design was even more incredible. It was so lean and interrupt-aware - I had data logging at 10kHz (we never missed a clock), with a lab full of people on terminals doing (very early) word processing. Cutler's legacy was clouded by the strange beast that VMS became, IMO.

A friend in another department ran UNIX on a much faster machine, but it couldn't even keep up with the line clock at 60Hz - the system clock drifted. In those days at least, the UNIX kernel pretty much ignored interrupts. So, the UNIX machine was a bust for real time lab work, but their productivity in other areas was a lot better than ours - while I wrote in MACRO-11 assembler and Fortran, they had this fancy new language called "C"...

Cutler's most elegant project?

In terms of technical elegance and pioneeringness, Cutler's eventual best DEC project was probably VAXELN, a distributed realtime kernel for VAX (with VMS as development host) which had features way ahead of its time. But it was a bit niche and predates the Internet so there's not much out there written about it [1]. Parts of its design are actually visible inside the original NT kernel - at least as visible as the alleged VMS heritage, in fact, if you understand the internals of both.

Thanks to the small number of really good people on the VAXELN development team post-Cutler, the user experience with VAXELN as it matured (eg V4.4) was much nicer than the early experience but the basic concepts underneath weren't greatly different.

"I wrote in MACRO-11 assembler and Fortran, they had this fancy new language called "C".."

Someone wasn't looking "outside the box"? The user group, DECUS, would have provided (for not much more than cost of media) stuff which would have moved your development productivity closer to the UNIX folks, maybe beyond.

For FORTRAN, e.g. there were extended versions of Kernighan's (as in K+R) RATFOR preprocessor. DEC's own FORTRAN IV and FORTRAN 77 each had their own advantages and disadvantages. For quiche eaters desiring Pascal, the list included Tanenbaum's Amsterdam Pascal for free, and at least two commercial Pascals (OMSI, UCSD).

There was C too. For free, there was at least DECUS C. There was at least one commercial offering from Whitesmiths (who decided that K+R's IO wasn't for them and nor was K+R's string-handling. Odd).

We shall not see those days again. Of course, sensible people will be quite happy never to see overlays again.

[1] A few paragraphs of user-contributed intro, with a link to DEC's own VAXELN Technical Summary:

http://unix.derkeiler.com/Newsgroups/comp.os.vms/2007-03/msg00014.html