I'm curious what speeds freenet can pull down from individual users.
Why would that matter? Freenet has enormous latency on single-file operations, because it has to go find the file, which may or may not exist, and if it still exists, may or may not be easy to locate. The actual time spent transferring a single HTML or text or even JPEG/GIF file is usually small compared to the search latency.
But if you're downloading a large splitfile, then it will spawn as many simultaneous download threads as you ask it to (if your computer can handle it). I've personally seen non-cached splitfiles download at 18 kB/s averaged over a period of several minutes, with the default 10 threads. For me, that's pretty good. As the story at the top of the page says (you did read it, yes?), other people have seen speeds of up to 100 kB/s on cable modems. (But I wouldn't count on that in the typical case.)
So why can't China or Your-favorite-regime simply block/ban downloading freenet itself?
They can block freenetproject.org and sourceforge.net, but it's also possible to download Freenet from any existing user's node, if that user sets up the distribution servlet. Or if you're old-fashioned, you could just ask your American pen-pal to e-mail it to you. (Well, that won't work if all e-mail is read by the government, as it is in China. But in some other places, it might work.)
It could also simply block the appropriate ports
No, this won't work because Freenet nodes use arbitrary TCP ports. When you install a Freenet node, a random port number is chosen (which you can accept or change) during the configuration.
My guess is Freenet would spiral long run into music and porn.
Instead of guessing, why don't you come find out? Write up, or find, some content you deem to be worthwhile. Publish it in Freenet. See how long it remains available without your having to reinsert it. Set up a cron job to reinsert it automatically every week, every month, or whatever.
If you make it plainly obvious how to avoid sharing your own files, people will quickly turn that off and you'll get a bunch of leeches.
This is not how Freenet operates. Freenet doesn't "share files". It's a publishing system, like the World Wide Web, or Usenet. It's a distributed encrypted hard drive. It doesn't have any content until you put something in it.
The analogous term for "leeches" in Freenet is "transient nodes" -- nodes which can initiate requests (and insert content), but which never answer or route requests for other people. If you're on dial-up, fine, run a transient. Nobody will expect a dial-up user to participate in Freenet. But beware that when you run a transient node, you lose plausible deniability. Any data in your node's data store came from your own requests or inserts. This is a trade-off that you may choose to make, or not.
If you want to participate in Freenet, then just set up a permanent node and let it run. Make sure it's not firewalled, and that you won't exceed any bandwidth limits your ISP may impose on you (Australia, etc.). You don't need to insert any content of your own, unless you wish to.
The freenet distribution comes with (iirc) 100 'seed' nodes. These are exported periodically from the hawk.freenetproject.org node. Your node picks 50 of them and uses them as peers.
Actually, it's 273 at the time of this writing. Hawk is set to keep 300, but 273 is all it could find at the moment.
But yes, your point is well taken. There's always a burst of instability when a lot of new users join the network. It was really bad when 0.5 was released, but the changes in 0.5.1 (and having 273 seed nodes in the snapshots directory instead of 100) make it less of a horror than it was last year.
Ideally, though, once Freenet has grown and matured, the slashdot effect for a given piece of content (say, Ian Clarke's video on the pornBush Freenet site) should not exist.
I want to be able to run peer-to-peer services and not have to worry about exceeding some stupid number of bits per unit time. I want a static IP address. I want no ports to be blocked, in either direction; in fact, I want no restrictions on the nature of the traffic I allow, in terms of direction, port numbers, protocols, content, etc., beyond what the lawyers will require ("don't do anything that would get you arrested or sued; if you do, we let you hang for it").
Right now, I'm getting all of this with a 512/256 ADSL connection from my local telco. It's a bit expensive/slow ($40 USD/month for the service, $20/month for the static IP), but I have no complaints. I've gotten what I paid for. How many people in, say, Australia can claim that?
A 100 Mbps fiber connection to my house? Sounds good. A 5 GB/month bandwidth cap? Not allowed to run services? Hahahahaha! Excuse me while I ignore your company.
