Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-22 Thread Timm Murray

 But why wouldn't a node just stop allowing connections when it sensed it
 was being used to much? Limit bandwidth and CPU usage within a given time
 frame and maybe watch for DoS type issues by only letting a certain number
 of connections per host per timeframe through? Then you know enough for a
 node to keep itself from being overloaded.

There already is bandwidth limiting.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-22 Thread Michael

 There already is bandwidth limiting.

*nods* I thought there was but haven't gone back to look at any such
details. Does it allow limiting incoming and outgoing bandwidth each on
their own?


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-21 Thread Timm Murray

  Centralization.  While it's not so bad if a better machine does a
  little more work, it get's bad when that same machine is getting
  requests from half the network.
 
 Wouldn't that be more a problem of routing requests than saying it is
 horrible if you want your node to have more data? I can see why
 centralization is bad but to be realistic there are just certain places
 that put out or take in a lot more information than others. Freedom
 shouldn't mean forced equality I guess. Couldn't the protocol just notice
 if there was centralization going on and assuming you had enough nodes in
 the network as a whole rebalance things somewhat?

If I understand what you're saying, no it can't.  Freenet is designed so that you 
can't know what is going on in other parts of the network.  A single node might 
notice that a certian other node tends to pop up a lot in it's table, and then go 
out and find a new node with similar data, but I don't think you could do this network-
wide and still maintain anonyimity.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-20 Thread Timm Murray

 Timm Murray:
  Over time, the large node simply accumulates more data from Freenet.  This means 
  there should be more nodes which point to data on the large node. Thus, there will 
  be more requests routed to the large node.
 
 Uhh... so? The node's big; it can handle lots of requests. That's
 not a problem. What's a problem is if Freenet says: Hey, your node
 is 10% better than usual, so let's send every request to it! I
 don't see why this would happen.

What about the guy who has a 40 GB hard drive, devotes half of it to Freenet, but only 
has a 56Kinda modem?  I'm pretty sure this *will* happen, probably already has.  If 
the average node is around 200 MB (which is what was suggested by Oskar and Scott at 
some point, IIRC), and you have a 20 GB node, your node is (1024*20)/200 = 102.4 times 
better then average, quite a bit more then the 10% you cite.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-20 Thread Timm Murray

Now you've got it :)

Yes, you're right that artificaly limiting the size of the store is a bad idea 
(zero-one-
infinity rule), but that doesn't stop us from advocating that people keep their store 
size down.

Setting the datasource diffrently when overloaded is an intresting idea.  I'll have to 
twist 
my brain around it more.

 Mark J Roberts:
  Timm Murray:
   Over time, the large node simply accumulates more data from Freenet.  This means 
   there should be more nodes which point to data on the large node. Thus, there 
will 
   be more requests routed to the large node.
  
  Uhh... so? The node's big; it can handle lots of requests. That's
  not a problem. What's a problem is if Freenet says: Hey, your node
  is 10% better than usual, so let's send every request to it! I
  don't see why this would happen.
 
 Looks like I totally, embarassingly missed your point the first time
 around... I'll try this again. :}
 
 You're right: it's true that the bigger your store, the more
 requests your node will receive. And overloaded nodes are bad
 because they make the network unreliable. So our objective is
 obviously to prevent overload.
 
 Well, I really detest artificially constricting the store size in
 order to regulate request load. But I _do_ understand your point
 now: larger stores demand more bandwidth. Low-bandwidth nodes need
 some way to avoid overload, and constricting the size of the store
 is one way to do it. The _wrong_ way to do it
 
 What's creepy about this are the various heuristics proposed to
 accomplish it: don't make a large store!, if you have Y
 bandwidth, use a X megabyte store!, etc. It's impossible to find an
 acceptable one. Which means that nodes will have to detect overload
 and adjust themselves. Which also means that we can stop promoting
 this dangerous small-store idea - the recommended size of the store
 should be based on the memory required to index its contents.
 
 One solution might be to, when overloaded, set the datasource not to
 yourself but to where you would've routed the request if you didn't
 have the data. You'd really actually use a probability scaled by the
 current throughput.
 
 Needs more processing.
 
 ___
 Chat mailing list
 [EMAIL PROTECTED]
 http://lists.freenetproject.org/mailman/listinfo/chat
 


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-20 Thread Timm Murray

   I haven't seen _any_ compelling argument why above-average nodes
   should attract more than their fair share of requests. What's
   yours?
  
