You seem to be contradicting yourself. Yes, of course we should be much more concerned with a nuclear war (or biological war or something with similarly global effects; and more and more scientists believe that it's too late to deal with the global warming anymore - we might have less than 10^2 here, not even dreaming about 10^3) than with big rocks flying by. But that doesn't make the "all the eggs in one basket" less valid, the very opposite - it enforces them significantly, pointing out a much more clear and realistic motivation.
Just try to be a little more imaginative and substitute the asteroid with any global disaster. The proximity is suddently not very low. Let's go for the rescue, perhaps the trick to deal with the beingkind evolution pitfalls is to cheat them instead of avoiding them.
It's not like the searching of the database in a regular library would take more than a second (and I assume the robot will have to have some good hardware inside anyway in order to be able to do all the movement calculations in real time, OCR fast etc). Still, I can agree the "manual navigation" mode would be useful too.
That's called "firewall". NAT acts as a bad firewall, but you can run a firewall even without a NAT. And it makes sense to replace NAT with a simple firewall because you can then choose to firewall only some of the nodes, you do not have problem with overloading the NAT with too many TCP connections, it is easier to allow P2P stuff through a firewall than through a NAT. And you lose no security.
Thanks to the autoconfiguration etc, IPv6 makes the renumbering much easier.
And what does it involve except simply switching the network part, then?
Also, if you discover a good scalable way to assign IP ranges directly to the end sites instead of ISPs, please tell us - (AFAIK) noone before you did.
No, (AFAIK:) Christians say that people cannot ever live a perfect life, the original sin causes everyone to make a sin - at least once. They cannot live a sinless life (with the exception of those who die in the baby age).
Ad Enoch, Christians believe that everyone must die, and that he will come back to the Earth at the end of days (and, well, presumably sin). Well, they at least believe Elias will return, I'm not 100% sure by Enoch now but I think they do believe he will return too.
Disclaimer: IANAT (I'm not a theologist), also these beliefs can vary between churches.
Well, SpamAssassin just failed on me two weeks ago - suddenly its effectivity dropped from a fair 95%-98% to less than 50% and spam started pouring into my mailbox. It already happenned once but upgrading to a new version with newer rulesets helped, but this time there was nothing to upgrade to. I suppose some new kinds of spam started hitting me, but there were no obvious rules for me to catch these. So I tried dspam, trained it for three days and it quickly raised to about these 98% and is still improving fast. It is also much faster and provides a rather nice quarantinebox CGI interface in the box. The only problem I still have with it is how hungry is it about the disk space. After 14 days of usage (5000 innocent mails, 2000 spams) my personal dspam dictionary + sig database raised over 135M, if it will raise this fast how big will it be in two months? I yet need to solve that but I'm sure there will be a way out - I plan to try other storage methods or getting that segfaulting database purge tool to work.;-)
First, the headers are not encrypted and that is most probably what are you filtering against on the server (subject, from, to, x-mailing-list,...).
Second, spam usually does not go through encrypted, therefore you do not need to scan encrypted messages for mail. I agree that you have a point here though, spams will get encrypted when enough people will start using encrypted messages extensively, by harvesting the keys from some public keyservers.
But I suspect that this would just bring the keyservers to death, I think their infrastructure is not thick enough to carry a massive amount of queries at once. An idea coming on my mind is to let only people with registered PGP key to check out keys of others, and allow PGP key registration only through some "hardened" mechanism (ie. that infamous images with garbled text). Then you can easily implement various reasonable per-user throttling and so.
Then it will be the spammers' turn to find another way to cheat us.
And there is another thing - one of the very important advantages of the source-based distributions, I think.
The scenario: I maintain a text-mode web browser, with quite a rich feature set and many optional dependencies; but it can run in very thin installations just w/ libc as well, without a problem. However, ie. in Debian it depends on gpm, X11, lua, gnutls, and_so_on... And that kind of defeats its philosophy! (OTOH when you _have_ these packages, the enabled features can be very nice.)
