22 Sep
2014
22 Sep
'14
5:55 p.m.
From: Al Kossow <aek at bitsavers.org>
You would obviously only talk to one device at a time. PC would be at
128.0.0.16 on
a
private un-routable interface (which is why it's
128.0.0.x).
128.0.0.0/24 isn't unroutable. It's allocated to a company called "Jump
Management SRL" out of Romania (and it's in the BGP tables). Unroutables
are 10/8, 172.16/12 & 192.168/16 (per RFC1918).
KJ
27 Sep
27 Sep
2:09 a.m.
On Mon, Sep 22, 2014 at 6:55 PM, Ken Seefried <seefriek at gmail.com> wrote:
From: Al Kossow <aek at bitsavers.org>
Presumably what was meant was 127.0.0.x, part of the 127.0.0.0/8 block
reserved by RFC 6890 and its predecessors for loopback. While one
could certainly argue that it's a non-compliant usage, many systems
have used various addresses in that block for things that are
deliberately intended to be unroutable.
The RFC 1918 ranges are intended for local use and are supposed to be
publicly unroutable, but are often locally routable. 127/8 shouldn't
even be locally routable, though stranger things have happened.
You would obviously only talk to one device at a
time. PC would be at
128.0.0.16 on a
private un-routable interface (which is why
it's 128.0.0.x).
128.0.0.0/24 isn't unroutable. It's allocated to a
company called "Jump
Management SRL" out of Romania (and it's in the BGP tables). Unroutables
are 10/8, 172.16/12 & 192.168/16 (per RFC1918). 
7:51 a.m.
On 9/27/14 2:09 AM, Eric Smith wrote:
Presumably what was meant was 127.0.0.x, part of the
127.0.0.0/8 block
reserved by RFC 6890 and its predecessors for loopback.
It was a typo. I have been working with some devices that have 128. hardwired
addresses.
It is also sad that this is the only thing that anyone said about my post.
I thought maybe someone might think UDP on a couple of ports might be a
good lightweight portable alternative to USB.
This will be my last post on this subject.
9:33 a.m.
It is also sad that this is the only thing that anyone
said about my post.
I thought maybe someone might think UDP on a couple of ports might be a
good lightweight portable alternative to USB.
I for one think that Ethernet with UDP is an excellent alternative to USB
for most, if not all, applications.
BLS
10:44 a.m.
On Sep 27, 2014, at 12:33 PM, Brian L. Stuart wrote:
Indeed, Stuart Cheshire (who may have already been working for Apple
at the time) made the same point a few years before USB became the
dominant standard. I wish that idea had gotten more traction.
- Dave
It is also sad
that this is the only thing that anyone said about my post.
I thought maybe someone might think UDP on a couple of ports might be a
good lightweight portable alternative to USB.
I for one think that Ethernet with UDP is an excellent alternative to USB
for most, if not all, applications. 
28 Sep
28 Sep
5:57 a.m.
On Sat, Sep 27, 2014 at 01:44:02PM -0400, David Riley wrote:
I don't. Consider the options:
a) A small plug, which provides 2.5W of power and has a simple 1.2Mb/s
protocol, which is more than adequate for keyboards and mice. Costs beer
money.
b) T-pieces, terminators, mini-hubs, crappy Realtek chipsets, and a separate
wall wart because there's no bus power. Costs hundreds of pounds.
Fast-forward two decades, and Ethernet *still* doesn't provide bus power by
default, requiring ugly PoE injectors or expensive enterprise switches. Worse,
it clocks the link even when idle, drawing a surprising amount of power. Fancy
a laptop with 10-20% less battery life? Just plug it into a gigabit network!
On Sep 27, 2014, at 12:33 PM, Brian L. Stuart wrote:
[...]
I for one
think that Ethernet with UDP is an excellent alternative to USB
for most, if not all, applications.
Indeed, Stuart Cheshire (who may have already
been working for Apple at the
time) made the same point a few years before USB became the dominant
standard. I wish that idea had gotten more traction. 
6:54 a.m.
I for one think that Ethernet with UDP is an excellent
alternative
to USB for most, if not all, applications.
Indeed, Stuart Cheshire [...] made the
same point a few years before
USB became the dominant standard. I wish that idea had gotten more
traction. Fast-forward two decades, and Ethernet *still*
doesn't provide bus
power by default, requiring ugly PoE injectors or expensive enterprise
switches. Worse, it clocks the link even when idle, drawing a
surprising amount of power.
All because it is still being designed for the use cases it was
originally designed for. (Broadly speaking, at least.) If it had won
over USB then it would have evolved towards good support for low-end
and/or low-power applications.
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
1 Oct
1 Oct
2:10 a.m.
On Sun, Sep 28, 2014 at 09:54:19AM -0400, Mouse wrote:
[...]
Both of those would be different now if Ethernet had
won over USB. If
Ethernet were used for the things USB is now used for, PoE would be
ubiquitous instead of niche, Ethernet would be dirt cheap because it's made
in trillions of units, and USB, if it still existed at all, would be
expensive because it would be a low-volume niche thing.
