OK. So lets get to the meat of the argument, which is that message passing micokernels are slow by design and still prone to failure if a critical userland device driver process dies.
Slow... yes, I suppose you will always take a performance hit.
If a userland device driver dies; in that case the microkernel would continue to run just fine... I'd say it's entirely possible to design the system so that all critical services could be restarted. But I'm being a smartass.
Further, please show me how security is improved. If a userland process (say a shell) elevates privileges to root, how is this any different from a monolithic kernel based system?
Well, if you have a "root" user that can exert absolute control over the system, then - surprise! - you're screwed. How does a monolithic kernel protect you from a privilege escalation exploit in the shell?
Possible MK advantage, though:
I am not a security expert, but isn't the idea of having a super user considered A Bad Idea nowadays? Isn't more fine-grained security the way everyone is heading?
If so, it seems reasonable to use those "newfangled" techniques (i know, they're not new) to control access not only to applications on the system, but to the system servers themselves.
That privilege escalation bug in the shell? Still there, except now you can have hard separation between servers intended to interact with users, and critical system servers. And the only privilege you can escalate to is "user is able to start/stop the GUI and network". I'd consider that an improvement.
Feel free to tell me why this wouldn't work. Like I said, security is not my area of expertise.
If the disk subsystem crashes, how exactly do you get at the on disk program to start it with? Are you going to try to restart it without actually reloading it? Would you trust it?
Well, duh, you load the "bootstrap" disk subsystem from ROM, obviously. Then you can load the "real" (full-featured, with permission checking, journalling) disk subsystem; start that up, and deactivate the bootstrap system.
In a manner of speaking, you are rebooting. Except that that single really important process doesn't have to be brought down, even for a second.
The question remains, what is the value in running a system with one of its basic subsystems (disk I/O in the above example) crashed? [...] If the disk subsystem of an OS craps out, well, thats bad. You're probably going to have to reboot it anyway and you certainly don't want the rest of the system to run as if nbothing happened because you have no idea what the disk subsystem did or did not write to disk before it crashed. Continuing as if nothing happened would be disasterous.
Why wouldn't I want to keep the rest of the system running as if nothing happened? Surely, that would be the ultimate goal for any system that needs to be available all the time. "You're probably going to have to reboot it anyway" is simply not true if you can restart all the parts completely independantly from each other.
Obviously, you would have to keep in mind that the assumptions have changed. It would no longer be acceptable for a critical program to crash if the whole file system was yanked out from under it.
Regarding phones, it's probably a bigger change that the word has changed meaning.
(It used to mean "that thing that you plug into the wall that lets you talk to other people far away", for you kids. Yes, outlets in the wall, at your home.)
I wonder how you'd explain todays über-gadgets to someone from the eighties. "This? Oh, it's my.. um, tricorder. Yeah, that's it."
So radioactive isotopes occur naturally, everywhere. Okay. Fine.
But why do you think that surgically implanting a foreign object is the only way we are going to be affected by them? As far as I can see, they are part of our natural environment, and we are adapted to dealing with that.
It's like with oxygen, normally not something you'd think would be very nice to have around (because it's so damn reactive).. but, well, we're adapted to it, and can't even live without it.
Living organisms take up radioactive isotopes all the time. Carbon-14 anyone?
You might have a really good reason to worry about alpha emitters, but you don't make a very good argument.
What you say is true. But that still means it will take lots of effort writing emulators, bypassing copying restrictions, etc. It's far better making sure today people won't have to do all that tomorrow.
Which means preserving the emulators (and cracks) we have now, and actually encouraging people to write emulators too. Fortunately, that sort of thing has mostly taken care of itself up until now, but what about the coming generation of consoles, for example?
Course a little nudge nudge wink wink could do wonders with getting them to write Linnox drivers.
I'd hope they publish the specs instead, and let the proverbial million monkeys with keyboards hammer out the drivers. Being open is how they could make a difference; if "having drivers" is all that matters Linux users might as well go with the bigger players.
And it might mean I can finally get a card that is well supported. It's not all Linux, you know.
