I kinda hate to break this to you, but Oskar isn't exactly going to be doing grunt work. He's been in Freenet development, IIRC, for about two years and has always done the Ugly Innards of Freenet. He wrote documentation once "when I was young and stupid" (so he says). There's no way he's going to be doing grunt work. From what I've read on the Freenet mailing lists, he'll be working on getting the next version of Freenet in working order, particularly a new system of node announcement which will do away with Freenet's last centralization (inform.php).
He was doing for free a long time before he started getting paid for it. So it used to be his leisure time, and most likely will be again once the summer is up.
Oskar was working on Freenet long before he started getting paid to do it. One shouldn't be getting involved in Free Software development because there might be money in it; it should be because you like to do it.
To go off on a tangent, it seems that Linus used to have a much thicker accent then he does now, which is kinda a shame. There is an audio file of him pronouncing "Linux" back before he moved to America. I think it's in "/pub/linux/kernel/silly-sounds" on any Linux kernel mirror.
99% of everything is crap. I hate almost everything coming out of the music industry, but I don't really like a lot of indy music, either. However, since I would much rather give money to a small, aspiring group then a huge, faceless corperation, I'll take the indy artists over the corperate monstrosities any day.
with Gnutella you can claim you just accidentally downloaded it (I searched for Adobe and downloaded the resulting list, how was I to know Photoshop6 was in there) and Gnutella automatically shares files that you've downloaded.
And you can't say you "accidentally downloaded" something off Freenet??? In either case, I don't think the feds will care.
Furthermore, you must realize that your orginal post suggested an idea (requesting files by their hash in order to verify the contents of the file) that has been in Freenet for a long time, and then you flamed Freenet for being behind in development. Um, yeah.
Look, Freenet and Gnutella are about equal in how well developed their respective ideas are, Gnutella just got a click-and-drool GUI longer, so it gives the appearance of being better developed. In truth, both are very beta code.
BTW--Are you the same guy I've been having this long, drawn out flame war with on K5? Or do you two just happen to use the same bunk arguments? Or maybe its a VAST GNUTELLA CONSPERICY:)
As another poster mentioned, Freenet is actualy older then Gnutella. In Freenet, we just prefer to think about what we're doing first before implementing every idea that comes forward. This makes development slower, but it's a better network overall.
Oh, and yes, your idea about "searching for a [hash] instead of a name" has been in Freenet for a long time. It's called a "Content Hash Key" (CHK). See http://www.freenetproject.org/index.php?page=keygu ide for details (section 0.3.1).
In Freenet, this is solved by what we call a "subspace" (which goes under the name of "SVK Subspace Key", or "SSK", an SVK being a public/private keypair). You can only post to a subspace if you have the private encryption key. To request from it, you need the public encryption key. So, you just need to insert under "SSK@/newsTellaDDMMYY.mp3" and no one can hijack it (provided you keep the private key private).
Just a few more episodes to go. I have no idea if the dream thing is true.
Actualy, I think some of the last few episodes have been really good. One that stick out in my mind is one where they saved a ship full of prisoners and the most violent prisoner of them all gets reformed due to some medical treatment from converted Borg nanites (this guy had some sort of genetic disorder, and the nanites fixed it). IMHO, it's one of the best episodes of trek, worthy of TOS. Too bad they didn't have more like it.
Andromeda is good. The characters and dialog suck, but the story line makes up for it.
Seriously, 'Andromeda' is Roddenberry's view of the Star Trek universe 300 years after the fall of the Federation
Really? I didn't know that.
"Earth: Final Conflict" is also really good, if you can keep up with it every week, never missing an episode. It manages to beat "Babylon 5" for the most complex, interwoven plot ever.
There has been some attempt at a GCJ compile. The result, overall, is worse then using a JIT compiler. Maybe as GCJ matures this will change, but not now.
There is a GCJ (which can compile directly to machine code) compile being worked on, and there is quite a bit of intrest in it on the mailing list. Currently, a GCJ-compiled node has worse reliablity then a "real" Java node, and some of the crypto goes slower then what IBM's VM can do.
There is also a C++ implementation called "Whiterose". It is skipping the 0.3 series (which was/is very broken) and going right for 0.4 compliance.
------
Re:Hey Taco, when we gonna get a slash port?
on
SQL Over FreeNet
·
· Score: 1
A lot of the work for this is already done as part of the Everything Over Freenet (EOF) project. Check out the "News" support, which basicly allows you to look at certian documents on Freenet as if they were USENET posts. I bet it would be easy to put this into slash.
Freenet is designed to scale up. It's not very well designed to scale down. Right now, the total size of the network is fairly small and the current referance implementation (0.3) is very broken.