I think it would be ideal to have a hybrid system - having several precompiled package mutations and if none matches the requirements, just get the source and compile it (if a compiler is installed). It should be fully automatic, except possibly for letting the user choose between compiling to fit the current set of fulfilled optional dependencies or following some of the dependencies.
Then you can go on and on, marking some apps to be "recommended to compile" when it could give a real performance boost with finetuned optimizations, etc.
THAT would be something! Well, I know, DIY. But didn't anyone do it already, anywhere? It shouldn't be very hard, after all, and ie. for Debian it could be a great boost.
Uh, why should I? I personally work in textmode not out of masochism, but because I can work more effectively there. Therefore I tend to use a web browser which isn't PITA to look at and actually use, like lynx - hey, you cannot do things while the document is loading! Awful. Well, I'm using ELinks most of the time;-).
Well, that's hard to answer if we don't know what games do you play under Windows. Some games run natively on Linux (ie. Unreal Tournament), and rather big number of other games runs under Wine or Winex. Do you want your Diablo, Baldur's Gate, Civ3, Neverwinter Nights, CounterStrike, FooBar? It works. Not everything works, but I'd say that majority of games does.
Re:Nah, Education is the Future
on
Linux in 2004?
·
· Score: 1
Interesting. I've installed Mandrake 9.1 on retailed OmniBook w/ some PII running at 300MHz and with 64M RAM (!), just out of interest how slow will KDE3 be there. The thing is, it is actually quite fast. Certainly a difference to Athlon 2.4GHz w/ 512M RAM, but it is definitively usable, and overally works smoothly. Even with the translucent menus etc (I didn't feel any noticeable slowdown).
Of course, I weren't even thinking about OpenOffice or Mozilla at all. Also, Gnome was unusably slow, but KDE3 just worked, worked smoothly and I had no reasons to complain.
Well, it's true that I don't mind konsole taking three or four seconds to start up, and I'm willing to wait 60 to 90 seconds to have the system booted up. (Then having it live ever after, or at least for n days until I take it to some longer travel and the batteries run out.)
What mask? In v6, the masks are described only by the bitcount. And most of the time, the network part of the address is 64 as far as you are concerned.
What gate? IPv6 supports router discovery - sort of plug'n'play. You set up the router to advertise itself on the ethernet segment, and then you just connect the computer to the ethernet. And that's it, you don't have to configure anything manually. It will automatically find the router, set it as its gateway and if the router is configured so, the machine will autoconfigure its IPv6 addy as well, gluing the network prefix it receives from the router with some simple transformation of the ethernet card hardware address.
What DNS addresses? Guess what, these can be advertised as well (although this one is not so widespread yet, and IIRC it is still in the process of standardization - IPv6 is still work in progress in some regards).
The more addresses thing is one part of it, but there is a lot more sweet stuff not so visible.
The change is actually happening. Slowly, silently, gradually, like all the articles/posts/blogs were actually suggesting. IPv6 _IS_ deployed between ISPs (_not_ only universities/academic networks) more and more widely in Asia (especially Japan, but I guess Taiwan or China as well), but lately also in Europe (from Finland through Netherlands to Austria). You can get native v6 connectivity to home in Japan or Netherlands. Most of the IXes are prepared for routing IPv6 traffic (or are already doing it). You can go from Czech Republic to France over IPv6 all-native, and it's actually faster than IPv4.
It's perhaps a bit worse with "the big corporations", but I think there are some movements visible as well, hopefully at least in Japan if not quite in Europe yet. It just all depends on the market demand. Asia is short on addresses, so it is leading in the IPv6 world, the press is the greatest. Europe is not so bad on it, but the movement is visible here as well. America feels comfortable with the addresses it has (mostly), so the motivation is not quite there (I know that IPv6 is not only about addresses, of course; it is the prime driving factor anyway, though). So you don't see much happenning in America. It is just at the other side of the cutting edge front, currently.
FYI, you can find specifications of open EEG equipment at http://openeeg.sf.net/. I didn't actually try it out myself (yet), but it looks quite decent and they say it is working nicely According to the specs, they are using 9V battery source and the actual electrodes-related circuits are optoelectrically separated from the computer.