Are the slashdot editors really such incompetent Linux fanboys that they all think the 2.2 kernel series is dead? My god... they probably even think 2.0 is dead!
[kernel.org] The latest stable version of the Linux kernel is: 2.4.20 The latest prepatch for the stable Linux kernel tree is: 2.4.21-pre5 The latest beta version of the Linux kernel is: 2.5.64 The latest snapshot for the beta Linux kernel tree is: 2.5.64-bk5 The latest 2.2 version of the Linux kernel is: 2.2.24 The latest 2.0 version of the Linux kernel is: 2.0.39
I run 2.2.x at home whenever possible. I have one box on which I must run 2.4.x in order to be able to run X, because it's got an i810 video chipset in it, and 2.2.x doesn't have agpgart. So I have to run 2.4.18 [1] on that box. But when I run 2.4.18, I can't use both X and the NFS client at the same time. Starting up X and then attempting to access files in an NFS-mounted file system causes whatever process is touching NFS to go into La-La-Land and never come back. This even prevents a clean reboot -- I have to sit through an fsck (fortunately ext3 makes it faster). So I stopped running NFS on that box.
And people wonder why I switched my server to OpenBSD!
At work, I administer a Red Hat server which I recently updated to kernel 2.0.39. From kernel 2.0.36. Unfortunately, that by itself didn't fix the problem I was seeing, but adding more swap space seems to have done the trick. The other Linux systems at work all run 2.2.* kernels.
Linus lied to you. There is no 2.4.20. It's really 2.4.0-pre20. If you read the 2.4.* kernel names that way, you will understand why your boxes keep having the problems they're having. And while you're sitting there, waiting for 2.4.18 to compile, you might as well read up on the IDE driver problems in 2.4.19 and 2.4.20. Higher numbers are not better.
[1] 2.4.18: the least benightedly unstable 2.4.x kernel it's even been my profound displeasure not to be able to avoid running.
using Debian stable would give you a standard 2.2.
Yes, if you're smart enough to choose the 2.2.20 kernel, or if you're dumb enough not to have read the install guide that tells you how to choose the 2.4.18 kernel.
There's a huge middle ground of people who are neither smart enough to choose 2.2.20 nor dumb enough to have had 2.2.20 thrust upon them, who are running 2.4.18-bf2.4. And then they come into #debian and ask why they don't have any modules available. But the same applies to the dummies who got 2.2.20-idepci thrust upon them, because that one doesn't have any modules either.
Moral of the story: build your own kernel. Do it right. And remember, apt-get install libncurses5-dev!
Whoever decided that 2.2.20-idepci should be the default kernel for idiots who just press Enter without reading what's written on the screen: I hope you got a good laugh out of our misery.
I've had the privilege of using one of these at work. But it seems that I'm using it for a completely different purpose than the typical/.er.
At work, I manage several HP-UX workstations. These are older models (B132L+, B180L) and only have SCSI interfaces -- no IDE.
We're currently looking into DVD-RW and related media for data archiving. But all of the reasonably priced DVD writing drives are IDE, not SCSI. The only SCSI DVD writer I found, last time I looked around the web, ran $2500!
A Sony DVD+RW IDE drive costs $300. An IDE-to-SCSI converter costs around $65. You also need a Y-cable for the power, since the B-series workstations don't have a third power cable for the adapter. (The one we're using requires external power.) Anyway, cram that all inside the case (not trivial, but possible) and you get a SCSI DVD writer that works just fine in HP-UX for less than $400 USD.
Now, if only there were actually DVD+RW software available for non-Linux systems... that would make my life much easier. But I'll settle for DVD-RW.;)
The real ident servers need to run as root (since they're running on a low port), or if you want to be fancy, they can be started by root and assume another (perhaps jailed) user's identity. Let's assume they all running as root, since I've not seen one that doesn't do so. They need to access/etc/passwd (or the NIS equivalent) as well.
You're just scratching the surface of the problem here. It goes even deeper than that. An ident daemon needs to be able to read kernel memory. Why? Think about it. You connect from host A port a to host B port b. Host B then connects to host A port 113 and says "What is the name of the user who connected to me from port a?" A normal userspace daemon has no way to answer this question. It needs to go poking around inside kernel memory in order to learn who has bound to port a.
