[extropy-chat] when will computers improve?

[Eugen Leitl]

On Sun, Dec 21, 2003 at 05:59:08AM +1000, Alejandro Dubrovsky wrote:

> Yes, any bad driver can crash a monolithic OS's kawasaki.  Got new

Yeah, this is one of the reasons monolithic systems suck so badly.
Context switches blahblah, you don't design an OS for a given architecture, and
Fiasco shows you can do much better even on x86.

There's zero evolutionary pressure for micro/nanokernels in a desktopcentric
universe.

> hardware, expect problems (and if you got old bits too, crap hardware
> doesn't help).  But it doesn't tend to stay in that state for long. Does
> it make OS design a trillion times easier though?  I'm sure it does. 

I'm not sure there's much to design in a kernel few kBytes code large.
Everything else is userland.

> Look at the state of all desktop microkernels.  

There aren't any. I keep looking, apart from the fringe there's nothing
there.

> On my side, i haven't had a crash in about a year, and that was playing
> with probably (not provably) broken hardware (or at least win2000 on
> another machine didn't like it either). (btw, this doesn't mean my
> uptime is a year. I reboot on average every couple of months after a
> power outage, or to upgrade the kernel)

The server is running fine, it's the desktop that's not so stable
(Radeon-driver-related crashes in some OpenGL screen blankers, rare). Doesn't
matter, I shut it down for the night for noise reasons.

> (ah, yes, this is vanilla 2.4.${MAX(x)} linux, with tainted NVidia
> drivers).  I don't miss those wasted couple of minutes terribly much. 
> Writing this email probably took more time than all of those reboots
> combined.

I'd rather have infinite uptime on a mission-critical.
 
> > I'm disgusted with the engineering aspect of all modern consumer-grade
> > systems, whether hardware or software.
> > 
> From the point of view of a not-yet-calcified member of the younger
> generation (which missed the awesomness of the 70s VMS, and the

I've seen VMS run, but it is massively unsexy.

> everything-has-been-done-in lisp machines) it is going quite well.  DOS
> 4.0 -> linux-2.4.23 and TRS-80 -> duron 1gig have been up all the way
> (as well as every single other aspect of either consumer hardware or
> software (graphics card, network speed, storage capacity, monitor
> size/quality, content viewers, content writers, content organisers,
> content distributors, content availability, programming language
> compatibility with brain (not by much, i admit), interfaces (mice,
> scanners, sound)). I expect (naively perhaps) for these trends to
> continue.

Things are not bad. They could do be a whole lot better, though, and it's this
unborn future I'm mourning.

-- Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07078, 11.61144            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
http://moleculardevices.org         http://nanomachines.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20031229/1cce8bd1/attachment.bin>