  Over time, the large node simply accumulates more data from Freenet.  
  This means there should be more nodes which point to data on the large
  node. Thus, there will be more requests routed to the large node.
 
 Does FreeNet not take into account reliability, 

Not for routing an individual request, no.  However, nodes that don't respond when 
requested will be flushed from the datastore.

proximity,

No.

speed,

No.

 etc of
 nodes when fetching files? If it does I would think it could automaticlly
 sense if a node was becoming saturated and find a more satisfactory link
 and/or clone the resource?
 
 Why is it a bad thing if machines with more resources do more work?

Centralization.  While it's not so bad if a better machine does a little more work, 
it get's bad when that same machine is getting requests from half the network.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Michael

 Not useless.
 
 Dangerous.

And how would it be dangerous if it were a feature you had to use by
choice? I'm not saying change the default behavior. I'm saying add an
option when you've rounded out the more vital parts of the program. Not
that it matters I suppose. If you don't add it then someone else can
always add it for themselves.

Rather than waste 100's of gigs of space caching data I've already written
to disk and wish to keep available to other apps on my disk (not to
mention a second copy of all data mirrored to another drive for backup)
it's a perfectly reasonable thing to have as an option. Unless you're
planning to write a FreeNet filesystem so that other apps can work with
said files without having to be made FreeNet aware. Even then there are
benefits to keeping the original on my disk in the form I originally
placed them in.

I am going to have a copy of the data stored on the disk anyway for other
apps to work with.. so using links is no more dangerous than having copies
but is lots nicer on my diskspace. Anything I downloaded from FreeNet I'd
still want stored in the normal FreeNet cache. Anything I decided to
insert as a copy rather than a link would still be in the cache. Anything
I decided to insert as a link would still be stored in the cache of
whatever nodes I was connected to. The link only has value on the local
machine.


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Michael

 Implementing scores of kinda-somewhat-maybe-useful features is a
 great way to destroy any application.

One feature that should be extremely simple to implement and has a long
history of use is going to destroy FreeNet? It shouldn't even effect the
protocol at all. Surely you're being over dramatic. I'm not asking you to
add a talking paper clip. Certainly having no features leaves you with no
application at all.

 And the transparent store thing is _clearly_ useless.

Maybe if you never want to have more than a couple gigs of files on your
network. If you want massive numbers of files available then it's a very
useful feature. Not everyone is willing to clone every single file they
have just for FreeNet. Certainly not people with any large number of files
they'd like to insert.

I'm certain I'd be rather annoyed if my webserver wanted me to cache all
my files in a certain manner that makes them inaccessible to other
programs. Or how about my file server? Damn all this time I've been making
my files available on my network via web, scp, ftp, NFS, Samba, and
Appletalk. Now I'll have to make 6 different copies of every file. It just
wouldn't make sense and it doesn't make sense for FreeNet either.

I understand the need for secure caches but it's an entirely different
issue in this case. I have no need to hide the fact that I'm making
these files available. It'd be fully evident to anyone interested because
there'd be fully visible copies still on the machine.


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Timm Murray

 Timm Murray ([EMAIL PROTECTED]) wrote:
 
 
  Sorry Mark, but the one computer, one node rule is something I
  picked up from Oskar (I think), and I decided I agreed with him (and
  not just because he's Oskar).  Freenet works better with lots of
  smaller nodes to spread the data out then to have a few really big
  nodes.  I think the optimal solution is to have each computer on the
  LAN to have a 50-200 MB store (non-transient) with the main node
  having a 1-2 GB store.
 
 (Damn, dude!  Hit Enter once in a while!)

No

 
 It seems quite likely that in the near future, Freenet is going to be
 (more commonly) used as a transport mechanism for Very Large Files
 (~650 MB).  If the local node's data store is less than the size of the
 file you're trying to retrieve, I doubt that the results are going to
 be pleasant.

This is not a problem with split files.  Event before split files, the node 
would just save it to the hard drive, pass it on to the next node, then delete it.  
You may have problems if the file is larger then your total free hard drive space.  
Again, this is not a problem with 0.4/5 split files.

 I can't see any reason why you'd want a data store less than 1 GB,
 unless your hard drive is simply so small that you can't have a node
 that big.  (In which case you won't be downloading ISO images, so you
 won't face these issues in the first place.)

