[Shameless Plug] A New Book

It was known as roff for short. NROFF and TROFF were patterned after it. The macros were dot so much different as not yet standardized. I had several NROFF .ms files for different printers, even one for an early HP inkjet.

Jerry

No kidding? For some reason, I had always thought it was the other way around -- probably just because I used Unix nroff before I used DEC runoff.

Yes. I used a VAX cluster during the 80s and it was my tool of choice on that machine until we obtained Scribe. TeX/LaTeX was also available on it, but I didn't get into it until I bought my first PC in 1988. I remember Nelson Blackman (GTE Government Sytsems) using it for his monthly scientist report.

By the way, I absolutely loved VMS.

I've never used a VAX cluster, but I know a story about one. Since we're totally off topic from an off-topic post, I figure I can digress some more -- right?

I took a multiprocessor class in the late 80's. The class project was to write a program to be benchmarked and run on a single processor, then a multiprocessor system. It was easy the year I did it, because the campus had a brand new Encore computer system with as many National Semi

At any rate, you write your little puzzle solving program on one processor and benchmark the results. Then you obtain access to a multiprocessor system, tweak your code, and benchmark your results again. In theory, with N processors the final answer should come out in

The instructor kept two records. At the time that I took the class the best speedup was something like 99.9% of theoretical, with a hand-built two-processor 8086 system with all dual-port RAM. The most slowdown was on a 4-processor VAX cluster, which would solve the puzzle on a single processor in about 40 minutes. When the hapless programmer tried it with all four, the plug got pulled after 24 hours.

It seems that on a VAX cluster any interprocessor communications that overflows the (small) buffer goes to disk...

It could be a tuning parameter that was mis-tuned. It seems that these DEC systems were very tunable.

I think ours were mainly used for load balancing. It was pretty amazing considering a cluster of 4 vaxes was serving 100+ simultaneous users at times - in ~mid-80s. Try doing that now, even with a 3+ GHz Pentium IV processor.

Shall I also tell about our PDP 11/70s and RSX11-M?

^^^

Jerry,

Was that deliberate, given that most of the macros began with a period, e.g. '.p' for a paragraph?

Knuth has always offered a bounty of $2.56 for errors found in his books. But I think most people frame the check rather than cash it. I know I would...

Regards, -=Dave

Jerry,

I always remember Xerex as being a type of antifreeze.

Clay

No, dot deliberate at all. I habe a code.

Jerry

So it was! For trademark purposes, is Xerex the same as XereX? The 'xer' for "dry" is necessary. Otherwise is could have been xorox. Oh, well!

Jerry

As a matter of principle, please preserve the attributions for material you quote.

... snip ...

I am not overly impressed. In the same time period we were servicing something in the range of 30 to 40 terminals with a single HP3000, not to mention several i/o processes fed by and to my embedded machines. I must admit most of the users were hooked into a single data base process, and as a rule no more than 4 to 6 were doing program development and other CPU intensive things.

The HP3000 is/was a 16 bit stack oriented machine. Todays Pentia are spending all their time drawing filthy pictures :-) Feed them a decent text oriented OS and reasonable i/o ports, especially ones that don't create an interrupt per character, and they should do very well.

I think that was Zerex.

(snipped)

Hi Jim,

interesting. I'm gonna guess that the typical new textbook has 40-100 errors that should be corrected.

So you're talkin' a fair amount of money there.

EBCDIC, BAUDOT, HOLLERITH or ASCII

I seem to remember something about setting the buffer length differently and getting significantly faster results -- but that makes the story much less interesting.

That was a good course -- the guy ping-ponged between starting companies and teaching, so his course was solidly grounded in reality. He had _lots_ of stories about wide-eyed technical innocence running smack into hard engineering realities.

Why leave out RT-11?

And since this is far OT -- any TECO fans [that almost came as fanatics] out there. In early 70's I worked in Maynard [ML5-5 to be exact]. Came in one Saturday to finish up a 2 page report using TECO on a KL-10 system that during week handled major portion of DEC's production control. Misplaced a ";" IIRCC thus copying file to itself ;[

Got thrown off system by upper level supervisor program that someone with my permissions should not have even new knew existed. However many scratch disks there were, I attempted to fill them all. A systems expert acquaitence told be I had managed to skip 2-3 levels of protection. Gee, wonder why DEC tried vainly to discourage use of that program -- you could do just about anything in it.

But shouldn't that come out of the publisher's cut? After all, what are editors for? ;]

Yes, and a google search on Xerex turns up some marginally interesting links, none of which is antifreeze. ;-) Zerex is, in fact, antifreeze.

Cheers! Rich