Am I the only person that thinks piping objects between processes instead of an raw byte stream sounds very, very awesome?
No, it's just that the rest of us got turned off by the first few pages of "but the console windows aren't transparent" and "backslash instead of slash for a pathname separator SUCKS!".
And modding all those posts off topic would drain the supply of mod points so quick the whole universe would disappear in a puff of division-by-zero, which is too bad.
"i play a pivotal role in a grand conspiracy to cripple the free software movement from within, by covertly embedding an unnecessary, yet seductively useful, patented technology in the very heart of the linux operating system's second most popular desktop environment."
I think it's about time that the EU did something about all these nonsense TLDs. (the EU being the only part that could reasonably pull it off) They could say "Ok, com, org, net, int, mil, edu and gov are okay, plus country TLDs, but the rest is just bullshit, and we're not going to consider them canonical any more" and introduce a bunch of TLDs of their own, and push them hard.
Wouldn't that be fun? EU ISPs would probably end up supporting all of them anyway, and the main result would be to point out that the DNS is just an arbitrary agreement. (Want a new TLD? Create it yourself and tell your friends!)
Yeah, it probably wouldn't lead to much, but at least maybe DENIC could get to run.net after the dust settles? And maybe, just maybe, the people currently in charge would start to see the DNS as a piece of critical infrastructure, not as their own personal cash cow.
Personally, I'd like to get rid of anything but country-specific TLDs too, with the obvious exceptions of.int and.arpa, but I guess that'll never happen...
To register in the.int domain, the applicant must be an intergovernmental organization that meets the requirements found in RFC 1591. In brief, the.int domain is used for registering organizations established by international treaties between or among national governments. Only one registration is allowed for each organization. There is no fee for registering an.int domain name.
Just look at that! Sheesh. No fee? No wonder all the spam comes from.int.
I took the time to pick out some misspellings in the comments. Here's my
findings:
"There are two major pc builders in China. Lenevo and..."
"The experience of getting my Lenevo ThinkPad was actually much lamer,..."
"As I said, I've got a Levano and not an IBM version."
"... This was before Leveno entered the picture."
"If someone checked out an HP and a Lonovo and liked the HP that's cool,..."
"The Levano models come out, they're supposedly the same...."
"Lonovo: Get your act straight or..."
"We moved from Dell Laptops to Lenova... (Score:1)"
"SHOCKER:Lenova not as well known or trusted as IBM (Score:1)"
I completely agree on wanting to have more modifier keys. At least the first part, less "crazy keys" seems to become more common, with some newer keyboards losing PrintSc/SysRq, ScrollLk, Pause/Break.
But if you want to get rid of some keys and have some new ones too, you could look into remapping those you don't want to doing something useful. It's certainly possible on Unix-like operating systems, and I'd imagine there's Windows software to do all sorts of weird things too. Sure, using Scroll Lock as a shift/modifier key would be awkward, but I suppose you could remap "the third Windows key", and place it in some unused corner instead. Or something. Num Lock is something you could probably get rid of too - when did you last use the keypad for cursor movement? About the only key you can't remap is Print Screen/SysRq, I believe.
I've been thinking about this a bit.:-)
Of course, I want something a bit more like this: Emulator 122. (Which also brings us back to the topic of IBM, and indestructible hardware.)
But the biggest use I find for it is as a shortcut key. It's a blank key with relatively few bindings (since it's technically optional) that you can bind to your heart's content. Every program I use day-to-day is bound to a windows+[key] combo, and it's incredibly convenient. I could use the typical Control+Alt+[key], but there's a lot of conflicts with that and other programs. Then there's Control+Alt+Shift+[key], still some conflicts and now a lot less convenient. But the windows key is a wide open field.
It's wide open only until people start using it, and a de-facto or operating system standard starts setting out what you can use it for. Typewriters didn't have all the junk that a modern keyboard has. So why all these Alt/Ctrl/AltGr/NumLock/SysRq etc keys? Because at some point in time we needed (or, "needed") them to do some specific task, and no other key was available.