OS/2 and NT are actualy the same OS. Yes, there are some surface diffrences, but down deep, they're the same. MS and IBM developed OS/2 for years. When the deal broke up, IBM and MS took what code they had and pushed them to compete with each other. Thats why OS/2 and NT have "binary compatiblity" with each other.
I'm something of a jack-of-all-computer-trades at the small bussiness I work at, and that includes system administration.
Our one file server is currently a GNU/Linux box (Debian), but the PHB wanted to move it to Windows 2000. Why? "Because it's stable and I can take care of it if you're gone" he said (I'm still in college and do this work part-time). My boss knows just enough about Windows to shoot himself in the foot.
And so he goes and gets a consultant to put Win 2000 on it, who shows my boss the price tag for Win 2000 Server (for the number of users we need, it costs more then the server itself orginaly cost). When my boss saw this, he said "um, lets get Win NT 4.0 instead". Great idea! Get an OS even the Microserfs say is a piece of crap!
So we see here how much support can be important. My PHB was willing (for better or for worse; probably worse) to put up with an inferior server OS just because of support. I hope he knows that, because he'll be needing that support with NT 4.0!
IMHO: I have quite a bit of background in trying to get Win9x systems to work (I'm not yet willing to give up gaming). I found myself constantly having problems getting somethings to work on Win 2000 (little things, mostly; setting up network settings, installation, driver support, etc.). I've never worked with it before in any detail, but from what I've heard, NT 4.0 would be even worse in this regard. For someone like my boss, who only what I would consider basic knoweldge of Win9x, these diffrences are bound to make support through him much more difficult.
A toaster is a really simple device. Some heating elements, a timer, and a spring to pop the toast up and down. How do people compare the maintnance of such a simple device to a computer network with hundreds, perhaps thousands of individual systems, each one having bazillions of transisters made to fit on a very small chunk of sillicon.
SOMEONE had to design all that. It probably took a small team to design a toaster. It takes a much larger team to design a huge corperate Intranet.
This has been gone over on the Freenet mailing lists to quite some extent. When you request some bad evil data through the network, you propagate it to each node in the request path. There is a "DataSource" field in the Freenet protocol that gives you the node that orginaly had it, but there is a 1/30th chance that each node along the request path will reset this field to point to itself (setting this probablity is tricky; too high of a chance and Freenet routing falls apart, but too low of a chance and it makes an attacker more and more sure that such and such data came from node x).
Thus, their are three nodes that you can be sure of that have the data being requested. The first is your own (assuming the attacker is actualy running a node). The second is the node that is next your yours in the request path. The third is the node addressed in the DataSource field (of which their is a 1/30th chance that it is the next node in the request path, thus bringing the list down to two if that happend).
Obviously, the prosecuter won't attack their own node, so the first one is off the list. The second and third node are diffrent matters. The prosecuter can't be sure if the data orginaly came from those two nodes or if they actualy "pulled" the data from somewhere else on the network. That means the data may never have been on that node until the attacker put it there.
This would be akin to a cop planting a baggie of cocaine on someone before making an arrest. This concept is called "entrapment" and is highly illegal. Likely, such a case would be thrown out of court.
Thus, it is in the attacker's intrest to make sure the case never gets to court. Thus, an attacker must resort to the same old trick of send a bark letter to an ISP saying "this evil person is trading on bad data, cut his accsess OR ELSE!" As always, the ISP will be completely spineless and cut the user's accsess.
The only way to fight this is with a widely deployed Freenet. If enough people are doing ligitamet things on Freenet, then we're safe.
This was a matter of great debate on the Freenet lists. Not as relates to this program specificly, but to FProxy (which does a lot of the same things as David McNab's program here, but his has a few more features). FProxy was orginaly designed to be used through a browser, but since it communicates to the browser via HTTP, it can be used right now by any program that already understands HTTP (like apt-get).
The argument against it was that if we created a plugin for getting "freenet://" and used it by default, then we would be forced to make plugins for everything that may want to acsess Freenet through FProxy. This means you need a plugin, not just for browsers, but for XMMS and apt-get and RealPlayer and a dozen other applications. The argument for such plugins was that, while FProxy defaults to using port 8081, there may be anoynimity reasons why you wouldn't want to do that. The only other way to accsess FProxy (with out a "freenet://" plugin) would be through "http://localhost:8081", but that requires FProxy to run on port 8081.
There are plugins to allow "freenet://" on various browsers, but they are unused by default so as to discourage content authors from linking to them.
Slashdot Mirror
Yes, Debian apt-get.
------
I kinda hate to break this to you, but Oskar isn't exactly going to be doing grunt work. He's been in Freenet development, IIRC, for about two years and has always done the Ugly Innards of Freenet. He wrote documentation once "when I was young and stupid" (so he says). There's no way he's going to be doing grunt work. From what I've read on the Freenet mailing lists, he'll be working on getting the next version of Freenet in working order, particularly a new system of node announcement which will do away with Freenet's last centralization (inform.php).
