Re: Tech centralisation

From: Eugen Leitl (eugen@leitl.org)
Date: Sun Oct 20 2002 - 18:56:27 MDT


On Sun, 20 Oct 2002, Phil Osborn wrote:

> Mandrake? As in Linux? Apart from the fact that you

Mandrake as in dirty french pinguin.

> are in a small minority, as virtually everyone with
> whom I have ever discussed this who worked with both
> NT and Linux vastly preferred Linux, including system
> administrators for many of the BIG university or

Linux is not THE solution. Those who're comfortable with the alternative
OSses, and don't mind the consequences long term should stick to whatever
system they have.

I personally value stability and being in control far more than bells and
whistles.

> corporate networks, I don't recall that I specifically
> mentioned Linux as being what I would compare NT - or
> any MicroSloth product to. In fact, I use The GIMP
> and am starting to learn Blender <www.blender.org> on
> my Win2000 system at work.

Btw, a week ago or so Blender finally went open source. I see that they no
longer are slashdotted, and show the full page. Verrry nice.

> To me, it is utterly bizarre to see a reasonably fast
> Pentium with almost a Gig of RAM slow to a c r a w l,
> just because it's printing to a PostScript printer,

Yes, it wants me want to puke, too.

> or, worse yet, formatting a floppy. On my Amiga 1000,
> running at 7.14Mhz (clock - the processor ran at half
> that), with 1 Meg of RAM, in 1986, I ran over 30
> programs simultaneously once just as a test... THAT
> finally bogged the system down.

There were at least as many system tasks running on my A2000 in 1988 as
what ps aux | wc just told me. It's really nice to run full-featured
binaries which were few 100 bytes to a few kBytes large.
 
> But, on the Amiga, I could set the task priorities for
> virtually anything on the system. I could format

Amiga OS was noticeably more advanced than your average (some like it in a
pot, three decades old) old man OS. Frankly, both our software and our
hardware stinks. It vividly demonstrates that on the short run progress
can and actually does reverse. We obviously have to wait for the trend to
pull us out of this swamp.

> multiple floppies - or hard drives - while printing,
> while downloading, while rendering, and while entering
> new data, all at once. And, I could create macro
> programs piping and controlling applications thru
> interprocess languages like REXX. AND, I could add
> new hardware while the system was running. AND, I
> could redefine "logical devices" on the fly, which the
> system would treat just like partitions on a hard
> drive, so that I could add a fonts directory from a
> floppy or zip to the fonts directory ten levels deep
> on the hard drive and the system would see them as one
> device. Thus, saving countless hours of work
> relinking files on a "modern" MicroSloth idiot box.
>
> We should have been 15 years beyond that point by now.

Look, I've used a 12 (FORTH) MIPS box from 1985, or so, which had a 2
kByte multitasking OS, including the compiler. Judging from descriptions,
Lisp machines from 1983 were really, really nifty.

A number of the better alternative reality branches never happened.

Do I whine? Nope. I mostly keep my mouth shut, and use the best tools I
can find for whetever $$$s I have. I could do much better with a couple
more k$. Apart from clustering, which scales in a fine-grain fashion the
next step of improvement would cost an order of magnitude more, and offer
a much worse ROI. Anything beyond that would be ridiculously expensive
(100 M$ to several G$), and would involve custom hardware.



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:40 MST