That being said, the Windows keys are probably less likely to be "used up" in the same way, especially if it becomes a standard having them as app launchers.
For those that don't like the win key, is it because you don't use it this way? Or would you still hate it even if you used it as a bind key?
It's emotional. All old DOS gamers remembers accidentally hitting the Windows key and ending up looking at the Windows 95 desktop instead of firing, and losing mental focus in the process.:-)
And the rest of us don't like the keys because they clutter the spacebar row and make it asymmetrical.
If bandwidth shaping is here to stay, as many posts here indicate, then use your software to work around it.
No, don't. If you deny ISPs the opportunity to limit traffic by protocol they'll simply limit traffic by user instead. So you'll end up not with slow torrents, but with slow everything -- web browsing, checking your email, SSHing home from work/school, whatever.
It's better to try to get some competition into the market, and by favouring the ISPs that are open with what they do. A previous poster working at an ISP mentioned giving P2P traffic a lower priority (but still letting it use all bandwidth), and that's not unreasonable. You want VoIP, SSH, email, and even HTTP to take precedence anyway.
On the other hand, if ISPs are being unreasonable and try to squeeze "bad" traffic out entirely, then by all means work around it. But they'll find a way to limit your bandwidth sooner or later. And there's no shortage of companies willing to provide technical solutions to their problems...
I have had to resort to blocking popular bittorrent ports on my linksys router just to keep the 5MBit cable connection from choking. Once the connection is close to being saturated, _nothing_ works because too many packets are getting lost or timing out.
On an ISP scale, you _never_ want to get to the point where you are using 100% of your bandwidth, because the network will slow down to a crawl.
Wait... Didn't we all just agree, a couple of threads back, that Ethernet (high burst speed, good common case, cheap hardware) was Good Stuff; and that Token Ring (deterministic behaviour, good worst case, expensive) was Teh Suxorz? Did I miss a memo or something?
(Yeah, I was going for "Funny" there. Sorry.)
On a more serious note, it sounds like your router isn't really up for the job.
Anyway, what I was going to say was that the scenario with lots of peer-to-peer bandwidth "inside" your ISP and lower (choked) bandwidth to the outside is a reasonably good parallel to that original scenario. So your up/download rate is limited. So what? As long as you still can have a decent speed to your closer peers, you're fine. You might get crappy speed to some faraway peers, but you and your closer peers can effectively pool your collective bandwidth and end up with something pretty decent anyway.
Lets say everyone want to download the just released new movie^H^H^H^H^Hdistro. Instead of N number of customers each using FTP/HTTP and sucking up as much of the ISPs outside bandwidth as they can, slowing it down for everyone to (outside bandwidth/N); or N number of customers peering equally within the ISP network and outside it, generating (outside peers * peer speed) amount of traffic, saturating your uplink, you create a big incentive for "local" peers to exchange data with each other, and only rarely going outside -- just like the Bittorrent protocol wants you to do. What's the problem?
Of course, I don't know how ISPs operate, but it seems to me that it wouldn't be at all hard to have a system that is both fair to all users and pretty fast for torrent leechers, if that is what you want. And yeah, it (traffic shaping) obviously opens up some evil possibilities too, depending on who owns the ISP, and what else they have to sell.
Actually with a large enough ISP network Bittorrent ain't all that bad, since you're likely to find many peers within the ISP. It's not local bandwidth that's scarce, it's the uplinks that are strangled [...]
Bittorrent generates lots of traffic yes, but the only difference now is that the traffic is coming from all over the place.
There is one main difference with Bittorrent, which is maximizing the total bandwidth. In my previous example, if 100 people downloaded the same file from an FTP server, the combined speed of all transfers was limited by that FTP's uplink, i.e. 10mbit, and everyone got a small slice of that bandwidth so it took longer to finish the transfer. Bittorrent does the opposite, while the initial seeder might only have 10mbit available, there are 99 other peers with anywhere from 1 to 5 mbit each, yielding an aggregate swarm speed that is several times faster than the FTP host could put out.
So, isn't bandwidth throttling via QoS actually a good thing? Why doesn't people realize that?