------
He was doing for free a long time before he started getting paid for it. So it used to be his leisure time, and most likely will be again once the summer is up.
------
Oskar was working on Freenet long before he started getting paid to do it. One shouldn't be getting involved in Free Software development because there might be money in it; it should be because you like to do it.
------
Never heard him other than through phospher
To go off on a tangent, it seems that Linus used to have a much thicker accent then he does now, which is kinda a shame. There is an audio file of him pronouncing "Linux" back before he moved to America. I think it's in "/pub/linux/kernel/silly-sounds" on any Linux kernel mirror.
------
99% of everything is crap. I hate almost everything coming out of the music industry, but I don't really like a lot of indy music, either. However, since I would much rather give money to a small, aspiring group then a huge, faceless corperation, I'll take the indy artists over the corperate monstrosities any day.
------
with Gnutella you can claim you just accidentally downloaded it (I searched for Adobe and downloaded the resulting list, how was I to know Photoshop6 was in there) and Gnutella automatically shares files that you've downloaded.
And you can't say you "accidentally downloaded" something off Freenet??? In either case, I don't think the feds will care.
Furthermore, you must realize that your orginal post suggested an idea (requesting files by their hash in order to verify the contents of the file) that has been in Freenet for a long time, and then you flamed Freenet for being behind in development. Um, yeah.
Look, Freenet and Gnutella are about equal in how well developed their respective ideas are, Gnutella just got a click-and-drool GUI longer, so it gives the appearance of being better developed. In truth, both are very beta code.
BTW--Are you the same guy I've been having this long, drawn out flame war with on K5? Or do you two just happen to use the same bunk arguments? Or maybe its a VAST GNUTELLA CONSPERICY :)
------
Actualy, the above link is a bit too basic. See http://www.freenetproject.org/oldsite/index.php?pa ge=keys for a much more verbose description.
------
As another poster mentioned, Freenet is actualy older then Gnutella. In Freenet, we just prefer to think about what we're doing first before implementing every idea that comes forward. This makes development slower, but it's a better network overall.
Oh, and yes, your idea about "searching for a [hash] instead of a name" has been in Freenet for a long time. It's called a "Content Hash Key" (CHK). See http://www.freenetproject.org/index.php?page=keygu ide for details (section 0.3.1).
------
I'm doing something similar to Freenet, although we didn't require any new code to get apt-get to work. See http://eof.sourceforge.net/APT/index.html.
------
In Freenet, this is solved by what we call a "subspace" (which goes under the name of "SVK Subspace Key", or "SSK", an SVK being a public/private keypair). You can only post to a subspace if you have the private encryption key. To request from it, you need the public encryption key. So, you just need to insert under "SSK@/newsTellaDDMMYY.mp3" and no one can hijack it (provided you keep the private key private).
------
We're doing this in Freenet. Check out http://eof.sourceforge.net/APT/index.html.
------
Just a few more episodes to go. I have no idea if the dream thing is true.
Actualy, I think some of the last few episodes have been really good. One that stick out in my mind is one where they saved a ship full of prisoners and the most violent prisoner of them all gets reformed due to some medical treatment from converted Borg nanites (this guy had some sort of genetic disorder, and the nanites fixed it). IMHO, it's one of the best episodes of trek, worthy of TOS. Too bad they didn't have more like it.
------
And speaking of warp 10, what was up with their explaination? "We have some very special dilithium". Oh.
I long for the days of TNG when you technobabble ran the ship.
------
You want 'Andromeda'
Andromeda is good. The characters and dialog suck, but the story line makes up for it.
Seriously, 'Andromeda' is Roddenberry's view of the Star Trek universe 300 years after the fall of the Federation
Really? I didn't know that.
"Earth: Final Conflict" is also really good, if you can keep up with it every week, never missing an episode. It manages to beat "Babylon 5" for the most complex, interwoven plot ever.
------
There has been some attempt at a GCJ compile. The result, overall, is worse then using a JIT compiler. Maybe as GCJ matures this will change, but not now.
------
There is a GCJ (which can compile directly to machine code) compile being worked on, and there is quite a bit of intrest in it on the mailing list. Currently, a GCJ-compiled node has worse reliablity then a "real" Java node, and some of the crypto goes slower then what IBM's VM can do.
There is also a C++ implementation called "Whiterose". It is skipping the 0.3 series (which was/is very broken) and going right for 0.4 compliance.
------
A lot of the work for this is already done as part of the Everything Over Freenet (EOF) project. Check out the "News" support, which basicly allows you to look at certian documents on Freenet as if they were USENET posts. I bet it would be easy to put this into slash.