This is kind of interesting --- when you will look where IPv6 was started to be adopted, first you will see Asia, mainly Japan. Then, slowly, Europe joined --- in fact from January on, things start to massively speed up here, a lot of providers decided at once that they want to try the thing out. Then there is North America, where somehow... well it doesn't seem that some remarkably wide IPv6 adoption is going on there.
The main reason is availability of IPv4 addresses (whole Japan has IIRC less than MIT, overally North America is where the addresses shortage is least apparent), but the side effect is that the centre of progress and cutting-edge front is moving from America to Asia and Europe. That is where probably the most of the further development is going to happen.
That resembles IPv6 router advertisment and node autoconfiguration to me. Actually you don't even need to select yes, just feed the router with a simple configuration (and it _is_ much less than 100 lines;-).
I think this article wasn't exactly focused on IP-layer networks, though.
I believe that this practice is not that good --- especially when the mail was NOT sent by a real human, but ie. mailing list subscription robot, some account creation verification mechanism sending you password by mail etc. Also you may easily annoy those who write you a mail by bothering them with this automation, it can be easily considered as impolite. And the real fun starts when two people with this filtering try to get in touch; you protect yourself from mails not generated by a real human, but at the same time you produce such mails yourself.
I think the middle path is the best one: automatically add the spam negatives to the approved sender db and send the verifications only to senders of mails identified as spams. You will get no false positives, in 99% of cases you will be able to even subscribe to a mailing list easily, and you won't bother the other people that much (and if you will specifically mention in the verifications that they are sent because the mail looked as spam, I think they will be much less angry at you).
About PCI modems, I don't think this will change dramatically.
I wonder what you mean by "the fucking cli" - what do you mean by booting directly into X? Lo, the/etc/inittab and the default runlevel entry, let 5 be it.
About switching kernels, it's not so trivial, but there's patch floating around LKML and suffering under Linus' uncharitable eye implementing a kexec() syscall, which should do basically what you want. Naturally you won't get the state (processes, sockets etc) preserved, but software suspend (already merged) can maybe help you here.
Well, the old LVM wasn't working at all in 2.5 kernel series at all - AFAIK the infrastructure all around gradually changed and noone stood up to keep LVM up to date. Thus, around the feature freeze there were three options - drop LVM, accept LVM2 or accept EVMS. It would be too bad to drop such a significant feature, and EVMS probably looked too complex to merge so lately or so.. it's not really clear to me, the decision was probably based on what Linus' lieutanants advised him, but at the end LVM2 ended up in the 2.5 series.
It would be maybe nice to have possibility in browsers to set whether referer should be actually sent and what should be sent inside. In fact, I know only about ELinks and Links now having this - you're free to set referer to "fake referer" (some string you're going to write down there is going to be sent), "normal referer" or referer containing URL of the page being loaded ("self-pointing referer";-). The "self-pointing referer" is set as default, and it helps workarounding most of the "protection" mechanisms, while effectively disclosing no possibly private information. Would be nice to see this in other browsers as well..
Strange that actually it looks the probably biggest IRC network (over 100 000 users in busy parts of day) (I'm not sure if dalnet isn't actually a bit bigger, but it really doesn't matter;) IRCnet is not very known. In fact it also includes *.fi, including irc.oulu.fi, so the very first core of IRC is the member of it as well, and this supports the claim of IRCnet to be original IRC network. It suffers from many problems as EFnet has as there's no chanserv or nickserv at IRCnet as well, however I think the DoS attacks aren't so frequent at IRCnet (altough they happens), even when it's larger than EFnet is.
There's story about The Great Split in '96 (EFnet vs. IRCnet) at http://www.irc.org/history_docs/TheGreatSplit.html
There's also IRCnet's homepage, surprisingly at http://www.ircnet.com/.
Slashdot Mirror
You seem to be contradicting yourself. Yes, of course we should be much more concerned with a nuclear war (or biological war or something with similarly global effects; and more and more scientists believe that it's too late to deal with the global warming anymore - we might have less than 10^2 here, not even dreaming about 10^3) than with big rocks flying by. But that doesn't make the "all the eggs in one basket" less valid, the very opposite - it enforces them significantly, pointing out a much more clear and realistic motivation.
