| |
PA2AGA > TCPDIG 26.11.96 18:19l 182 Lines 7135 Bytes #-10808 (0) @ EU
BID : TCP_96_247C
Read: DG7DAH GUEST
Subj: TCP-Group Digest 96/247C
Path: DB0AAB<DB0PV<DB0MAK<DB0BOX<DB0FP<DB0DNI<DB0OCA<DB0FC<DB0HOL<DB0SHG<
DB0OBK<DB0CL<DB0NDR<PI8AWT<PI8JYL<PI8WFL<PI8MBQ<PI8VNW
Sent: 961125/2231 45648@PI8VNW.#ZH2.NLD.EU
From: PA2AGA@PI8VNW.#ZH2.NLD.EU
To : TCPDIG@EU
Received: from pa2aga by pi1hvh with SMTP
id AA21364 ; Mon, 25 Nov 96 19:49:21 UTC
Received: from pa2aga by pa2aga (NET/Mac 2.3.62/7.1) with SMTP
id AA00000092 ; Mon, 25 Nov 96 19:44:43 MET
Received: from pa2aga-1 by pa2aga with SMTP
id AA00000027 ; Mon, 25 Nov 96 19:31:01 MET
Received: from pa2aga-1 by pa2aga-1 (NET/Mac 2.3.62/7.5.5) with SMTP
id AA00008582 ; Mon, 25 Nov 96 19:30:59 MET
Date: Mon, 25 Nov 96 19:21:06 MET
Message-Id: <tcp_96_247C>
From: pa2aga
To: tcp_broadcast@pa2aga-1
Subject: TCP-Group Digest 96/247C
X-BBS-Msg-Type: B
via The ARRL Letter and QST.
Earlier this year, company chairman Mike Lamb, N7ML, took over day-to-day
control of AEA, following the hasty departure of former president Rod
Proctor, KI7ZI. Lamb had hopes of turning the company around and had
introduced several new products on the Amateur Radio market, including the
long-awaited DSP-232 multimode TNC, a 6-meter halo antenna, and a dedicated
9600-baud packet transceiver. In recent weeks, Lamb has made no secret of
the fact that he has been seeking a buyer or new investors for AEA; over the
past two months, the company has laid off most of its employees. While the
company no longer handles customer service requests, Lamb said this week
that AEA still is taking orders and even continues to manufacture some
items. AEA products also continue to be available from some dealers. The
company's three product lines include antennas, test (analyst) equipment,
and data products/software.
* * *
Material from The ARRL Letter may be reproduced in whole or in part, in any
form, including photoreproduction and electronic databanks, provided that
credit is given to The ARRL Letter and The American Radio Relay League.
___
- Origin: N1BEE BBS +1 401 944 8498 V.34/V.FC/V.32bis/HST16.8 (1:323/107)
------------------------------
Date: Sat, 23 Nov 1996 19:23:02 -0600
From: "Steve Sampson" <ssampson@oklahoma.net>
Subject: AEA up for sale; customer service closed
> Advanced Electronic Applications Inc--better known as AEA--is on the
market
> and is no longer handling customer service requests, as callers to the
> company's Lynnwood, Washington, headquarters discovered this week.
People won't spend $1000 for a modem. Even if it does contain a DSP chip.
AEA lost it's market share many years ago, when it distributed its products
to protected zones. You had to order by the mail to some 6 state protected
zone store. You couldn't see one in your local store.
With Ham stores failing and going under every week now, there is no place
to wholesale to.
Once we get down to two super distributers, I guess the prices will go back
up,
and then Ham radio will be a mail-order thing. Right now it's no fun to
work
all day for $2 markup on a radio, and have all the real money go to Japan
:-)
Steve
------------------------------
Date: Tue, 19 Nov 1996 08:23:57 +0000 (GMT)
From: Alan Cox <alan@cymru.net>
Subject: File loss
> like the filesystem should always actively schedule writes, albeit at
> a low priority to give them a chance to be updated by further task
> writes to the same blocks before the actual disk write takes place.
It can be very counterintutive but its nowdays held best to write the data
first then the metadata (control information). That means your file system
fixing program can see that there are errors rather than building valid
file systems with wrong data.
There are file systems that write in a computed order of dependencies. Thats
not easy to do as you can get loops in the dependencies and also not useful
in the USA because as is traditional everything in the USA (this included)
is subject to bogus software patents.
Now shall we go back to amateur radio and tcp/ip ;)
Alan
------------------------------
Date: Mon, 18 Nov 1996 22:09:45 -0800 (PST)
From: Phil Karn <karn@qualcomm.com>
Subject: File loss
>They do write back over a couple of seconds time, but if you lose power
>at an annoying moment you lose data. And we all know what 2 seconds the
>power will go off dont we.
Well, that's not too bad. You'd expect to lose *something* if power
goes off while the disk is active. More important in that case is that
the file system is left in a reasonably consistent state so further damage
won't occur when the system is restarted.l
What you *can* avoid is losing data due to modified stuff sitting in
cache while the disk sits idle waiting for a timeout to come along and
flush it out. Or worse, no timeout, which is what would happen in
generic UNIX without an update task running. It always seemed to me
like the filesystem should always actively schedule writes, albeit at
a low priority to give them a chance to be updated by further task
writes to the same blocks before the actual disk write takes place.
Phil
------------------------------
Date: Fri, 22 Nov 1996 17:33:47 GMT
From: brian@nothing.ucsd.edu (Brian Kantor)
Subject: Gateway from changing IP
>I'm trying to understand why traffic has to go through mirrorshades...
Historically, because of limitations of the early routing protocols,
there could only be one point of attachment between networks. If the
AMPRNet (network 44) is viewed as a monolithic network [which in the
absence of classless routing, is the only way it can be viewed], there
can only be the one connection point, and that is the system known as
'mirrorshades.ucsd.edu' or 'ampr.org'.
We divide network 44 up into smaller bits. It has recently become
possible to advertise (i.e., send routing info) to much of the internet
that a particular bit is reachable via a specific gateway.
For example, it would be possible to use non-Class-A routing to
advertise that instead of all of 44/8 going to San Diego, the piece of
net 44 that serves Los Angeles (44.16/18) is to be sent to the Los
Angeles gateway system, currently w6vio.jpl.nasa.gov (128.149.37.26).
And the piece for San Diego goes to our gateway. And the Bahamas,
theirs.
The problem is that this doesn't work everywhere yet. It's especially
difficult when we have lots of varying subnets - nearly a thousand of
them right now - and more to come as more gateways get set up. And
they change DAILY.
So what we're doing is continuing for a while to use encapsulated
routing (aka "tunnelling") where packets destined for any piece of
network 44 are routed to 44.0.0.1, and then encapsulated and shipped
onward to the appropriate gateway.
People can bypass this, of course, simply by installing "shortcut"
routes in their local region of the net. They just shouldn't advertise
those shortcut routes to the internet "core" to avoid confusing
things.
>> The technology exists now to make mirrorshades either obsolete,
>> or just a transition system for legacy systems. The technology
>> has existed for years. The core internet routers support classless
To be continued in digest: tcp_96_247D
sts now to make mirrorshades either obsolete,
>> or just a transition system for legacy systems. The technology
>> has existed for years. The core internet routers support classless
To be continued in digest: tcp_96_247D
Read previous mail | Read next mail
| |