USB was *designed* to be an extremely low-cost interface exactly so it could be
high-volume. For Ethernet to have "won over" USB, it would have had to be
redesigned so that it was as naff as USB, at which point everybody loses.
8:44 a.m.
New subject: Ethernet and USB (was cctalk Digest, Vol 133, Issue 43)
Both of those
would be different now if Ethernet had won over USB. If
Ethernet were used for the things USB is now used for, PoE would be
ubiquitous instead of niche, Ethernet would be dirt cheap because it's made
in trillions of units, and USB, if it still existed at all, would be
expensive because it would be a low-volume niche\ thing.
USB was *designed* to be an extremely low-cost interface exactly so it could be
high-volume. For Ethernet to have "won over" USB, it
would have had to be
redesigned so that it was as naff as USB, at which point
everybody loses.
I'm not so sure about that. It appears to me that one could design
a lower cost physical interface for Ethernet that didn't require all
the magnetics by requiring that anything more than one-to-one had
to go through a switch instead of a hub or a bus. That would also
appear to simplify the MAC silicon. Now one could argue whether
that's really still Ethernet without CSMA/CD, and quite frankly, I
don't know that I care. What would make me happy about that is
that as far as controller and driver design go, the salient feature
is that we still have 802.3 frames on the wire. Then I can do
whatever I want with those frames, just like I can with Ethernet.
I'm not forced into doing something horribly complex to accomplish
something simple.
It's not the physical layer of USB that I object to (except of course
that the name doesn't match the physical characteristics). It's
everything in all the layers above the physical that makes me want
to wretch every time I find myself working with it.
BLS
9:47 a.m.
New subject: Ethernet and USB (was cctalk Digest, Vol 133, Issue 43)
On Oct 1, 2014, at 11:44 AM, Brian L. Stuart <blstuart at bellsouth.net> wrote:
That was certainly the intent, and I've certainly read worse standards
than USB, but IMHO, they failed to achieve what they intended. USB
ended up with a protocol that is much more complex and ugly than
Ethernet and required more complex silicon.
There certainly are few protocols as simple as Ethernet.
The essence of Ethernet is the frame formats. WiFi carries those over a completely
different physical layer, but since the same packets apply, 802.11 succeeded because of
Ethernet. Conversely, 802.4 and 802.5 failed in part by being different than Ethernet.
(Excessive and useless complexity is another component, of course.)
The various flavors of Ethernet are not all that similar. For each speed grade, the
physical layer and bit level encoding is completely different. Gigabit does away with
half duplex (for practical purposes) and 10G kills it completely ? but realistically, half
duplex has been obsolete for many years now.
A full duplex Ethernet MAC is a very simple beast. A matching PHY for modest speeds (100
Mb/s or so), a few meters, and common ground, is also very simple indeed. For example,
the magnetics come from the goal of supporting isolated grounds; if you don?t need that,
which a USB-like application doesn?t, then you can do without them just as USB does.
This would make a good EE project, come to think of it. Build an FPGA that uses the
Ethernet framing but with a PHY optimized for USB-like settings, and discuss what it would
cost in high volumes. Any EE 201 profs around here who would like to try that?
paul
Both of those would be different now if Ethernet had
won over USB. If
Ethernet were used for the things USB is now used for, PoE would be
ubiquitous instead of niche, Ethernet would be dirt cheap because it's made
in trillions of units, and USB, if it still existed at all, would be
expensive because it would be a low-volume niche\ thing.
USB was *designed* to be an extremely low-cost interface exactly so it could be
high-volume. For Ethernet to have "won over" USB, it
would have had to be
redesigned so that it was as naff as USB, at which point
everybody loses.
I'm not so sure about that. It appears to me that one could design
a lower cost physical interface for Ethernet that didn't require all
the magnetics by requiring that anything more than one-to-one had
to go through a switch instead of a hub or a bus. That would also
appear to simplify the MAC silicon. Now one could argue whether
that's really still Ethernet without CSMA/CD, and quite frankly, I
don't know that I care. What would make me happy about that is
that as far as controller and driver design go, the salient feature
is that we still have 802.3 frames on the wire. Then I can do
whatever I want with those frames, just like I can with Ethernet.
I'm not forced into doing something horribly complex to accomplish
something simple.
It's not the physical layer of USB that I object to (except of course
that the name doesn't match the physical characteristics). It's
everything in all the layers above the physical that makes me want
to wretch every time I find myself working with it.
BLS
3735
days inactive
3743
days old
9 comments
8 participants
participants (8)
-
abuse＠cabal.org.uk -
aek＠bitsavers.org -
blstuart＠bellsouth.net -
fraveydank＠gmail.com -
mouse＠Rodents-Montreal.ORG -
paulkoning＠comcast.net -
seefriek＠gmail.com -
spacewar＠gmail.com