Large datastores tend to centralize the network.  Datastores don't fill up as 
quickly and your node caches more data and less data falls out.  On the surface, 
this seems like an advantage;  indeed, for a node operator's short term gain, it 
is an advantage.  However, over the long term it tends to hurt routing.  Nodes 
won't be requesting as much data from other nodes, and thus won't discover new 
nodes through requests.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Mark J Roberts

Timm Murray:
 Large datastores tend to centralize the network.  Datastores don't fill up as 
 quickly and your node caches more data and less data falls out.  On the surface, 
 this seems like an advantage;  indeed, for a node operator's short term gain, it 
 is an advantage.  However, over the long term it tends to hurt routing.  Nodes 
 won't be requesting as much data from other nodes, and thus won't discover new 
 nodes through requests.

I haven't seen _any_ compelling argument why above-average nodes
should attract more than their fair share of requests. What's
yours?

I've heard the whispered rumors about simulations suggesting that
Freenet will deny some nodes traffic while overloading others, and I
think the problem, if there actually is one, can easily be fixed by
varying the datasource-reset frequency inversely with request load.

Anyway, I'm not too impressed with arguments that nodes won't see
enough requests. Freenet routing is grossly inefficient. Nodes are
going to be _falling_over_ with requests when it's actually used.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Mark J Roberts

Timm Murray:
 Over time, the large node simply accumulates more data from Freenet.  This means 
 there should be more nodes which point to data on the large node. Thus, there will 
 be more requests routed to the large node.

Uhh... so? The node's big; it can handle lots of requests. That's
not a problem. What's a problem is if Freenet says: Hey, your node
is 10% better than usual, so let's send every request to it! I
don't see why this would happen.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Mark J Roberts

Mark J Roberts:
 Timm Murray:
  Over time, the large node simply accumulates more data from Freenet.  This means 
  there should be more nodes which point to data on the large node. Thus, there will 
  be more requests routed to the large node.
 
 Uhh... so? The node's big; it can handle lots of requests. That's
 not a problem. What's a problem is if Freenet says: Hey, your node
 is 10% better than usual, so let's send every request to it! I
 don't see why this would happen.

Looks like I totally, embarassingly missed your point the first time
around... I'll try this again. :}

You're right: it's true that the bigger your store, the more
requests your node will receive. And overloaded nodes are bad
because they make the network unreliable. So our objective is
obviously to prevent overload.

Well, I really detest artificially constricting the store size in
order to regulate request load. But I _do_ understand your point
now: larger stores demand more bandwidth. Low-bandwidth nodes need
some way to avoid overload, and constricting the size of the store
is one way to do it. The _wrong_ way to do it

What's creepy about this are the various heuristics proposed to
accomplish it: don't make a large store!, if you have Y
bandwidth, use a X megabyte store!, etc. It's impossible to find an
acceptable one. Which means that nodes will have to detect overload
and adjust themselves. Which also means that we can stop promoting
this dangerous small-store idea - the recommended size of the store
should be based on the memory required to index its contents.

One solution might be to, when overloaded, set the datasource not to
yourself but to where you would've routed the request if you didn't
have the data. You'd really actually use a probability scaled by the
current throughput.

Needs more processing.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-19 Thread Michael

 This feature is completely incompatible with Freenet's goals and
 architecture (insofar as I understand the latter).  Freenet isn't
 Gnutella.

Thank gawd. Gnutella really does suck. Mostly because of users who take
without giving.

 In fact, the closest approximation to Freenet that I'm aware of is
 Usenet.  When you insert an article into Usenet, you don't expect
 your news client to store the article in your home directory, then
 tell the news server what the file name is.  The news server might
 not even be on the same computer as the client!  (And might not share
 a filespace, e.g. with NFS or SMB.)

I fully understand this. I don't see how that changes anything.

 Freenet is a client/server architecture, like (modern, post-NNTP) Usenet.
 Your client has some data; it might have come from a file, or you might
 have piped it in on stdin, or it might have spontaneously generated itself
 for all we know.  The client opens an FCP connection to a Freenet node --
 which may be on the same machine or not.  It does some protocol
 handshaking, then sends the data to the node.

Obviously the symlink would only work on local machines unless for some
reason you wanted to make it able to addle URL's also but I don't see how
that would help anybody. The protocol is fine. It's just a simple issue of
file storage exactly as storing any other file would be. The rest of the
Freenet network shouldn't be able to tell any difference.

 If you think your feature is going to be easy to implement (which I
 personally doubt), then go ahead and submit your patch for peer review
 after you've added it.