In addition to this, I'd like to point out another problem associated with identd: firewalls. Right now, I'm at work, behind a firewall. When I connect from my workstation to a remote service, it might wish to connect back to me for purposes of querying my ident service. But it can't -- port 113 is blocked at the router. Even if I wanted to run ident (which I don't), I wouldn't be able to respond to that query, because I never receive it. Eventually the request times out. This means that connecting to any service that blocks awaiting an ident response has a built-in delay. And if the ident response is mandatory, it means I'm completely shut out.
At home, I just run nullidentd. I return the string rfc1413 for the response. Most people don't react to that, but I've had a few people ask "Which RFC is that?", and one or two have complimented me on it.
gnome-panel version 1.4.x works on my setup. gnome-panel 2.x does not, because it uses this gconf2 thing. (And as you can see, the Debian maintainers for gconf2 are a bit less than helpful in this situation.)
My NFS server is OpenBSD. I've configured NFS version 3 support on the Linux client. Locking just does not work. I've spoken with another Debian+OpenBSD user in IRC and he's confirmed that locking does not work. I consider the problem intractable -- I just ssh to the OpenBSD box to run mutt instead of trying to run mutt on the Debian client.
Hahahahah! Oh yeah, I can see it now. GNOME 2, which totally and utterly refuses to work if your $HOME is NFS-mounted, will become the default for Solaris, from Sun, the people who invented NFS.
This should be amazingly funny to watch. I'll have to stock up on popcorn.
08:58 greycat> ~laugh at gnome 2
08:58 @apt> HAHAHA! AH-HAHA! gnome 2 just cracks me up!
I visited the site with Mozilla 1.1 on HP-UX 10.20 (which I am also using for Slashdot). The front page consists of some text and a single animated-GIF button in the middle of the page. When I click the button I get a more complex page with a login box on the right, a navigation menu across the top, etc. It looks like a fairly normal web site. I saw no pop-ups, and I wasn't called any rude names or told to go away, etc.
Oh, did I mention that I have Javascript turned completely off? Really, Javascript serves no legitimate purpose. It is solely a tool of malice.
This is the fourth time I've downloaded a Phoenix build, and the second time that running it does nothing visible. My machine is a Pentium II running Windows 95. (I think it's Win95B. Not sure. "VER" says "Windows 95. [Version 4.00.1111]".)
The first one (a nightly build somewhere around 0.1) and the last one (0.4 milestone) do not run. They both give me the same behavior: after typing "PHOENIX", the mouse cursor changes to the arrow-plus-hourglass for a few seconds, then changes back to normal. There is no new task in the task manager after that. For all visible purposes, phoenix crashed with no errors and no core file. (Of course, there's no core file. This is Windows. Why would we be able to debug it? No strace, no gdb, no truss,....)
Phoenix 0.2 and 0.3 both run perfectly well on this same computer. So I've gone back to 0.3.
"PHOENIX -ProfileManager" also does nothing. Same symptoms. I've even tried deleting the 0.3 profiles using Phoenix 0.3's ProfileManager before running Phoenix 0.4. No change.
middle-click on links to open them in a new tab [...] Now I'd just like the same behavior on form buttons
Drooool. I've been wishing for that for about 4 years now. (I might actually participate in more Slashdot polls if I didn't have to destroy my copy of the main page to do so.)
And why oh why can't I hit ESC to stop animated GIFs? Netscape 4.x can do this. Debian's versions of Mozilla can do it. Yet upstream Mozilla, for some reason, cannot. (And I can't get Phoenix 0.4 to run, so....)
Where is the hardware going to come from to build these Freenet archives? Is it just going to fall out of the sky?
It is already here. It's provided by the users of Freenet.
But you failed to notice that writings in these categories published anonymously would be basically of no value to anyone.
The second category in which I used the word "uncensorable" was opinions. Opinions are of value to whomever values them. Period.