I mean, this is the sort of situation that Bittorrent is actually designed to handle. One uploader, who has his own website, but a somewhat limited bandwidth, decides to use Bittorrent for a few big files. He sets up a tracker and creates.torrent files. Now when someone starts downloading one of those files, he'll get a download speed equalling at least the site owner's maximum upload speed, or more if there are others with partial copies.
(All the nonsense today about "keep your client open", and "always seed at least to a 1:1 ratio" are really results of applying Bittorrent to a scenario it was never really designed to handle. The original goal was to ease the load on the server, and nothing else. Speed is just a bonus.)
Interesting. Maybe you should look into getting a Swedish bank, they seem to have stopped with that nonsense now.
Or more realistically, an FF extension to change the User-agent string. My bank (Föreningssparbanken) used to lock me out before, but with an extension that was quickly fixed. Then they had a period of putting up a warning instead ("We can not guarantee the security of..." -- yeah right..) but now it's no problem. They even keep track of new versions and tell users they might want to upgrade, at least for Opera.
And the marginalization of other browsers? Her argument basically runs that other browsers don't stand a chance against IE's installed base, while conveniently overlooking the fact that IE itself was once an "other" browser and citing ways that IE got the leg-up on Netscape without ever noting that those other browsers are doing the same things to IE. The argument basically runs "Yes, things changed in the past, but things will remain as they are now because they're the way they are now." Buh?
Not really; her argument is that back then, only a small part of the userbase even had a browser installed, and MS could make a good offer to all the ones who didn't. Things are different now, when everyone already has a browser. (Why bother with getting a deluxe browser when what you've already got works for you?)
So.. for hydrogen as fuel, there's going to be more energy consumed in its making, than would be available to be extracted by the car. And, there are many studies that indicate that ethanol has a similar problem.
The only near term way either works is either by using coal plants or nuclear plants to produce it.
Or just "plants". As in green plants, processed to produce ethanol.
Somewhat off topic, but anyway:
Where I live (Sweden) there are seemingly lots of just-starting-up ethanol production plants along with lots and lots of ethanol buses in our towns. And the number of "green" cars is steadily rising, many of which run on the stuff. (Tax breaks and exemption from congestion charges being the main incentives.)
I realize it's a poor argument, but if there was something fundamentally wrong with using ethanol as a medium- or large-scale alternative to fossil fuels I think someone would have pointed it out by now. There is the occasional criticism; it's are mostly "but will it be competitive with ethanol made in Brazil from sugar canes?", though.
On the other hand, I bet it's a lot more expensive than refining the black stuff that comes out of the ground. Fortunately for us, our ~= 200% taxes has prepared us for that...
Hm, Jeroen Tel's excellent tune "Alternative Fuel" is coming up on Kohina as I'm typing this. Coincidence? Of course not. I guess I'll have to post this now.
Slashdot Mirror
If a userland device driver dies; in that case the microkernel would continue to run just fine... I'd say it's entirely possible to design the system so that all critical services could be restarted. But I'm being a smartass.
Well, if you have a "root" user that can exert absolute control over the system, then - surprise! - you're screwed. How does a monolithic kernel protect you from a privilege escalation exploit in the shell?Possible MK advantage, though:
I am not a security expert, but isn't the idea of having a super user considered A Bad Idea nowadays? Isn't more fine-grained security the way everyone is heading?
If so, it seems reasonable to use those "newfangled" techniques (i know, they're not new) to control access not only to applications on the system, but to the system servers themselves.
That privilege escalation bug in the shell? Still there, except now you can have hard separation between servers intended to interact with users, and critical system servers. And the only privilege you can escalate to is "user is able to start/stop the GUI and network". I'd consider that an improvement.
Feel free to tell me why this wouldn't work. Like I said, security is not my area of expertise.
In a manner of speaking, you are rebooting. Except that that single really important process doesn't have to be brought down, even for a second.
