Re: *nix

[Gene Heskett <gheskett@shentel.net>]

On Saturday 15 February 2020 16:03:02 ghe wrote:

> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
> On Friday, February 14, 2020 10:56 PM, Gene Heskett
>
> <gheskett@shentel.net> wrote:
> > On Friday 14 February 2020 22:56:11 Richard Owlett wrote:
> > > On 02/14/2020 12:52 PM, Gene Heskett wrote:
>
> FYI, fogies, in the Jul-Aug, 1978 Bellsystem Technical Journal,
> announcing Unix, in the Style section of the Foreward is a list of
> "maxims...gained currency among the builders and users..." The first
> sentence of the first maxim in the list is, "Make each program do one
> thing well."
>
> The second sentence is "To do a new job, build afresh rather than
> complicate old programs by adding new 'features.'"
>
> Until recently, the *nix communities have stuck pretty well to these
> recommendations -- they're just descriptions of competent programming,
> after all. There may be some discussion over the definitions of "one
> thing" and "well" but there is software in our Linux that, I think,
> doesn't conform to anybody's understanding of these maxims.

No, it hasn't been anything resembling recent,  Glenn. Remember that in 
1978, the 6809, z80 were in the future, and the then current crop of 
smart sand was things like the 8085, the cdp-1802, and possibly the 
ti9900 were state of the art, memory was still gawdauful expensive and 
Dynax was the king of disk drives spinning a 14" disk with the head 
positioning done by a very recalcitrantly driven voice coil about 4" in 
diameter Disk capacity of such a monster was 5 megabytes. The tv station 
I was the ACE at the time had a to9900 based mainframe with 2 or 3 
terminals in traffic had an uptime of maybe 12 hours a week because one 
of its 2 disk drives needed a recalibration that took 6 to 12 hours to 
do. The 9900 had a multitasking os that was also used in the ti-99/4 
home computer, an interesting software approach to how to do 
multitasking, not by stacking its registers but by switching its one 
memory pointer that pointed at its register images in memory, so an 
interrupt was handled by pointing the register at the memory occupied by 
the interrupt handler. All this wasn't unix but it outran the unix of 
the day by a fair margin. But it eventually lost that battle as unix got 
better at running multiple copies of cpu's tp gain thruput. My first 
intro to unix was the AT&T 3B2. A moto 68000 based this that was 
basically a fire starter looking for a place to burn up as it did so 
several times while serving as the first email server dedicated to 
receive messages from CBS about programming data. After the first smoke 
party we put it on a metal table. But in those days unix only ran on 
those moto cpu's.

Fast fwd from 78-84 to 2020 and our linux is running on the darndest 
collection of cpu and machine architecture, whose variations to run on 
whatever is in the box, variations that seem to be approaching the 
famous !69 test of good ti calculators. It is in some circles built from 
src and runs only on that motherboard, or its built to run on certain 
families of similar architectures, but every driver today has to have a 
lookup table to determine how to run on this variation. But in becoming 
the universal operating system also means the average install has a 
couple gigabytes in the boot files that never see an execution because 
its not built to run on his hardware. But today, gigabytes of memory are 
cheap, I just pair about half of what I paid for 4k of static ram I had 
to solder up on an s-100 card, that 4k of static ram cost $400 then but 
a couple months back I bought 32GIGs of memory forthis machine after a 
fire in a usb2 port on the mainboard put it out of business.  Just how 
effective is this run on what you find today?

This HD I'm booting from has an install from the old board with debian 9 
on it, but when this drive found it was not running on a 4 core phenom, 
it never even had to clear its throat to boot up on a new Asus 
motherboard with an 9th generation core i5 6 core cpu.

I think thats doing pretty good and I wouldn't change it w/o a fight.

Right now I've in the last week, I've built and installed a new 
preempt-rt kernel on that rpi4, needed to run any moving machinery, then 
git cloned linuxcnc-master from github, to that rpi4's added ssd's, set 
a couple environment vars, and half an hour later I have debs that 
install LinuxCNC on this rpi4 and its running a 75 yo 11x54 Sheldon 
lathe I converted to cnc, making it do tricks the original could not do 
in 75 years of trying.

And I could take both src trees to x86 hardware and build and run it just 
as sweet.

I have built 3 other machines that are doing exactly that. But why bother 
when I can get those from the buidbot.linuxcnc.org site. x86 hardware is 
available for peanuts at just about any yard sale.

Its progress and light years ahead of the situation in 1978 when I had to 
solder every part to make a dumb little computer run a u-matic tape deck 
to reduce the aired commercial by one dub level and improve our aired 
image a lot ANd install the cue tones to make stuffing 5 30 second 
commercials into a 2:30 network break, a 1 button operation with enough 
time left over to show a 1 second station id.  Yeah, I'm a retired 
broadcast engineer. And a CET.

> ​--
> Glenn English


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page <http://geneslinuxbox.net:6309/gene>