And the first category was "criticism" (of various authoritarian or pseudo-authoritarian organizations). Now, you're implying that expression of criticism is of no value unless you know who wrote it. I disagree -- I think that an anonymous tip is of enormous value. Imagine what would happen if the police never investigated anonymous tips. Imagine the opposite scenario, in which an anonymous publication describes corruption inside a government office and gives details. Those details can be checked and amplified by regular journalists once the story has been exposed.
But in a Freenet world, where anybody with a computer could publish anything, the lies would soon vastly outnumber the truths. Freenet would be a vast collection of bullshit with a few pearls buried in it.
You've just described the entire world -- books, magazines, newspapers, television shows, radio shows, world wide web sites, Freenet sites, slashdot. All of it.
If you can't handle the burden of analyzing written material yourself to see whether it's of value to you, then go ahead and continue in your little spoon-fed world where only the texts on the Big Brother Approved List (To Save The Children) are worth reading. But for me, I'll take freedom of choice, thank you.
I eagerly await your superior layman's explanation, Mr. Coward. I'd really love to hear it. I'm sure Ian Clarke would love to hear it as well. He could add it to the sorely lacking Freenet newbie documentation.
It sure would have been nice for that gpl outfit to make certain their program would run with the current kaffe stable series, eh?
I run a node on OpenBSD 3.1 i386 using Kaffe 1.0.7 with a couple asserts commented out. It only crashes once or twice a week at most.
mjr claims that he has never had his Freenet node crash, and he's running Kaffe CVS from 20010902.
hobx claims that he got Freenet to run on Kaffe CVS head with only one small patch to Kaffe. (He didn't say whether it would stay running.)
All Java implementations suck. Your job is to find one that sucks as little as possible in your environment.
(I am a lifetime member of the "I Hate Java" club. If Freenet didn't exist, or wasn't written in Java, I would purge the evil Java from all of my computers, while capering with a demented glee. If nothing else, my months of experience with Freenet have taught me just how bad -- and how unportable -- Java is.)
Oh yeah, because searching freenet for "deep throat" is sure to give you secret information right at the top of the list.
You can't search Freenet. You surf it, the way you surf the World Wide Web when you don't have Google. (Don't any of you remember the WWW before search engines?) Then you bookmark the good sites you find.
If you want to search Freenet, then you'll have to implement a Google of your own. Spider it, then index everything you find, then search your index. I know at least one person is working on this right now.
Correct me if I'm wrong, but the only way to find something on freenet is to search for it.
Actually, you can't "search" as most people use that term. You retrieve documents from Freenet by specifying their key. You have to learn the key somehow, usually from another Freenet document.
The whole point in whistle blowing is that nobody knows about whatever you're uncovering. If they don't know about it, how can they search for it?
Some of the popular Freenet site authors have a way to send them messages using KSK@ keys. This is normally used for Freesite submission -- for example, it's how TFE learns about new Freenet sites so that he/she can list them.
So if I were going to do some whistle-blowing, I'd create my Freenet site (could even be a single text file), insert it into Freenet as a one-shot or edition site (certainly not a DBR), and then submit the key to TFE's submissions bin. And possibly a few other Freesite authors' submissions bins as well.
Almost. It's ASCII after the first four bytes, but for some reason that I don't quite grasp, the Freenet developers decided that you would have to send \000\000\000\002 at the beginning of each FCP request.
Slashdot Mirror
I'm sitting on a good broadband connection [...] speeds up to 1.5 MB/s
Holy shit. That's EIGHT TIMES the speed of a T1. That goes a little tiny bit beyond what I'd call "good".
/me mumbles something about 1200 baud modems, and uphill both ways.
I'm curious what speeds freenet can pull down from individual users.
Why would that matter? Freenet has enormous latency on single-file operations, because it has to go find the file, which may or may not exist, and if it still exists, may or may not be easy to locate. The actual time spent transferring a single HTML or text or even JPEG/GIF file is usually small compared to the search latency.