Why wouldn't I want to keep the rest of the system running as if nothing happened? Surely, that would be the ultimate goal for any system that needs to be available all the time. "You're probably going to have to reboot it anyway" is simply not true if you can restart all the parts completely independantly from each other.Obviously, you would have to keep in mind that the assumptions have changed. It would no longer be acceptable for a critical program to crash if the whole file system was yanked out from under it.
(It used to mean "that thing that you plug into the wall that lets you talk to other people far away", for you kids. Yes, outlets in the wall, at your home.)
I wonder how you'd explain todays über-gadgets to someone from the eighties. "This? Oh, it's my .. um, tricorder. Yeah, that's it."
But why do you think that surgically implanting a foreign object is the only way we are going to be affected by them? As far as I can see, they are part of our natural environment, and we are adapted to dealing with that.
It's like with oxygen, normally not something you'd think would be very nice to have around (because it's so damn reactive) .. but, well, we're adapted to it, and can't even live without it.
Living organisms take up radioactive isotopes all the time. Carbon-14 anyone?
You might have a really good reason to worry about alpha emitters, but you don't make a very good argument.
Which means preserving the emulators (and cracks) we have now, and actually encouraging people to write emulators too. Fortunately, that sort of thing has mostly taken care of itself up until now, but what about the coming generation of consoles, for example?
And it might mean I can finally get a card that is well supported. It's not all Linux, you know.
(Sorry for the off topic post, but someone had to say it.)
And modding all those posts off topic would drain the supply of mod points so quick the whole universe would disappear in a puff of division-by-zero, which is too bad.
User page: http://slashdot.org/comments.pl?sid=164643&cid=13
Wouldn't that be fun? EU ISPs would probably end up supporting all of them anyway, and the main result would be to point out that the DNS is just an arbitrary agreement. (Want a new TLD? Create it yourself and tell your friends!)
Yeah, it probably wouldn't lead to much, but at least maybe DENIC could get to run .net after the dust settles? And maybe, just maybe, the people currently in charge would start to see the DNS as a piece of critical infrastructure, not as their own personal cash cow.
Personally, I'd like to get rid of anything but country-specific TLDs too, with the obvious exceptions of .int and .arpa, but I guess that'll never happen...
The United Nations
The European Union
NATO
Interpol
World Health Organization
International Civil Aviation Organization
The International Telecommunications Union
The Red Cross
I don't know what to say about this one though:
International Network for Bamboo and Rattan (INBAR)
And more: Google it
Not to mention the sloppy rules for registration:
Just look at that! Sheesh. No fee? No wonder all the spam comes from"There are two major pc builders in China. Lenevo and ..." ..." ..." ..." ..."
"The experience of getting my Lenevo ThinkPad was actually much lamer,
"As I said, I've got a Levano and not an IBM version."
"... This was before Leveno entered the picture."
"If someone checked out an HP and a Lonovo and liked the HP that's cool,
"The Levano models come out, they're supposedly the same.
"Lonovo: Get your act straight or
"We moved from Dell Laptops to Lenova... (Score:1)"
"SHOCKER:Lenova not as well known or trusted as IBM (Score:1)"
Counts:
Lenevo: 1 (same poster twice)
Leveno: 1
Lonovo: 2
Lenova: 2
Levano: 3
Lenovo: Lots.
Lenovo clearly needs to work on its brand. For comparison, I did not once see someone misspell "IBM".
But if you want to get rid of some keys and have some new ones too, you could look into remapping those you don't want to doing something useful. It's certainly possible on Unix-like operating systems, and I'd imagine there's Windows software to do all sorts of weird things too. Sure, using Scroll Lock as a shift/modifier key would be awkward, but I suppose you could remap "the third Windows key", and place it in some unused corner instead. Or something. Num Lock is something you could probably get rid of too - when did you last use the keypad for cursor movement? About the only key you can't remap is Print Screen/SysRq, I believe.
I've been thinking about this a bit. :-)
Of course, I want something a bit more like this: Emulator 122. (Which also brings us back to the topic of IBM, and indestructible hardware.)
That being said, the Windows keys are probably less likely to be "used up" in the same way, especially if it becomes a standard having them as app launchers.