Just try to be a little more imaginative and substitute the asteroid with any global disaster. The proximity is suddently not very low. Let's go for the rescue, perhaps the trick to deal with the beingkind evolution pitfalls is to cheat them instead of avoiding them.
It's not like the searching of the database in a regular library would take more than a second (and I assume the robot will have to have some good hardware inside anyway in order to be able to do all the movement calculations in real time, OCR fast etc). Still, I can agree the "manual navigation" mode would be useful too.
That's called "firewall". NAT acts as a bad firewall, but you can run a firewall even without a NAT. And it makes sense to replace NAT with a simple firewall because you can then choose to firewall only some of the nodes, you do not have problem with overloading the NAT with too many TCP connections, it is easier to allow P2P stuff through a firewall than through a NAT. And you lose no security.
Thanks to the autoconfiguration etc, IPv6 makes the renumbering much easier.
And what does it involve except simply switching the network part, then?
Also, if you discover a good scalable way to assign IP ranges directly to the end sites instead of ISPs, please tell us - (AFAIK) noone before you did.
No, (AFAIK :) Christians say that people cannot ever live a perfect life, the original sin causes everyone to make a sin - at least once. They cannot live a sinless life (with the exception of those who die in the baby age).
Ad Enoch, Christians believe that everyone must die, and that he will come back to the Earth at the end of days (and, well, presumably sin). Well, they at least believe Elias will return, I'm not 100% sure by Enoch now but I think they do believe he will return too.
Disclaimer: IANAT (I'm not a theologist), also these beliefs can vary between churches.
Ad Noah: This was at the start of the journey. Flip the page and read what happenned after the waters were gone. The wine affair etc.
Ad Zacharias and Elisabeth: The same thing.
Another example of this omit-context argumentation:
osgeek:9740177 "What a cool god you have!"
Glad you praise the God! (Although rather unconventionally....)
Well, SpamAssassin just failed on me two weeks ago - suddenly its effectivity dropped from a fair 95%-98% to less than 50% and spam started pouring into my mailbox. It already happenned once but upgrading to a new version with newer rulesets helped, but this time there was nothing to upgrade to. I suppose some new kinds of spam started hitting me, but there were no obvious rules for me to catch these. So I tried dspam, trained it for three days and it quickly raised to about these 98% and is still improving fast. It is also much faster and provides a rather nice quarantinebox CGI interface in the box. The only problem I still have with it is how hungry is it about the disk space. After 14 days of usage (5000 innocent mails, 2000 spams) my personal dspam dictionary + sig database raised over 135M, if it will raise this fast how big will it be in two months? I yet need to solve that but I'm sure there will be a way out - I plan to try other storage methods or getting that segfaulting database purge tool to work. ;-)
First, the headers are not encrypted and that is most probably what are you filtering against on the server (subject, from, to, x-mailing-list, ...).
Second, spam usually does not go through encrypted, therefore you do not need to scan encrypted messages for mail. I agree that you have a point here though, spams will get encrypted when enough people will start using encrypted messages extensively, by harvesting the keys from some public keyservers.
But I suspect that this would just bring the keyservers to death, I think their infrastructure is not thick enough to carry a massive amount of queries at once. An idea coming on my mind is to let only people with registered PGP key to check out keys of others, and allow PGP key registration only through some "hardened" mechanism (ie. that infamous images with garbled text). Then you can easily implement various reasonable per-user throttling and so.
Then it will be the spammers' turn to find another way to cheat us.
And there is another thing - one of the very important advantages of the source-based distributions, I think.
The scenario: I maintain a text-mode web browser, with quite a rich feature set and many optional dependencies; but it can run in very thin installations just w/ libc as well, without a problem. However, ie. in Debian it depends on gpm, X11, lua, gnutls, and_so_on... And that kind of defeats its philosophy! (OTOH when you _have_ these packages, the enabled features can be very nice.)