But if you're downloading a large splitfile, then it will spawn as many simultaneous download threads as you ask it to (if your computer can handle it). I've personally seen non-cached splitfiles download at 18 kB/s averaged over a period of several minutes, with the default 10 threads. For me, that's pretty good. As the story at the top of the page says (you did read it, yes?), other people have seen speeds of up to 100 kB/s on cable modems. (But I wouldn't count on that in the typical case.)
So why can't China or Your-favorite-regime simply block/ban downloading freenet itself?
They can block freenetproject.org and sourceforge.net, but it's also possible to download Freenet from any existing user's node, if that user sets up the distribution servlet. Or if you're old-fashioned, you could just ask your American pen-pal to e-mail it to you. (Well, that won't work if all e-mail is read by the government, as it is in China. But in some other places, it might work.)
It could also simply block the appropriate ports
No, this won't work because Freenet nodes use arbitrary TCP ports. When you install a Freenet node, a random port number is chosen (which you can accept or change) during the configuration.
My guess is Freenet would spiral long run into music and porn.
Instead of guessing, why don't you come find out? Write up, or find, some content you deem to be worthwhile. Publish it in Freenet. See how long it remains available without your having to reinsert it. Set up a cron job to reinsert it automatically every week, every month, or whatever.
If you make it plainly obvious how to avoid sharing your own files, people will quickly turn that off and you'll get a bunch of leeches.
This is not how Freenet operates. Freenet doesn't "share files". It's a publishing system, like the World Wide Web, or Usenet. It's a distributed encrypted hard drive. It doesn't have any content until you put something in it.
The analogous term for "leeches" in Freenet is "transient nodes" -- nodes which can initiate requests (and insert content), but which never answer or route requests for other people. If you're on dial-up, fine, run a transient. Nobody will expect a dial-up user to participate in Freenet. But beware that when you run a transient node, you lose plausible deniability. Any data in your node's data store came from your own requests or inserts. This is a trade-off that you may choose to make, or not.
If you want to participate in Freenet, then just set up a permanent node and let it run. Make sure it's not firewalled, and that you won't exceed any bandwidth limits your ISP may impose on you (Australia, etc.). You don't need to insert any content of your own, unless you wish to.
The freenet distribution comes with (iirc) 100 'seed' nodes. These are exported periodically from the hawk.freenetproject.org node. Your node picks 50 of them and uses them as peers.
Actually, it's 273 at the time of this writing. Hawk is set to keep 300, but 273 is all it could find at the moment.
But yes, your point is well taken. There's always a burst of instability when a lot of new users join the network. It was really bad when 0.5 was released, but the changes in 0.5.1 (and having 273 seed nodes in the snapshots directory instead of 100) make it less of a horror than it was last year.
Ideally, though, once Freenet has grown and matured, the slashdot effect for a given piece of content (say, Ian Clarke's video on the pornBush Freenet site) should not exist.
So, what do you want [...]?
(Why isn't every company asking me this?)
I want to be able to run peer-to-peer services and not have to worry about exceeding some stupid number of bits per unit time. I want a static IP address. I want no ports to be blocked, in either direction; in fact, I want no restrictions on the nature of the traffic I allow, in terms of direction, port numbers, protocols, content, etc., beyond what the lawyers will require ("don't do anything that would get you arrested or sued; if you do, we let you hang for it").
Right now, I'm getting all of this with a 512/256 ADSL connection from my local telco. It's a bit expensive/slow ($40 USD/month for the service, $20/month for the static IP), but I have no complaints. I've gotten what I paid for. How many people in, say, Australia can claim that?
A 100 Mbps fiber connection to my house? Sounds good. A 5 GB/month bandwidth cap? Not allowed to run services? Hahahahaha! Excuse me while I ignore your company.
Are the slashdot editors really such incompetent Linux fanboys that they all think the 2.2 kernel series is dead? My god... they probably even think 2.0 is dead!
