Re: Welcome to 'tool' confusion

From: Eugene.Leitl@lrz.uni-muenchen.de
Date: Mon Apr 23 2001 - 15:01:26 PDT


Jeff Bone wrote:

> If all you need is a shell and / or X, and you've "got pipe," then fine. IMO,

Plain ssh is a bit spartanic, but brings you surprisingly far.
ssh/X combo means 100% of GUI in *nix. (Either that, or it's VNC).
Either way round, if you don't see the latency, you're home. Kindly don't
barf on the floormat.

> this isn't the problem I was musing about and therefore your solution above is
> not the right solution. Why? What happens when you need to access (at least

If 100% of *nix GUI is not the answer, what is then the question?
Totally emphatic holistically pangalactic YlemGui? Which clean
blows off the top of your skull and splatters the brains at the
ceiling, instead of a meek core dump?

> some part of) your "computing environment" and you have no IP flow?

If you don't have any (as in: zip; zilch) packet flux, then you're
limited to 1) a big fat honking nonvolatile storage unit (CDR is not enough,
USB drive? Where art the drivers? Got admin?) in a totalitarian 100%
totally compliant "we're in full control of your gonads" fashistoid
environment 2) portable environment (either Javur, or Winders. Both
major poopie). 3) Wearable (haulable, luggable). Best solution, which
I've -- duh -- mentioned.
 
> Disconnected use *is* a feature of the world we live in, and is likely to remain
> so indefinitely; even as wireless connectivity becomes ubiquitous and faster,

I don't think so. I already consider the state of being disconnected
as deeply pathological, and I haven't even gotten tetherless yet. Soon
I'll snort them packets, if I can't get a clean needle push. Vatch *DIS*

> the wired connection will continue to be faster still, meaning that some

Faster, yes.

> asymmetry remains between tethered and untethered. Total disconnection is an

I think the functionality is asymptotic. A little flatrate connectivity brings
you a long way already. Fat pipes are nice, but saturate when you can beam
3d telepresence in realtime, which is largely latency, not bandwidth. A lot
of this is tighter coding, which is a DSP problem. Portable DSP is largely
a power problem, which is in the process of being solved by fuel cells.

> edge case of this asymmetry.
>
> The bottom line is that solutions which assume connectivity will in the long term
> fail to be truly portable computing environments. We need a broader base of

I can't parse this, but presume you mean connectivity is not a solution. I disagree.
Since the altenatives stink so overpoweringly, connectivity, poor solution that it is,
is the only panaceiaform one.

> technology in areas like replication, optimistic pre-fetch, configuration
> management, etc. In the past, I've referred to this kind of thing as "dispersed"
> rather than "distributed" computing.
>
> Just some more random musing...

I see where you're coming from, but I consider the problem set unsolvable.
What you're talking about is full control of other people's hardware, Always.
Sorry, ain't gonna see it happen, unless it's Redmond, Redmond uber alles.
(Heil Gates; hello Godwin!).



This archive was generated by hypermail 2b29 : Sun Apr 29 2001 - 20:26:06 PDT