I think it would be ideal to have a hybrid system - having several precompiled package mutations and if none matches the requirements, just get the source and compile it (if a compiler is installed). It should be fully automatic, except possibly for letting the user choose between compiling to fit the current set of fulfilled optional dependencies or following some of the dependencies.
Then you can go on and on, marking some apps to be "recommended to compile" when it could give a real performance boost with finetuned optimizations, etc.
THAT would be something! Well, I know, DIY. But didn't anyone do it already, anywhere? It shouldn't be very hard, after all, and ie. for Debian it could be a great boost.
Uh, why should I? I personally work in textmode not out of masochism, but because I can work more effectively there. Therefore I tend to use a web browser which isn't PITA to look at and actually use, like lynx - hey, you cannot do things while the document is loading! Awful. Well, I'm using ELinks most of the time ;-).
Yes, at least early 2.0 kernels were vulnerable to the (in)famous Ping of Death
Well, that's hard to answer if we don't know what games do you play under Windows. Some games run natively on Linux (ie. Unreal Tournament), and rather big number of other games runs under Wine or Winex. Do you want your Diablo, Baldur's Gate, Civ3, Neverwinter Nights, CounterStrike, FooBar? It works. Not everything works, but I'd say that majority of games does.
Interesting. I've installed Mandrake 9.1 on retailed OmniBook w/ some PII running at 300MHz and with 64M RAM (!), just out of interest how slow will KDE3 be there. The thing is, it is actually quite fast. Certainly a difference to Athlon 2.4GHz w/ 512M RAM, but it is definitively usable, and overally works smoothly. Even with the translucent menus etc (I didn't feel any noticeable slowdown).
Of course, I weren't even thinking about OpenOffice or Mozilla at all. Also, Gnome was unusably slow, but KDE3 just worked, worked smoothly and I had no reasons to complain.
Well, it's true that I don't mind konsole taking three or four seconds to start up, and I'm willing to wait 60 to 90 seconds to have the system booted up. (Then having it live ever after, or at least for n days until I take it to some longer travel and the batteries run out.)
What mask? In v6, the masks are described only by the bitcount. And most of the time, the network part of the address is 64 as far as you are concerned.
What gate? IPv6 supports router discovery - sort of plug'n'play. You set up the router to advertise itself on the ethernet segment, and then you just connect the computer to the ethernet. And that's it, you don't have to configure anything manually. It will automatically find the router, set it as its gateway and if the router is configured so, the machine will autoconfigure its IPv6 addy as well, gluing the network prefix it receives from the router with some simple transformation of the ethernet card hardware address.
What DNS addresses? Guess what, these can be advertised as well (although this one is not so widespread yet, and IIRC it is still in the process of standardization - IPv6 is still work in progress in some regards).
The more addresses thing is one part of it, but there is a lot more sweet stuff not so visible.
Sue!
The change is actually happening. Slowly, silently, gradually, like all the articles/posts/blogs were actually suggesting. IPv6 _IS_ deployed between ISPs (_not_ only universities/academic networks) more and more widely in Asia (especially Japan, but I guess Taiwan or China as well), but lately also in Europe (from Finland through Netherlands to Austria). You can get native v6 connectivity to home in Japan or Netherlands. Most of the IXes are prepared for routing IPv6 traffic (or are already doing it). You can go from Czech Republic to France over IPv6 all-native, and it's actually faster than IPv4.
It's perhaps a bit worse with "the big corporations", but I think there are some movements visible as well, hopefully at least in Japan if not quite in Europe yet. It just all depends on the market demand. Asia is short on addresses, so it is leading in the IPv6 world, the press is the greatest. Europe is not so bad on it, but the movement is visible here as well. America feels comfortable with the addresses it has (mostly), so the motivation is not quite there (I know that IPv6 is not only about addresses, of course; it is the prime driving factor anyway, though). So you don't see much happenning in America. It is just at the other side of the cutting edge front, currently.