[kernel.org]
The latest stable version of the Linux kernel is: 2.4.20
The latest prepatch for the stable Linux kernel tree is: 2.4.21-pre5
The latest beta version of the Linux kernel is: 2.5.64
The latest snapshot for the beta Linux kernel tree is: 2.5.64-bk5
The latest 2.2 version of the Linux kernel is: 2.2.24
The latest 2.0 version of the Linux kernel is: 2.0.39
I run 2.2.x at home whenever possible. I have one box on which I must run 2.4.x in order to be able to run X, because it's got an i810 video chipset in it, and 2.2.x doesn't have agpgart. So I have to run 2.4.18 [1] on that box. But when I run 2.4.18, I can't use both X and the NFS client at the same time. Starting up X and then attempting to access files in an NFS-mounted file system causes whatever process is touching NFS to go into La-La-Land and never come back. This even prevents a clean reboot -- I have to sit through an fsck (fortunately ext3 makes it faster). So I stopped running NFS on that box.
And people wonder why I switched my server to OpenBSD!
At work, I administer a Red Hat server which I recently updated to kernel 2.0.39. From kernel 2.0.36. Unfortunately, that by itself didn't fix the problem I was seeing, but adding more swap space seems to have done the trick. The other Linux systems at work all run 2.2.* kernels.
Linus lied to you. There is no 2.4.20. It's really 2.4.0-pre20. If you read the 2.4.* kernel names that way, you will understand why your boxes keep having the problems they're having. And while you're sitting there, waiting for 2.4.18 to compile, you might as well read up on the IDE driver problems in 2.4.19 and 2.4.20. Higher numbers are not better.
[1] 2.4.18: the least benightedly unstable 2.4.x kernel it's even been my profound displeasure not to be able to avoid running.
using Debian stable would give you a standard 2.2.
Yes, if you're smart enough to choose the 2.2.20 kernel, or if you're dumb enough not to have read the install guide that tells you how to choose the 2.4.18 kernel.
There's a huge middle ground of people who are neither smart enough to choose 2.2.20 nor dumb enough to have had 2.2.20 thrust upon them, who are running 2.4.18-bf2.4. And then they come into #debian and ask why they don't have any modules available. But the same applies to the dummies who got 2.2.20-idepci thrust upon them, because that one doesn't have any modules either.
Moral of the story: build your own kernel. Do it right. And remember, apt-get install libncurses5-dev!
Whoever decided that 2.2.20-idepci should be the default kernel for idiots who just press Enter without reading what's written on the screen: I hope you got a good laugh out of our misery.
I'm using this product: IDSC21-E. The current price at DeviceNet-USA is $68.00 USD.
Best Vaporware of 1997... 1998... 1999... 2000... 2001... 2002...: Grimoire! (Site requires Javascript.)
I've had the privilege of using one of these at work. But it seems that I'm using it for a completely different purpose than the typical /.er.
;)
At work, I manage several HP-UX workstations. These are older models (B132L+, B180L) and only have SCSI interfaces -- no IDE.
We're currently looking into DVD-RW and related media for data archiving. But all of the reasonably priced DVD writing drives are IDE, not SCSI. The only SCSI DVD writer I found, last time I looked around the web, ran $2500!
A Sony DVD+RW IDE drive costs $300. An IDE-to-SCSI converter costs around $65. You also need a Y-cable for the power, since the B-series workstations don't have a third power cable for the adapter. (The one we're using requires external power.) Anyway, cram that all inside the case (not trivial, but possible) and you get a SCSI DVD writer that works just fine in HP-UX for less than $400 USD.
Now, if only there were actually DVD+RW software available for non-Linux systems... that would make my life much easier. But I'll settle for DVD-RW.
The real ident servers need to run as root (since they're running on a low port), or if you want to be fancy, they can be started by root and assume another (perhaps jailed) user's identity. Let's assume they all running as root, since I've not seen one that doesn't do so. They need to access /etc/passwd (or the NIS equivalent) as well.
You're just scratching the surface of the problem here. It goes even deeper than that. An ident daemon needs to be able to read kernel memory. Why? Think about it. You connect from host A port a to host B port b. Host B then connects to host A port 113 and says "What is the name of the user who connected to me from port a?" A normal userspace daemon has no way to answer this question. It needs to go poking around inside kernel memory in order to learn who has bound to port a.