I haven't looked much at the FreeNet code but if it uses decent
abstractions for the caching portions it shouldn't be very difficult. I
have a half dozen projects I'm working on now so I really haven't time to
start hacking something new yet. When I come to the point of needing it
I'll add it. I'd rather wait for a 1.0 version from the guys upstairs
first anyway since it sounds like nobody else would want to keep the
feature in and I'd hafta rehack it for each new version. If I fall flat on
my ass in my efforts well I guess I'll owe everyone a beer or
something. ;


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-18 Thread Timm Murray



  Every computer should be running a Freenet node anyway. There are
  ways to make Freenet work behind firewalls, though it's a bit
  tricky last I heard. Something like a Squid proxy server for
  Freenet is actualy redundant.
 
 That's ridiculous. Run _one_ node per LAN.

Sorry Mark, but the one computer, one node rule is something I picked up from Oskar 
(I think), and I decided I agreed with him (and not just because he's Oskar).  Freenet 
works better with lots of smaller nodes to spread the data out then to have a few 
really big nodes.  I think the optimal solution is to have each computer on the LAN to 
have a 50-200 MB store (non-transient) with the main node having a 1-2 GB store.  Node 
announcements always go through the main node.  You can skip a few computers if you 
have security concerns (such as on a web server).  Then set your LANs firewall to 
block everything incoming and outgoing (except what is explicity allowed; you should 
do this anyway).  Allow the main node to go out on the various Freenet ports.  This 
should make it so that internal nodes don't learn about nodes outside the LAN (except 
for the main node).  They will learn about new nodes when requests are sent through 
the main node, but those new nodes will be flushed when they can't contact that node 
because of the firewall.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-18 Thread Mark J Roberts

Greg Wooledge:
 Timm Murray ([EMAIL PROTECTED]) wrote:
  Sorry Mark, but the one computer, one node rule is something I
  picked up from Oskar (I think), and I decided I agreed with him
  (and not just because he's Oskar).

Does Freenet routing _really_ seem all that attractive to you for
the average LAN? WHY??

  Freenet works better with lots of smaller nodes to spread the
  data out then to have a few really big nodes.

If, by works better, you mean resists attack through
decentralization, or prevents overload by mirroring data. But,
uhh..., who's attacking individual nodes on a fucking LAN? Is your
LAN _really_ big enough to suffer from slashdot effects... and if
so, how does adding MULTIPLE HOPS to each request help your
suffering network?

 It seems quite likely that in the near future, Freenet is going to be
 (more commonly) used as a transport mechanism for Very Large Files
 (~650 MB).  If the local node's data store is less than the size of the
 file you're trying to retrieve, I doubt that the results are going to
 be pleasant.

I don't believe in caching locally inserted or requested data. Say
you have a 1G store, and you download a 650M movie. It's cached.
Congratulations, you just _wiped_out_ half the useful data in your
store! Hope you don't download another one

Yes, I realize that this invites evil nodes to attempt to verify
that we've actually cached the data by checking how quickly their
request to us for it is fulfilled. However, caching many parts of
the same splitfile is almost as bad, and splitfiles aren't going
away. So I think the argument against caching wins, for now. It's an
unpleasant decision.

And storing partial data for later completion is the client's job.

 I can't see any reason why you'd want a data store less than 1 GB,
 unless your hard drive is simply so small that you can't have a node
 that big.  (In which case you won't be downloading ISO images, so you
 won't face these issues in the first place.)

Indeed. Each node should use all the disk space and bandwidth it
can. And those heretics who'd like to force-homogenize Freenet
should be force-homogenized in a large industrial blender.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



[freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread David McNab

I look forward to the day when I visit slashdot and see a story on Freenet,
with the 'hops' logo at the top right of the story - ie freenet having its
own category on /.




___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Michael

 Yes, in fact if you hang around long enough, you will find a lot of
 people suggesting using Freenet to overcome the Slashdot Effect.  
 Freenet is still slower then HTTP in the sense that it will use more
 bandwidth.  An HTTP server under normal bandwidth load will allways be
 faster then Freenet, but Freenet is infinatly faster then not being
 able to get the document at all.

Would it be possible to add FreeNet into something like Squid so that any
web browser inside a LAN that is already set to use our Squid proxy server
would be able to request FreeNet objects without needing to run nodes on
every machine or to configure each machine to use a special FreeNet
proxy? It'd seem to me that'd be a major issue on making it available
inside firewalls and such. Assuming that is even a desired goal for
FreeNet.