FYI, you can find specifications of open EEG equipment at http://openeeg.sf.net/. I didn't actually try it out myself (yet), but it looks quite decent and they say it is working nicely According to the specs, they are using 9V battery source and the actual electrodes-related circuits are optoelectrically separated from the computer.
This is kind of interesting --- when you will look where IPv6 was started to be adopted, first you will see Asia, mainly Japan. Then, slowly, Europe joined --- in fact from January on, things start to massively speed up here, a lot of providers decided at once that they want to try the thing out. Then there is North America, where somehow... well it doesn't seem that some remarkably wide IPv6 adoption is going on there.
The main reason is availability of IPv4 addresses (whole Japan has IIRC less than MIT, overally North America is where the addresses shortage is least apparent), but the side effect is that the centre of progress and cutting-edge front is moving from America to Asia and Europe. That is where probably the most of the further development is going to happen.
That resembles IPv6 router advertisment and node autoconfiguration to me. Actually you don't even need to select yes, just feed the router with a simple configuration (and it _is_ much less than 100 lines ;-).
I think this article wasn't exactly focused on IP-layer networks, though.
I believe that this practice is not that good --- especially when the mail was NOT sent by a real human, but ie. mailing list subscription robot, some account creation verification mechanism sending you password by mail etc. Also you may easily annoy those who write you a mail by bothering them with this automation, it can be easily considered as impolite. And the real fun starts when two people with this filtering try to get in touch; you protect yourself from mails not generated by a real human, but at the same time you produce such mails yourself.
I think the middle path is the best one: automatically add the spam negatives to the approved sender db and send the verifications only to senders of mails identified as spams. You will get no false positives, in 99% of cases you will be able to even subscribe to a mailing list easily, and you won't bother the other people that much (and if you will specifically mention in the verifications that they are sent because the mail looked as spam, I think they will be much less angry at you).
About PCI modems, I don't think this will change dramatically.
/etc/inittab and the default runlevel entry, let 5 be it.
I wonder what you mean by "the fucking cli" - what do you mean by booting directly into X? Lo, the
About switching kernels, it's not so trivial, but there's patch floating around LKML and suffering under Linus' uncharitable eye implementing a kexec() syscall, which should do basically what you want. Naturally you won't get the state (processes, sockets etc) preserved, but software suspend (already merged) can maybe help you here.
Well, the old LVM wasn't working at all in 2.5 kernel series at all - AFAIK the infrastructure all around gradually changed and noone stood up to keep LVM up to date. Thus, around the feature freeze there were three options - drop LVM, accept LVM2 or accept EVMS. It would be too bad to drop such a significant feature, and EVMS probably looked too complex to merge so lately or so.. it's not really clear to me, the decision was probably based on what Linus' lieutanants advised him, but at the end LVM2 ended up in the 2.5 series.
It would be maybe nice to have possibility in browsers to set whether referer should be actually sent and what should be sent inside. In fact, I know only about ELinks and Links now having this - you're free to set referer to "fake referer" (some string you're going to write down there is going to be sent), "normal referer" or referer containing URL of the page being loaded ("self-pointing referer" ;-). The "self-pointing referer" is set as default, and it helps workarounding most of the "protection" mechanisms, while effectively disclosing no possibly private information. Would be nice to see this in other browsers as well..
Take it as the place for those who will come and the main site will be already slashdotted :^)
Strange that actually it looks the probably biggest IRC network (over 100 000 users in busy parts of day) (I'm not sure if dalnet isn't actually a bit bigger, but it really doesn't matter ;) IRCnet is not very known. In fact it also includes *.fi, including irc.oulu.fi, so the very first core of IRC is the member of it as well, and this supports the claim of IRCnet to be original IRC network. It suffers from many problems as EFnet has as there's no chanserv or nickserv at IRCnet as well, however I think the DoS attacks aren't so frequent at IRCnet (altough they happens), even when it's larger than EFnet is.
l
There's story about The Great Split in '96 (EFnet vs. IRCnet) at http://www.irc.org/history_docs/TheGreatSplit.htm
There's also IRCnet's homepage, surprisingly at http://www.ircnet.com/.
Anyway, do anyone know what's up with Jarkko?