In addition to this, I'd like to point out another problem associated with identd: firewalls. Right now, I'm at work, behind a firewall. When I connect from my workstation to a remote service, it might wish to connect back to me for purposes of querying my ident service. But it can't -- port 113 is blocked at the router. Even if I wanted to run ident (which I don't), I wouldn't be able to respond to that query, because I never receive it. Eventually the request times out. This means that connecting to any service that blocks awaiting an ident response has a built-in delay. And if the ident response is mandatory, it means I'm completely shut out.
At home, I just run nullidentd. I return the string rfc1413 for the response. Most people don't react to that, but I've had a few people ask "Which RFC is that?", and one or two have complimented me on it.
gnome-panel version 1.4.x works on my setup. gnome-panel 2.x does not, because it uses this gconf2 thing. (And as you can see, the Debian maintainers for gconf2 are a bit less than helpful in this situation.)
My NFS server is OpenBSD. I've configured NFS version 3 support on the Linux client. Locking just does not work. I've spoken with another Debian+OpenBSD user in IRC and he's confirmed that locking does not work. I consider the problem intractable -- I just ssh to the OpenBSD box to run mutt instead of trying to run mutt on the Debian client.
Hahahahah! Oh yeah, I can see it now. GNOME 2, which totally and utterly refuses to work if your $HOME is NFS-mounted, will become the default for Solaris, from Sun, the people who invented NFS.
This should be amazingly funny to watch. I'll have to stock up on popcorn.
08:58 greycat> ~laugh at gnome 208:58 @apt> HAHAHA! AH-HAHA! gnome 2 just cracks me up!
I visited the site with Mozilla 1.1 on HP-UX 10.20 (which I am also using for Slashdot). The front page consists of some text and a single animated-GIF button in the middle of the page. When I click the button I get a more complex page with a login box on the right, a navigation menu across the top, etc. It looks like a fairly normal web site. I saw no pop-ups, and I wasn't called any rude names or told to go away, etc.
Oh, did I mention that I have Javascript turned completely off? Really, Javascript serves no legitimate purpose. It is solely a tool of malice.
This is the fourth time I've downloaded a Phoenix build, and the second time that running it does nothing visible. My machine is a Pentium II running Windows 95. (I think it's Win95B. Not sure. "VER" says "Windows 95. [Version 4.00.1111]".)
....)
PHOENI~1.ZIP 8,691,111 09-26-02 4:01p phoenix-win32.zip
PHOENI~2.ZIP 7,806,796 10-04-02 9:47a phoenix-0.2-win32.zip
PHOENI~3.ZIP 7,360,073 10-16-02 9:55a phoenix-0.3-win32.zip
PHOENI~4.ZIP 7.396,544 10-30-02 8:14a phoenix-0.4-win32.zip
The first one (a nightly build somewhere around 0.1) and the last one (0.4 milestone) do not run. They both give me the same behavior: after typing "PHOENIX", the mouse cursor changes to the arrow-plus-hourglass for a few seconds, then changes back to normal. There is no new task in the task manager after that. For all visible purposes, phoenix crashed with no errors and no core file. (Of course, there's no core file. This is Windows. Why would we be able to debug it? No strace, no gdb, no truss,
Phoenix 0.2 and 0.3 both run perfectly well on this same computer. So I've gone back to 0.3.
"PHOENIX -ProfileManager" also does nothing. Same symptoms. I've even tried deleting the 0.3 profiles using Phoenix 0.3's ProfileManager before running Phoenix 0.4. No change.
middle-click on links to open them in a new tab [...] Now I'd just like the same behavior on form buttons
Drooool. I've been wishing for that for about 4 years now. (I might actually participate in more Slashdot polls if I didn't have to destroy my copy of the main page to do so.)
And why oh why can't I hit ESC to stop animated GIFs? Netscape 4.x can do this. Debian's versions of Mozilla can do it. Yet upstream Mozilla, for some reason, cannot. (And I can't get Phoenix 0.4 to run, so....)
Where is the hardware going to come from to build these Freenet archives? Is it just going to fall out of the sky?
