> That business of terminating the cable has always amused me.
> There is no reason to terminate a short cable, the reflections at
> 100MHz will be stuff-all under about 10-metres, and with that
> silly thin cable they use, the losses are so great that any
> reflection will be lost at anything *over* that!

RG> Amusing or not, when I set up my first 'network' (2 computers)
RG> I (foolishly) thought I'd be able to plug the coax (1metre)
RG> straight into the cards (No "T" adapters, not terminators) and
RG> needless to say it didn't work. 

 You have to terminate it at *one* end, or the system will be running
into an open-circuit generating 20V! But that's got nothing to do with
the cable.

 Cables have a characteristic impedance set by the relationship
between the inductance of the wires and the capacitance between. If
you send a pulse along a cable and leave the end open (badly
mismatched), then the pulse reflects, and if you then run along that
cable measuring the pulse, you will see spikes on the front and back
edges. The return pulse is delayed, and if the cable is long enough,
with small loss, you might even find points where you have two sets of
pulses.

 The delay along a 10-metre 50-ohm cable would be something like 5
nanoseconds, but so what, when there are only two computers, one at
each end? The delay is goign to exist anyway, and when it reaches
the 50-ohm termination at either end it will be absorbed anyway. A
short cable will have negligible delay.

 The other thing I mentioned was the length. A long lossy cable (thin
cable) ends up looking like 50-ohms anyway. There are reflections, but
they are lost on the way back. For that thin cable, 10-metres will
be as good as infinity.

RG> It took me much experimenting before I discovered I needed the
RG> "T" at BOTH ends, and that both ends needed to be terminated
RG> before it would work. 

 Rubbish. On a one-metre cable, it wouldn't matter a rat's arse
*where* you terminated it, all the way up to a gigahertz.

> I know... and it really is fast!

RG> Not as fast as the lastest 1gig cards :-)

 It can't be any faster than the PC bus...

> but the weird part is that Windows uses 169.254.0.0 as their LAN
> and Linux uses 192.168.0.0!

RG> See my previous response to this question.

 Taht was a statement; not a question. Let me ask a question... why
does Windows use the 168.254.x.x and Linux 192.168.x.x when *neither*
are standard?

> It is easy to understand why Gates owns 99% of the O/S market
> when Linux is the main competition.

RG> You've twisted the facts a little Bob. Windoze has 99% of the
RG> *desktop* market. The *nix variations account for 80% or more
RG> of the *server* market. 

 You mean the market for 2-billion PCs against 500,000 servers?
That's not a market... it's a roadside stall.

> ROFL! You're kidding, aren't you? Routing tables and there are
> only two computers! ROFL!! Good one, Rod.

RG> I assume by this thoughtless remark that you are under the
RG> mistaken impression that routing tables aren't required if only
RG> two machines are connected? If so, once again you are showing
RG> your ignorance in regards to networking.

 ROFL! I can't believe you'd write a routing table for two
machines! Ignorant... or stupid?  

> That's the sort of unhelful crap I've been handling all week,

RG> OK, so it turned out to be useless information in your
RG> instance. You were having problems, you asked for help, and I
RG> gave you several possibilites. I wasn't expecting you to have
RG> made EVERY mistake a person could possibly make, but at least I
RG> tried to cover all the bases you ungreatful ingrate. 

 Not only can't you spell "ungrateful" but you repeat yourself. An
ingrate *is* ungrateful. Why not "ungracious ingrate"? That's much
more insulting and a bit poetic, too.

RG> For YOUR *future* reference though, had you enabled DHCP on
RG> your windoze box, which would have assigned a 169.x.x.x number,
RG> and then used the general default settings on your Linux box,
RG> and got a 172.x.x.x number, the two machines would have no
RG> connectivity unless you either assigned a default route, or a
RG> specific route for either of the addresses.

 Dur... now tell me something I *don't* know. In fact, my Linux uses
192.168.x.x.

 But in fact, my response to you was when you raised the furfy of
DNS...

RG> Oh, don't forget DNS when performing ping tests. Unless you
RG> have a working/accessable DNS you'll need to use the IP numbers
RG> when pinging because if you use host names you'll probably get
RG> a 'host not found' error.

 You do *not* need DNS to be able to use host names instead of IP
addresses on a network of two! I agre that the two IPs have to be on
the same page for a local net, but once having set them, you can
forget them, DNS or not.

RG> For YOUR *future* reference though, had you enabled DHCP on
RG> your windoze box, which would have assigned a 169.x.x.x number,
RG> and then used the general default settings on your Linux box,
RG> and got a 172.x.x.x number, the two machines would have no
RG> connectivity unless you either assigned a default route, or a
RG> specific route for either of the addresses.

 I took the easier option, and put them on the same page.

RG> No need to treat me as the idiot because you managed to avoid a
RG> mistake that many others have made in the past.

 Don't you like being treated like an idiot, Rod? I don't either,
which is why I have been right up your arse all month. If you talk
down to me and treat me like a dill, you'll get it right back and I
have to say, I do it better. I actually enjoy it!

RG> Screw you.

 Now you're getting close to a proper AVT response.

Regards,
Bob

--- BQWK Alpha 0.5