Intel doesn't have Dual Core

[George Macdonald]

By the client application.


I suppose they could go so far as to try to create viruses, worms and
such to disrupt internet traffic in general, but I think that would be
too much even for the **AA organizations!

Just junk files pretending to be music with garbage content seems to work -
if it can be automatized by the new mechanism which triggered this
discussion, so much the better for them... and the worse for the rest of
us.

There is a certain degree of pollution in P2P networks in general,
that has been one of their on-going tactics for some time. This is
also the reason why some networks have kind of failed in favor of some
more robust networks. Here a lot of this pollution is filtered on the
server and superpeer side of things, normally hosted on rather
high-bandwidth links.

The **AA's pollution is being filtered now? Where do the signatures come
from? What makes them appear different from a valid music file to a
superpeer?

It would be a bit of an annoyance, but more a waste of resources for
them than for those downloading the files. Their garbage sources
could get filtered out on either the server or client end. Perhaps
it's fortunate for many P2P networks that they've already been forced
to build some such capabilities into their tools in order to prevent
viruses/worms and spammers from spreading too much crap.

Filtering at the client end still clogs the channels and there is the
question of a signature to differentiate the **AA generated pest from bona
fide content - at least with viruses there is a necessary consistency for
any given virus since it has to do something well defined by its perp -
just garbage is not so easy to categorize.

 

[Henry Nettles]

Are you running PCI NICs on that setup? I ran a very similar system,
except using a full-fledge Debian GNU/Linux install, on a Pentium 100
for several years and almost never saw the CPU usage exceed 10%, even
when I had several people sharing this network and running plenty of
P2P connections. The only time it ever got up there was when I needed
to throw an ISA NIC into the machine. Err, that and when I was trying
to recompile the kernel (though even then it didn't seem to drop
packets).

Nope, two 10/100 PCI NICs, the red interface is on a Realtek 8139, and
the green interface is on an Intel 82558.

 

[Tony Hill]

The **AA's pollution is being filtered now? Where do the signatures come
from? What makes them appear different from a valid music file to a
superpeer?

There are a variety of techniques, a couple things I've seen include
blocking certain IPs, filtering according to some tell-tale overhead
packets and users manually tagging files they know to be invalid.
None of these methods are perfect by any means, but they do seem to
reduce the pollution sufficiently that's rarely been an issue for me.
Mind you, most of the music I download if from independent artists
anyway (I can hear enough of the mainstream stuff on the radio to know
whether or not I want to buy those albums), so maybe that skews my
personal experience somewhat.

 

[Robert Myers]

On Sun, 24 Apr 2005 08:48:00 -0400, Robert Myers wrote:


I wasn't suggesting that one build other's IP into hardware as a way to
skirt the law, rather build enough IP into the silicon that giving away
the software is a smart move. If someone wants to support your hardware,
fine.

Ayup. That would be a smart move, and that's one reason to want to
put software into silicon. Even though you might want to open source
it, you probably wouldn't want to use GPL because of its stickiness.

Am I losing it, or are we all buying into George's implied assumption
that open source = GPL? I don't know enough about BSD... I'll guess
(and I'll be shot down if wrong) that there's enough tool-kit type
stuff for BSD that you should be able to write x-windowing
applications that will run on linux certainly using no more than the
LGPL c-library, and maybe not even that.

RM

 

[YKhan]

Nope, two 10/100 PCI NICs, the red interface is on a Realtek 8139, and
the green interface is on an Intel 82558.

Yup, pretty much syncs with what I've been observing for the last
little while. Network routing is not as much of a trivial task as
people here think, even though the job has been taken over almost
completely by cute little set-top boxes. Those set-tops are
underpowered for this job.

Yousuf Khan

 

[keith]

Ayup. That would be a smart move, and that's one reason to want to
put software into silicon. Even though you might want to open source
it, you probably wouldn't want to use GPL because of its stickiness.

As long as the family jewels are hidden behind the API, who cares abut
GPL? Open the silly kimono. Let the world develope your product.

Am I losing it, or are we all buying into George's implied assumption
that open source = GPL?

I'm certainly not buying it. I know for a fact that people are making
money here, on high-value products, even. It *can* be done. AIUI, only
some libraries are strictly controlled by GPL, and there are LGPL answers
for much of this. If one seperates the core math routines from the UI,
mush of any license issues can be circumvented.

No, I don't buy George's arguments. The Existance Theorem tells me
otherwise.

I don't know enough about BSD... I'll guess
(and I'll be shot down if wrong) that there's enough tool-kit type stuff
for BSD that you should be able to write x-windowing applications that
will run on linux certainly using no more than the LGPL c-library, and
maybe not even that.

So release the UI under GPL and keep the secrets, secret.