It is already here. It's provided by the users of Freenet.
But you failed to notice that writings in these categories published anonymously would be basically of no value to anyone.
The second category in which I used the word "uncensorable" was opinions. Opinions are of value to whomever values them. Period.
And the first category was "criticism" (of various authoritarian or pseudo-authoritarian organizations). Now, you're implying that expression of criticism is of no value unless you know who wrote it. I disagree -- I think that an anonymous tip is of enormous value. Imagine what would happen if the police never investigated anonymous tips. Imagine the opposite scenario, in which an anonymous publication describes corruption inside a government office and gives details. Those details can be checked and amplified by regular journalists once the story has been exposed.
But in a Freenet world, where anybody with a computer could publish anything, the lies would soon vastly outnumber the truths. Freenet would be a vast collection of bullshit with a few pearls buried in it.
You've just described the entire world -- books, magazines, newspapers, television shows, radio shows, world wide web sites, Freenet sites, slashdot. All of it.
If you can't handle the burden of analyzing written material yourself to see whether it's of value to you, then go ahead and continue in your little spoon-fed world where only the texts on the Big Brother Approved List (To Save The Children) are worth reading. But for me, I'll take freedom of choice, thank you.
I eagerly await your superior layman's explanation, Mr. Coward. I'd really love to hear it. I'm sure Ian Clarke would love to hear it as well. He could add it to the sorely lacking Freenet newbie documentation.
It sure would have been nice for that gpl outfit to make certain their program would run with the current kaffe stable series, eh?
I run a node on OpenBSD 3.1 i386 using Kaffe 1.0.7 with a couple asserts commented out. It only crashes once or twice a week at most.
mjr claims that he has never had his Freenet node crash, and he's running Kaffe CVS from 20010902.
hobx claims that he got Freenet to run on Kaffe CVS head with only one small patch to Kaffe. (He didn't say whether it would stay running.)
All Java implementations suck. Your job is to find one that sucks as little as possible in your environment.
(I am a lifetime member of the "I Hate Java" club. If Freenet didn't exist, or wasn't written in Java, I would purge the evil Java from all of my computers, while capering with a demented glee. If nothing else, my months of experience with Freenet have taught me just how bad -- and how unportable -- Java is.)
Oh yeah, because searching freenet for "deep throat" is sure to give you secret information right at the top of the list.
You can't search Freenet. You surf it, the way you surf the World Wide Web when you don't have Google. (Don't any of you remember the WWW before search engines?) Then you bookmark the good sites you find.
If you want to search Freenet, then you'll have to implement a Google of your own. Spider it, then index everything you find, then search your index. I know at least one person is working on this right now.
Correct me if I'm wrong, but the only way to find something on freenet is to search for it.
Actually, you can't "search" as most people use that term. You retrieve documents from Freenet by specifying their key. You have to learn the key somehow, usually from another Freenet document.
The whole point in whistle blowing is that nobody knows about whatever you're uncovering. If they don't know about it, how can they search for it?
Some of the popular Freenet site authors have a way to send them messages using KSK@ keys. This is normally used for Freesite submission -- for example, it's how TFE learns about new Freenet sites so that he/she can list them.
So if I were going to do some whistle-blowing, I'd create my Freenet site (could even be a single text file), insert it into Freenet as a one-shot or edition site (certainly not a DBR), and then submit the key to TFE's submissions bin. And possibly a few other Freesite authors' submissions bins as well.
FCP is an ASCII protocol - very easy to use.
Almost. It's ASCII after the first four bytes, but for some reason that I don't quite grasp, the Freenet developers decided that you would have to send \000\000\000\002 at the beginning of each FCP request.
pegasus:~$ echo -e '\000\000\000\002ClientHello\nEndMessage' | nc 127.0.0.1 8481
NodeHello
Protocol=1.2
HighestSeenBuild=6
Node=Fred,0.5,1.46,524
EndMessage
It's open source, and written in java. It would take all of 30 seconds to port.
You apparently don't have much experience with Java. (Hint: try searching the Freenet development mailing list archives for "Heisenbug".)