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re:Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Timm Murray

  Yes, in fact if you hang around long enough, you will find a lot of
  people suggesting using Freenet to overcome the Slashdot Effect.  
  Freenet is still slower then HTTP in the sense that it will use more
  bandwidth.  An HTTP server under normal bandwidth load will allways be
  faster then Freenet, but Freenet is infinatly faster then not being
  able to get the document at all.
 
 Would it be possible to add FreeNet into something like Squid so that any
 web browser inside a LAN that is already set to use our Squid proxy server
 would be able to request FreeNet objects without needing to run nodes on
 every machine or to configure each machine to use a special FreeNet
 proxy? It'd seem to me that'd be a major issue on making it available
 inside firewalls and such. Assuming that is even a desired goal for
 FreeNet.

Every computer should be running a Freenet node anyway. There are ways to make Freenet 
work behind firewalls, though it's a bit tricky last I heard. Something like a Squid 
proxy server for Freenet is actualy redundant.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread David McNab

From: Michael [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Thursday, October 18, 2001 7:25 AM
Subject: Re: [freenet-chat] Slashdot needs Freenet category with Hops


 Would it be possible to add FreeNet into something like Squid so that any
 web browser inside a LAN that is already set to use our Squid proxy server
 would be able to request FreeNet objects without needing to run nodes on
 every machine or to configure each machine to use a special FreeNet
 proxy?

Very much, YES! :)
Have a look at fcpproxy. It's in freenet CVS - more precisely,
Contrib/fcptools/fcpproxy (NOT Freenet/Contrib/...).

fcpproxy in its present form is a hack of the Junkbusters proxy server. It's
got a hook that recognises when an http request should go to Freenet, and
instead requests the key using the ezFCPlib C functions.

No doubt Squid can be easily hacked in a similar way.

David




___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re:Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Michael

 Every computer should be running a Freenet node anyway. There are ways
 to make Freenet work behind firewalls, though it's a bit tricky last I
 heard. Something like a Squid proxy server for Freenet is actualy
 redundant.

I mean that from a security standpoint I would not want every machine on
my LAN running FreeNet. Not that I don't trust the programmers of FreeNet
but each new bit of ware you have to run on each machine is just one more
thing to worry about. The entire LAN is essentially running as a single
computer on the Net so it doesn't make sense to run more than one node and
it makes the most sense to run that node as part of the proxy server that
is already setup. There is already some proxy server FProxy or something
like that which lets you use a local machine to host FreeNet objects
within the web browser. Couldn't you just wrap that into an already
existing proxy server (once it reaches a stable API) to simplify
installation and admining?


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Mark J Roberts

Timm Murray:
   Yes, in fact if you hang around long enough, you will find a
   lot of people suggesting using Freenet to overcome the
   Slashdot Effect.  Freenet is still slower then HTTP in the
   sense that it will use more bandwidth.  An HTTP server under
   normal bandwidth load will allways be faster then Freenet, but
   Freenet is infinatly faster then not being able to get the
   document at all.

Splitfiles make Freenet faster than any webserver - the bandwidth
attainable should be limited by the size of the network and average
node throughput.

  Would it be possible to add FreeNet into something like Squid so
  that any web browser inside a LAN that is already set to use our
  Squid proxy server would be able to request FreeNet objects
  without needing to run nodes on every machine or to configure
  each machine to use a special FreeNet proxy? It'd seem to me
  that'd be a major issue on making it available inside firewalls
  and such. Assuming that is even a desired goal for FreeNet.

Yes.

 Every computer should be running a Freenet node anyway. There are
 ways to make Freenet work behind firewalls, though it's a bit
 tricky last I heard. Something like a Squid proxy server for
 Freenet is actualy redundant.

That's ridiculous. Run _one_ node per LAN.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Michael

 Very much, YES! :)
 Have a look at fcpproxy. It's in freenet CVS - more precisely,
 Contrib/fcptools/fcpproxy (NOT Freenet/Contrib/...).
 
 fcpproxy in its present form is a hack of the Junkbusters proxy server. It's
 got a hook that recognises when an http request should go to Freenet, and
 instead requests the key using the ezFCPlib C functions.
 
 No doubt Squid can be easily hacked in a similar way.

Thanks. Exactly the information I was looking for. Any idea if any of the
Squid developers have been involved in FreeNet at all? I always thought
the Junkbusters proxy was based on Squid but have never used it so I guess
I was wrong. :)

Am I correct in thinking that the Squid server would then run a node as
usual and using that C library would just push the file back to the user
as it would have with any other web request? Hrm.. Squid has the ability
to have certain portions of itself scripted out to do interesting
things. Possibly this could be implemented even as a shell script? Will
hafta look.. :)


___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Don Marti

begin Mark J Roberts quotation of Wed, Oct 17, 2001 at 05:29:11PM -0500:

 That's ridiculous. Run _one_ node per LAN.