It's emotional. All old DOS gamers remembers accidentally hitting the Windows key and ending up looking at the Windows 95 desktop instead of firing, and losing mental focus in the process.And the rest of us don't like the keys because they clutter the spacebar row and make it asymmetrical.
It's better to try to get some competition into the market, and by favouring the ISPs that are open with what they do. A previous poster working at an ISP mentioned giving P2P traffic a lower priority (but still letting it use all bandwidth), and that's not unreasonable. You want VoIP, SSH, email, and even HTTP to take precedence anyway.
On the other hand, if ISPs are being unreasonable and try to squeeze "bad" traffic out entirely, then by all means work around it. But they'll find a way to limit your bandwidth sooner or later. And there's no shortage of companies willing to provide technical solutions to their problems...
(Yeah, I was going for "Funny" there. Sorry.)
On a more serious note, it sounds like your router isn't really up for the job.
Anyway, what I was going to say was that the scenario with lots of peer-to-peer bandwidth "inside" your ISP and lower (choked) bandwidth to the outside is a reasonably good parallel to that original scenario. So your up/download rate is limited. So what? As long as you still can have a decent speed to your closer peers, you're fine. You might get crappy speed to some faraway peers, but you and your closer peers can effectively pool your collective bandwidth and end up with something pretty decent anyway.
Lets say everyone want to download the just released new movie^H^H^H^H^Hdistro. Instead of N number of customers each using FTP/HTTP and sucking up as much of the ISPs outside bandwidth as they can, slowing it down for everyone to (outside bandwidth/N); or N number of customers peering equally within the ISP network and outside it, generating (outside peers * peer speed) amount of traffic, saturating your uplink, you create a big incentive for "local" peers to exchange data with each other, and only rarely going outside -- just like the Bittorrent protocol wants you to do. What's the problem?
Of course, I don't know how ISPs operate, but it seems to me that it wouldn't be at all hard to have a system that is both fair to all users and pretty fast for torrent leechers, if that is what you want. And yeah, it (traffic shaping) obviously opens up some evil possibilities too, depending on who owns the ISP, and what else they have to sell.
I mean, this is the sort of situation that Bittorrent is actually designed to handle. One uploader, who has his own website, but a somewhat limited bandwidth, decides to use Bittorrent for a few big files. He sets up a tracker and creates .torrent files. Now when someone starts downloading one of those files, he'll get a download speed equalling at least the site owner's maximum upload speed, or more if there are others with partial copies.
(All the nonsense today about "keep your client open", and "always seed at least to a 1:1 ratio" are really results of applying Bittorrent to a scenario it was never really designed to handle. The original goal was to ease the load on the server, and nothing else. Speed is just a bonus.)
So when a thousand geeks unexpectedly show up, that means you've been slashmobbed?
Or more realistically, an FF extension to change the User-agent string. My bank (Föreningssparbanken) used to lock me out before, but with an extension that was quickly fixed. Then they had a period of putting up a warning instead ("We can not guarantee the security of ..." -- yeah right..) but now it's no problem. They even keep track of new versions and tell users they might want to upgrade, at least for Opera.
But there's apparently a good chance it will run, maybe even with native EFI support
I wish I had literature points right now...
Somewhat off topic, but anyway:
Where I live (Sweden) there are seemingly lots of just-starting-up ethanol production plants along with lots and lots of ethanol buses in our towns. And the number of "green" cars is steadily rising, many of which run on the stuff. (Tax breaks and exemption from congestion charges being the main incentives.)
I realize it's a poor argument, but if there was something fundamentally wrong with using ethanol as a medium- or large-scale alternative to fossil fuels I think someone would have pointed it out by now. There is the occasional criticism; it's are mostly "but will it be competitive with ethanol made in Brazil from sugar canes?", though.
On the other hand, I bet it's a lot more expensive than refining the black stuff that comes out of the ground. Fortunately for us, our ~= 200% taxes has prepared us for that...
Hm, Jeroen Tel's excellent tune "Alternative Fuel" is coming up on Kohina as I'm typing this. Coincidence? Of course not. I guess I'll have to post this now.