------
Freenet is designed to scale up. It's not very well designed to scale down. Right now, the total size of the network is fairly small and the current referance implementation (0.3) is very broken.
------
OS/2 and NT are actualy the same OS. Yes, there are some surface diffrences, but down deep, they're the same. MS and IBM developed OS/2 for years. When the deal broke up, IBM and MS took what code they had and pushed them to compete with each other. Thats why OS/2 and NT have "binary compatiblity" with each other.
------
I'm something of a jack-of-all-computer-trades at the small bussiness I work at, and that includes system administration.
Our one file server is currently a GNU/Linux box (Debian), but the PHB wanted to move it to Windows 2000. Why? "Because it's stable and I can take care of it if you're gone" he said (I'm still in college and do this work part-time). My boss knows just enough about Windows to shoot himself in the foot.
And so he goes and gets a consultant to put Win 2000 on it, who shows my boss the price tag for Win 2000 Server (for the number of users we need, it costs more then the server itself orginaly cost). When my boss saw this, he said "um, lets get Win NT 4.0 instead". Great idea! Get an OS even the Microserfs say is a piece of crap!
So we see here how much support can be important. My PHB was willing (for better or for worse; probably worse) to put up with an inferior server OS just because of support. I hope he knows that, because he'll be needing that support with NT 4.0!
IMHO: I have quite a bit of background in trying to get Win9x systems to work (I'm not yet willing to give up gaming). I found myself constantly having problems getting somethings to work on Win 2000 (little things, mostly; setting up network settings, installation, driver support, etc.). I've never worked with it before in any detail, but from what I've heard, NT 4.0 would be even worse in this regard. For someone like my boss, who only what I would consider basic knoweldge of Win9x, these diffrences are bound to make support through him much more difficult.
------
A toaster is a really simple device. Some heating elements, a timer, and a spring to pop the toast up and down. How do people compare the maintnance of such a simple device to a computer network with hundreds, perhaps thousands of individual systems, each one having bazillions of transisters made to fit on a very small chunk of sillicon.
SOMEONE had to design all that. It probably took a small team to design a toaster. It takes a much larger team to design a huge corperate Intranet.
------
It was applied for in Nov. '96. I believe the first Quake came out before then (I'll have to check on that, though).
------
This has been gone over on the Freenet mailing lists to quite some extent. When you request some bad evil data through the network, you propagate it to each node in the request path. There is a "DataSource" field in the Freenet protocol that gives you the node that orginaly had it, but there is a 1/30th chance that each node along the request path will reset this field to point to itself (setting this probablity is tricky; too high of a chance and Freenet routing falls apart, but too low of a chance and it makes an attacker more and more sure that such and such data came from node x).
Thus, their are three nodes that you can be sure of that have the data being requested. The first is your own (assuming the attacker is actualy running a node). The second is the node that is next your yours in the request path. The third is the node addressed in the DataSource field (of which their is a 1/30th chance that it is the next node in the request path, thus bringing the list down to two if that happend).
Obviously, the prosecuter won't attack their own node, so the first one is off the list. The second and third node are diffrent matters. The prosecuter can't be sure if the data orginaly came from those two nodes or if they actualy "pulled" the data from somewhere else on the network. That means the data may never have been on that node until the attacker put it there.
This would be akin to a cop planting a baggie of cocaine on someone before making an arrest. This concept is called "entrapment" and is highly illegal. Likely, such a case would be thrown out of court.
Thus, it is in the attacker's intrest to make sure the case never gets to court. Thus, an attacker must resort to the same old trick of send a bark letter to an ISP saying "this evil person is trading on bad data, cut his accsess OR ELSE!" As always, the ISP will be completely spineless and cut the user's accsess.
The only way to fight this is with a widely deployed Freenet. If enough people are doing ligitamet things on Freenet, then we're safe.
------
This was a matter of great debate on the Freenet lists. Not as relates to this program specificly, but to FProxy (which does a lot of the same things as David McNab's program here, but his has a few more features). FProxy was orginaly designed to be used through a browser, but since it communicates to the browser via HTTP, it can be used right now by any program that already understands HTTP (like apt-get).
The argument against it was that if we created a plugin for getting "freenet://" and used it by default, then we would be forced to make plugins for everything that may want to acsess Freenet through FProxy. This means you need a plugin, not just for browsers, but for XMMS and apt-get and RealPlayer and a dozen other applications. The argument for such plugins was that, while FProxy defaults to using port 8081, there may be anoynimity reasons why you wouldn't want to do that. The only other way to accsess FProxy (with out a "freenet://" plugin) would be through "http://localhost:8081", but that requires FProxy to run on port 8081.
There are plugins to allow "freenet://" on various browsers, but they are unused by default so as to discourage content authors from linking to them.
------