Only 2GB of Freenet storage for a LAN that may have terabytes of
unused hard drive space on desktop systems?  Why?

You could certainly make a security case for having one gateway
node that talks to the outside, but you might as well use that free
space on all those computers to make the Freenet experience better
and faster for all the users on the LAN.

-- 
Don Marti  What do we want?  Free Dmitry!  When do we want it?  Now! 
http://zgp.org/~dmarti  Free Dmitry: http://eff.org/
[EMAIL PROTECTED]  Free the web, burn all GIFs: http://burnallgifs.org/

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread David McNab

From: Don Marti [EMAIL PROTECTED]
  That's ridiculous. Run _one_ node per LAN.

 Only 2GB of Freenet storage for a LAN that may have terabytes of
 unused hard drive space on desktop systems?  Why?

Tavin is presently re-writing the datastore code, which will allow the
creation of massive datastores which will physically exist as a set of 2GB
files.

AFAIK, Tavin will be releasing the updated code within about a week.

David




___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Mark J Roberts

Michael:
 Would/is it be possible to serve files in place? Rather than packing them
 into cache files? Something more Napster/Gnutella like just because it
 seems painful to have to have two copies of the same files on my hdd.

Why do you think Freenet would ever ask your node for that data?

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread David McNab

From: Michael [EMAIL PROTECTED]

 Would/is it be possible to serve files in place? Rather than packing them
 into cache files? Something more Napster/Gnutella like just because it
 seems painful to have to have two copies of the same files on my hdd.

There are heavy legal problems with that:

1) It becomes possible to censor freenet, which makes it answerable to the
DMCA
2) If your PC gets hacked or confiscated, the datastore may contain
unencrypted files which are illegal in your country. You could go to jail,
or even be executed for possession of certain files in certain countries

Freenet encrypts the datastore and consolidates it into one or more large
files for very good reasons:
1) No one can determine exactly what's on your disk, not even you. Not
unless you request a given key at htl 0 while your node is disconnected.
2) There is no way you can eliminate any file from the datastore without
destroying the whole store. That keeps freenet uncensorable, and eliminates
the ability to monitor content (such ability is necessary for a node
operator to be answerable to the DMCA).
3) You cannot be held responsible for the contents of your datastore,
because it can't be proved that you requested or inserted such materials.
Legally, you should be able to fall into the category of a 'caching online
service provider'. Here, freenet provides a level of 'plausible
deniability'.

David





___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Mark J Roberts

David McNab:
 Freenet encrypts the datastore and consolidates it into one or more large
 files for very good reasons:

I don't see many good reasons at all. (Does anyone else?)

 1) No one can determine exactly what's on your disk, not even you. Not
 unless you request a given key at htl 0 while your node is disconnected.

Which is trivial.

 2) There is no way you can eliminate any file from the datastore without
 destroying the whole store. That keeps freenet uncensorable, and eliminates
 the ability to monitor content (such ability is necessary for a node
 operator to be answerable to the DMCA).

Wrong. Deleting a file from the store is easy.

 3) You cannot be held responsible for the contents of your datastore,
 because it can't be proved that you requested or inserted such materials.
 Legally, you should be able to fall into the category of a 'caching online
 service provider'. Here, freenet provides a level of 'plausible
 deniability'.

But this is also true for the transparent store that he suggested.

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat



Re: [freenet-chat] Slashdot needs Freenet category with Hops

2001-10-17 Thread Don Marti

begin Michael quotation of Wed, Oct 17, 2001 at 08:39:47PM -0500:

 Usually I've always tried to concentrate disk space at the servers. These
 days with 20gigger drives in even the most basic client machines it'd be
 great to use the extra diskspace but I'd still rather serve up that extra
 diskspace via Samba, NFS, or some similar technology that is already built
 into the client OS's and I know how to secure.

What happens when a desktop machine gets rebooted?  That One Big
Node has to deal with losing and regaining arbitrary large chunks
of datastore at any time.  Sounds hard.

-- 
Don Marti  What do we want?  Free Dmitry!  When do we want it?  Now! 
http://zgp.org/~dmarti  Free Dmitry: http://eff.org/
[EMAIL PROTECTED]  Free the web, burn all GIFs: http://burnallgifs.org/

___
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat