However, this doesn't work with bandwith. There is no free bandwith just laying around to be snatched up by anybody. You pay for your bandwith, websites pay for theirs, and if you use more bandwith, it costs more.
Well that was part of the question I was posing. You assert that everyone is paying their fair share. Perhaps not. Few people consume all the bandwidth they could and conversely the heavy users are getting a cheap ride since most (home) internet use is a flat fee that varies little (less than a factor of 2). So most folks don't pay more if you use more.
So this is indeed why as people's ability to easily consumbe bandwith rises and it becomes more prevalent we are going to have pressure on the resource to the point where if everyong used a little less then everyone would get faster service on the part they used. But of course as with the tragedy of the commons there's no incentive to reduce ones' personal use since there's not marginal cost for using more.
If indeed heavy users did pay more and that fully accounted for their use then indeed it should scale well. Of course now I raised the scaling issue. Does the internet cost strucure go up or down and people use more bandwith? Is it linear, super linear or sub linear. p2p is sub linear since the more routes off the backbone the less it should cost since the same hardware is more effectively ustilized. but backbone centric may be superlinear since now adding users means further distances too the back bone and then back out neccessitating more than a lineaar increase in infrastucture to add users.
We need to ask ourselves if this is a tragedy of the commons or a case where uniform access decreases costs or provides more public-goods. I don't know which it really is.
The tragedy of the commons is what happens when a resource is provided that lacks a proper mariginal cost for increased use. The classic example is private property versus unrstricted access to public grazing land. By charging a small price for admission per sheep to the land or by making it private, the incentive to overgraze it is removed and the total amount of meat sustainbly raise actually is higher. In this case if it's case where there is simply not enough baqndwith for everyone to do voip, and I don't pay any extra to do VOIP, then it's going to be over grazed and everyone gets a crappy connection. On the otherhand if the connection cost already is sufficient to expand the network to handle all the users that want voip or if we can prevent this from becomeing a power law network with critical links then it may be that the more users the better some sort of p2p works.
Thus another way of looking is this is that the thing we need to fear is too few corporation controlliing the internet and resulting in bottlenecks on backbones. In the long run to get high bandwidth we will need p2p that does not traverse a central backbone.
Assuming that the p2p scaling effect will not be sufficient and the tragedy of the commons wil happen then the way out is to have a pricing schedule. We can put that schedule on the users or on the content providers. the latter is what the backbone owners want since it means no net neutrality and control. The former would be better but I can imagine the cheap ass slashdotters used to paying a tiny sum for all-they-can eat internet won't like it.
I note that most vote talleying machines use windows software. By this I mean both the kiosk s you vote on and the central tabulation systems. In many cases the central tabulators are networked. In some cases the kiosks are.
The information is not classified, it's Official Use Only, which is a form of protected information. Personell records are usually, in part, execmt from freedom of information act requests, so they may enjoy a slightly higher level of protection than ordinary OUO.
However, nearly every govenrment computer in existence includiung laptops has gobs of OUO information on it. It's not encrypted because it's not that sort of information. It's just controlled dissemination. That does not mean it might be harmless to release it but it's way below classified.
It is not alarming the people occasionally accdentally disseminate or lose control of OUO. Employees are simply expcted not to do so wilfully or wantonly or carelessly. Its even permissible to share OUO with people outside the governemnt if the employee thinks it would be useful to do so. The fact that OUO was taken home is not a big deal.
In this case the only big distinctions are the massive quantity of the information, and the fact that it's personell records which do have higher levels of protection. Apparently it was also policy not to take these home.
f you look at the PI/PO scans in the media forum, the MID (Media ID code) is clearly shown. Learn to read these scans, you'll learn a lot.
Okay but how does this help me when I go to buy a CD or DVD-r. The stores don't list those.
You must not have visited the media forum in a long time. Ritek, especially the G05, is considered junk by most.
Here is a poll currently on Cdfreaks, regading manufacturers (DVD) quality.
You will see that Ritek is highly favored (third place) .
And here is another comprehensive test site that ranks Ritek among the best possible, and Ridata (also made by ritek) among the poorest for DVD media.
this is not to say that I don't doubt you when you say ritek has a bad reptuation. But I do doubt that you could learn that from reading CDfreaks. In fact I cant' even find a bad review of Ritek on the web except possibly on places like Epionions and in regard to their cheapest ten cent disks. Cdfreaks It's a confused jumble of on weighted uncurated information with almost no value since it focuses on positive experience and annecdote not on negative experiences and metrics.
I've had positive experiences with my Pioneer DVR-111D as well. It plugged and played when I replaced my old DRV-104 without any "patch". It's max burn rate is 8x DVD-R and 48x CD, which is plenty for me since I favor quality over speed: I've had exactly one coaster CD out of hundreds of burns. It reads faster than it writes.
The only problem I've noted is that it's slightly more fragile about reading highly scratched DVDs than my old 104.
If you go to CDfreaks you will see people rate Ritek highly. But which Ritek? Ritek makes lots of different versions with highly different prices. You will see that brands like verbatim have different plants with different disks. And some of the allegedly top shelf labels like Plextor don't even make their own disks.
Most manufactureres make both blue and green and sometimes gold CDs. So simply saying buy "Maxell" doesn't really clear up the matter. Furthermore when you scan the forums you mostly see unweighted voting. people vote for the ones they are using but theres no list of the the ones they used and did not like. Comparisons are not metric of put on an even footing.
And when you do see a metric comparison, and you froogle the media model number chances are it's no longer even being made. You can't just assume that one Maxell disk is like another so you can't just use a positive rating of an obsolete model number confers quality to the others.
I'd also like to see a site reviewing the quality of media for CD-R and DVD-R. The thing I care about is not speed, but reliability. And the Media may affect that more than the burner itself. There's a wide variety of prices on media but nothing to really guide you on quality and longevity.
Over the long run the cost of the burner may be small compared to the cost of the media, so there's no big reason to scrimp on the burner price. But there's a big reason to scrimp on the media. Plus of course unreliable media may lose very valuable data. So it's important to understand reliability of media.
I can't find any discussions of this that are not terribly outdated. It seems like every manufarcturer is constantly changing media names and makes several different lines. (e.g. look at Ritek). But on-line stores don't offer enough information to discern what might make one better than the other. (e.g. info on dyes, or disk construction).
Anyone have some reasonably fresh or comprehensive discussions of this. Or list the names of DVD or CD-r you had reliability problems with. Were the problems Batch-like (e.g. if one CD in the cake-box was bad were many of them bad) or random?
The thing that puzzles me is this whole bit about reeds and trees. Where where they growing. Unlike Antratica, which is built on rock, the arctic is an ocean with no rocks above the water. Or so I have read. Hence the term artic ocean.
Hardware and software backdoors are a reality. Look at the tiawanese Router maker that put a backdoor password in all the netgear routers. Consider that britain finally wised up and wont buy closed source software on their defense avionics. Consider the fact that slot machines get ripped off every year by programmrs putting in backdoors.
Sure it's more difficult to imagine how commondity hardware would be rigged but it's not implausible if the target warrants it. There's been some pretty big efforts staged for security interests. For example, the NSA's recent efforts and the British enigma cracking computers.
The total capitalization of Lenovo is a teeny teeny teeny fraction of the value of being able to have a backdoor to secret us government negotiating positions. teeny. it's would not only be truly worth the risk of exposure and loss of bussniess, it would be a dereliction of duty for the chinese not to try to rig the machines.
If device drivers are third-party plugable without a recompile, I wonder if this is going to open the same sort of security hole that windows and it's device driver overwrting causes.
16 terraflops is what the cheap $499 PS3 can do. You'd need a room full of pentiums to touch that and even then you'd never have the interrocnnection band width to equal it.
the PS3 is an unbeatable number cruncher. The question is only if they can wrangle it into a game with enough differences to matter to the consumer's experience.
As for costs, the POWER chips and Intel CPUs you find in an Xbox or a desktop will never ever be able to match the price per performance. Ever. There's no way to go from ten gigaflops to terraflops with those general purpose CPUs. And the Sony's will only get cheaper with time, and the games get better at taking advantage of it.
So really it's more like a race. Will Sony be able to hang on with early tepid sales till the games become worth it? Will developers come to their rescue?
It's not a question of the PS3 being overpriced. it's prices inexpesively for what it is.
Since people seem to think I was talking about shared memory in my post, let me clarify what I mean be a page socket. What I mean is that for two processes to communicate over a normal serial socket they do it one byte at a time. With a page socket, one process could in effect send an entire page of memory to another. It would infact be sending simply a pointer to some page of memory. This pointer could be to a read-only or read-write part of memory enforced in the MMU. Thus sending a page of a data structure would be instantenous.
But it's not free-form shared memory. One process strill has to request and or accept a socket from another. They are isolated. The other process only gets the page pointer when the first one sends it.
Thus not every processes can write to every bit of shared memory because they all live in seperate process spaces, not just a kernel space.
If a process dies, a re-incarnation server can restart it and it can once again send a message requesting the page of memory. It's still message based to request or send pages. Thus it maintains the robust, isolated, message based nature of a micro kernel in many ways.
Or is this how shared kernel memeory works now in Linux?
Perhaps I'm mistaken, but Isn't shared memory essentially available to the entire macro-kernel and all it's processes. Something more fine grained like a page based socket would let two processes agree to communicate. They would be sending messages to each other over a very wide channel: the entire page, not some serial socket.
Some other process could not butt-in on this channel however, since it's not registered to that socket.
Or is that how shared memory works?
Tnnebaum's point is that he can have a re-incarnation server that can recreate a stalled process. Thus by using exclusively message based communication he can assure that this won't disrupt the state of the operating system (because it's all isolated). The problem is when two processes need to share lots of data quickly. THen message passing gets in your way.
Something that had the negotiated nature of a socket, yet allowed two processes to syncronize their large data structures without passing the entire structure serially would be ideal. then you could still potenitally have things like a re-incarnation server. A new process would simply have to re-negotiate the socket.
I think the real issue here that brings this argument full circle is that when speed for basic operations was an issue then Linux wins. But in todays world we have plenty of fast CPUs. What we lack is stability and the ability to architect robust systems in a scalable manner. For applications that require every ounce of speed, well only those that actually interact with the OS will notice much differnce between Micro-kernel and linux. So this is not to say that the whole computer is slow under micro kernel.
It seems to me the whole issue boils down to memory isolation. If you always have to pass messages to communicate you have good isolation but costly syncronization of data/state and hence potential performance hits. And vica versa: Linux is prone to instability and security breaches from every non-iolated portion of it.
As I understand it, as a novice, the only way to communincate or syncronize data is via copies of data passed via something analogous to a socket. A Socket is a serial interface. If you think about this for a moment, you realize this could be thought of as one byte of shared memory. Thus a copy operation is in effect the iteration of this one byte over the data to share. At any one moment you can only syncronize that one byte.
But this suggests it's own solution. Why not share pages of memory in parallel between processes. This is short of full access to all of the state of another process. But it would allow locking and syncronization processes on entire system states and the rapid passing of data without copies.
Then it would seem like the isolation of mickrokernels would be fully gained without the complications that arrise in multi processing, or compartmentalization.
On the one hand there is clear neccessity for the governement to establish a foresnic identity system. Finger prints, photographs, age, weight, height, eye color, build, race and gender are all legitimate and well established metric the government collects and wisely uses in our collective best interest.
One the other hand, DNA is quite different. You can learn from DNA things the govenrment is not entitled to know. Your lineage, your health prospects, your allegries, and any number of personal attributes. From blood you can learn even more. e.g. are you HiV positive.
So saying DNA and bllod are one more in a long line of useful tools is not a gimme. We have to think it through.
It is quite clear that infinite knowledge of people is not neccessarily in societies best interest. Or at least our society does not agree that it is. And crime deterence is not the sole purpose of governement. protection of privacy and civil lberties needs to be considered. For example, even prisons and navy ships, the most well watched populations on the planet, do not fully prevent crime. And we certainly would not be willing to subject ourselves to that kind of scrutiny just to reduce crime. So there must be a trade between security and liberty and risk. One should not just blindly always trade liberty for security becuase the trade off is without limit.
Yet coming back to DNA. unlike everything except finger prints, it's something that ubquitously taints crime scenes, and it's utility is thus so much above any othe rmetric it's foolish not to atleast consider a DNA databse of former felons and possibly even citizens at large. One solution to this might be DNA hashing. perhaps there is a way to hash a DNA sequence in a manner that would be sufficient to establish presence at a crime scene. Or maybe atleast probable cause for further testing of a particular individual without actually having the governement retain DNA samples of innocent people.
An approach to this would be to identify a long list of biological diversity markers then weed out all the ones know to be associated with any health condition. Then hash these in a way that preserves just enough features to establish likely identity between two samples without revelaing any further details. The govenrment would be required to destroy the original samples and to delete any of the pre-hash specific information. This would have to be done in a manner we can trust them to actually execute this policy. I think this could be done and just to make the point, here's how. Have all testing done in labs in non-networked computers with small hard disks. This would be a physical layer to prevent overt records retention. One could of course imagine ways this could be subverted on a case by case basis but it would impede wholsale collection.
So if I ship my DVDs in a netflix enevelope wil they opne them. If they play them to see what they contain are they violating thr DRM? Do they he probable cause to perform this search?
Having lately been coding Interprocess communication structures, I think Linus nails the issue. The solution to this it seems to me is very simple in concept. We need something akin to a parallel socket connection. As it is stands copying memory, as one does with serial scoket communication, is the bottleneck in any message passing communication between processes.
Now one way to look at this is that serial communicaiton, like copying, is effectively like have one byte of shared memory between two processes. To transfer a data straucture you effectively have to serialize it across this one byte. You do have locking bewteen the processes, only its just on the one byte (the last byte copied or the last byte serialized). So it's not locking on the data structure which is the problem.
So the solution to this is to have a page of shared memory. Lots of bytes. and locking on this page.
One still has message passing and isolated processes since they are only sharing that page not their stacks or registers, or anything else.
I like to call this a parallel socket. But whatever it's called, if you implemented it, it would bridge the world of micro and monolithic kernels. And it would allow you to acrue all the advantages of micro kernels with only a slight weakinging of their isolation.
I know you were joking but you have a serious point. iTunes succeeded because of the quality of service. You could easily find massive (massive!) numbers of songs, with precise documentation, preview them at high quality, and then buy and download them in a snap without a single surpise. Outside of concentrated areas like univeristies, The filesharing community could not match that quality of service. The reason for this is probably due to two things first the catalog is so large, and the immatrutity of the bit torrent like networks at the time.
With movies and the more mature bit torrent this all changes. First there are not that many profitable movies out at any given time. Older movies don't retain their value like older songs do so the catalog of interest to the studios at any given time is smaller and more vulnerable. Because the number of movies of interest to viewers is so concentrated it makes it very easy to have a critical mass of bit torrent.
By critical mass I mean a community of peers sufficient to saturate your inbound or outbound transfer rates. Once you saturate those, then the studios themselves cannot compete on the basis of "ease of download" since they cannot possibly offer a superior network. COntrast this to audio recorndings where it was quite difficult to get a high download rate or even locate peers for every song of interest.
Moreover, right now the quality of movies being sent out is pretty much near maximal. That is when MPEG2 movies are being sent they generally sent as 4 gig movies due to the desire to place them on single layer DVD. The demand for 9 gig movies is almost non existant, not because of the donwload speed but because of this stoarage issue. The Studio's will face the same constraint so it's hard to see how they can offer any better quality than ripped movies can.
thus unlike itunes. It's hard to understand how the movie studio's will have an edge here.
I've never understood the attraction of Vim, maybe someone could explain. It seems like a throwback to keyboard command line editors with it's modal editing.
For my needs I either want a nice gui, in which case I will use kwrite, or bbedit, or some IDE. or I want something simple from the commandline, in which case pico is almost useful, though I prefer emacs for that. I am not an emacs power user. All I can do is do primive searches, cut and paste. But that's really all I need for quick command line edits.
The other reason I like emacs and it's non-modal behaviour is that on a mac, those simple key bindings are available in every cocoa test window.
If you never read A canticle for Leibowitz, well you need to, it's part of any liberal education.
In any case what is the most enduring instituion bar none. Religion. Start a religious order that protects the sites.
Slashdot Mirror
Well that was part of the question I was posing. You assert that everyone is paying their fair share. Perhaps not. Few people consume all the bandwidth they could and conversely the heavy users are getting a cheap ride since most (home) internet use is a flat fee that varies little (less than a factor of 2). So most folks don't pay more if you use more.
So this is indeed why as people's ability to easily consumbe bandwith rises and it becomes more prevalent we are going to have pressure on the resource to the point where if everyong used a little less then everyone would get faster service on the part they used. But of course as with the tragedy of the commons there's no incentive to reduce ones' personal use since there's not marginal cost for using more.
If indeed heavy users did pay more and that fully accounted for their use then indeed it should scale well. Of course now I raised the scaling issue. Does the internet cost strucure go up or down and people use more bandwith? Is it linear, super linear or sub linear. p2p is sub linear since the more routes off the backbone the less it should cost since the same hardware is more effectively ustilized. but backbone centric may be superlinear since now adding users means further distances too the back bone and then back out neccessitating more than a lineaar increase in infrastucture to add users.
We need to ask ourselves if this is a tragedy of the commons or a case where uniform access decreases costs or provides more public-goods. I don't know which it really is.
The tragedy of the commons is what happens when a resource is provided that lacks a proper mariginal cost for increased use. The classic example is private property versus unrstricted access to public grazing land. By charging a small price for admission per sheep to the land or by making it private, the incentive to overgraze it is removed and the total amount of meat sustainbly raise actually is higher. In this case if it's case where there is simply not enough baqndwith for everyone to do voip, and I don't pay any extra to do VOIP, then it's going to be over grazed and everyone gets a crappy connection. On the otherhand if the connection cost already is sufficient to expand the network to handle all the users that want voip or if we can prevent this from becomeing a power law network with critical links then it may be that the more users the better some sort of p2p works.
Thus another way of looking is this is that the thing we need to fear is too few corporation controlliing the internet and resulting in bottlenecks on backbones. In the long run to get high bandwidth we will need p2p that does not traverse a central backbone.
Assuming that the p2p scaling effect will not be sufficient and the tragedy of the commons wil happen then the way out is to have a pricing schedule. We can put that schedule on the users or on the content providers. the latter is what the backbone owners want since it means no net neutrality and control. The former would be better but I can imagine the cheap ass slashdotters used to paying a tiny sum for all-they-can eat internet won't like it.
I note that most vote talleying machines use windows software. By this I mean both the kiosk s you vote on and the central tabulation systems. In many cases the central tabulators are networked. In some cases the kiosks are.
The information is not classified, it's Official Use Only, which is a form of protected information. Personell records are usually, in part, execmt from freedom of information act requests, so they may enjoy a slightly higher level of protection than ordinary OUO.
However, nearly every govenrment computer in existence includiung laptops has gobs of OUO information on it. It's not encrypted because it's not that sort of information. It's just controlled dissemination. That does not mean it might be harmless to release it but it's way below classified.
It is not alarming the people occasionally accdentally disseminate or lose control of OUO. Employees are simply expcted not to do so wilfully or wantonly or carelessly. Its even permissible to share OUO with people outside the governemnt if the employee thinks it would be useful to do so. The fact that OUO was taken home is not a big deal.
In this case the only big distinctions are the massive quantity of the information, and the fact that it's personell records which do have higher levels of protection. Apparently it was also policy not to take these home.
Okay but how does this help me when I go to buy a CD or DVD-r. The stores don't list those.
You must not have visited the media forum in a long time. Ritek, especially the G05, is considered junk by most.
Here is a poll currently on Cdfreaks, regading manufacturers (DVD) quality. You will see that Ritek is highly favored (third place) . And here is another comprehensive test site that ranks Ritek among the best possible, and Ridata (also made by ritek) among the poorest for DVD media.
this is not to say that I don't doubt you when you say ritek has a bad reptuation. But I do doubt that you could learn that from reading CDfreaks. In fact I cant' even find a bad review of Ritek on the web except possibly on places like Epionions and in regard to their cheapest ten cent disks. Cdfreaks It's a confused jumble of on weighted uncurated information with almost no value since it focuses on positive experience and annecdote not on negative experiences and metrics.
I've had positive experiences with my Pioneer DVR-111D as well. It plugged and played when I replaced my old DRV-104 without any "patch". It's max burn rate is 8x DVD-R and 48x CD, which is plenty for me since I favor quality over speed: I've had exactly one coaster CD out of hundreds of burns. It reads faster than it writes.
The only problem I've noted is that it's slightly more fragile about reading highly scratched DVDs than my old 104.
If you go to CDfreaks you will see people rate Ritek highly. But which Ritek? Ritek makes lots of different versions with highly different prices. You will see that brands like verbatim have different plants with different disks. And some of the allegedly top shelf labels like Plextor don't even make their own disks.
Most manufactureres make both blue and green and sometimes gold CDs. So simply saying buy "Maxell" doesn't really clear up the matter. Furthermore when you scan the forums you mostly see unweighted voting. people vote for the ones they are using but theres no list of the the ones they used and did not like. Comparisons are not metric of put on an even footing.
And when you do see a metric comparison, and you froogle the media model number chances are it's no longer even being made. You can't just assume that one Maxell disk is like another so you can't just use a positive rating of an obsolete model number confers quality to the others.
So no CD freaks really does not clear things up.
I'd also like to see a site reviewing the quality of media for CD-R and DVD-R. The thing I care about is not speed, but reliability. And the Media may affect that more than the burner itself. There's a wide variety of prices on media but nothing to really guide you on quality and longevity.
Over the long run the cost of the burner may be small compared to the cost of the media, so there's no big reason to scrimp on the burner price. But there's a big reason to scrimp on the media. Plus of course unreliable media may lose very valuable data. So it's important to understand reliability of media.
I can't find any discussions of this that are not terribly outdated. It seems like every manufarcturer is constantly changing media names and makes several different lines. (e.g. look at Ritek). But on-line stores don't offer enough information to discern what might make one better than the other. (e.g. info on dyes, or disk construction).
Anyone have some reasonably fresh or comprehensive discussions of this. Or list the names of DVD or CD-r you had reliability problems with. Were the problems Batch-like (e.g. if one CD in the cake-box was bad were many of them bad) or random?
No this is all prehistorical times. So No humans and No atlantis in the warm ocean period. Nice though though.
The thing that puzzles me is this whole bit about reeds and trees. Where where they growing. Unlike Antratica, which is built on rock, the arctic is an ocean with no rocks above the water. Or so I have read. Hence the term artic ocean.
So where were these trees growing?
Hardware and software backdoors are a reality. Look at the tiawanese Router maker that put a backdoor password in all the netgear routers. Consider that britain finally wised up and wont buy closed source software on their defense avionics. Consider the fact that slot machines get ripped off every year by programmrs putting in backdoors.
Sure it's more difficult to imagine how commondity hardware would be rigged but it's not implausible if the target warrants it. There's been some pretty big efforts staged for security interests. For example, the NSA's recent efforts and the British enigma cracking computers.
The total capitalization of Lenovo is a teeny teeny teeny fraction of the value of being able to have a backdoor to secret us government negotiating positions. teeny. it's would not only be truly worth the risk of exposure and loss of bussniess, it would be a dereliction of duty for the chinese not to try to rig the machines.
If device drivers are third-party plugable without a recompile, I wonder if this is going to open the same sort of security hole that windows and it's device driver overwrting causes.
16 terraflops is what the cheap $499 PS3 can do. You'd need a room full of pentiums to touch that and even then you'd never have the interrocnnection band width to equal it.
the PS3 is an unbeatable number cruncher. The question is only if they can wrangle it into a game with enough differences to matter to the consumer's experience.
As for costs, the POWER chips and Intel CPUs you find in an Xbox or a desktop will never ever be able to match the price per performance. Ever. There's no way to go from ten gigaflops to terraflops with those general purpose CPUs. And the Sony's will only get cheaper with time, and the games get better at taking advantage of it.
So really it's more like a race. Will Sony be able to hang on with early tepid sales till the games become worth it? Will developers come to their rescue?
It's not a question of the PS3 being overpriced. it's prices inexpesively for what it is.
Since people seem to think I was talking about shared memory in my post, let me clarify what I mean be a page socket. What I mean is that for two processes to communicate over a normal serial socket they do it one byte at a time. With a page socket, one process could in effect send an entire page of memory to another. It would infact be sending simply a pointer to some page of memory. This pointer could be to a read-only or read-write part of memory enforced in the MMU. Thus sending a page of a data structure would be instantenous.
But it's not free-form shared memory. One process strill has to request and or accept a socket from another. They are isolated. The other process only gets the page pointer when the first one sends it.
Thus not every processes can write to every bit of shared memory because they all live in seperate process spaces, not just a kernel space.
If a process dies, a re-incarnation server can restart it and it can once again send a message requesting the page of memory. It's still message based to request or send pages. Thus it maintains the robust, isolated, message based nature of a micro kernel in many ways.
Or is this how shared kernel memeory works now in Linux?
Perhaps I'm mistaken, but Isn't shared memory essentially available to the entire macro-kernel and all it's processes. Something more fine grained like a page based socket would let two processes agree to communicate. They would be sending messages to each other over a very wide channel: the entire page, not some serial socket.
Some other process could not butt-in on this channel however, since it's not registered to that socket.
Or is that how shared memory works?
Tnnebaum's point is that he can have a re-incarnation server that can recreate a stalled process. Thus by using exclusively message based communication he can assure that this won't disrupt the state of the operating system (because it's all isolated). The problem is when two processes need to share lots of data quickly. THen message passing gets in your way.
Something that had the negotiated nature of a socket, yet allowed two processes to syncronize their large data structures without passing the entire structure serially would be ideal. then you could still potenitally have things like a re-incarnation server. A new process would simply have to re-negotiate the socket.
I think the real issue here that brings this argument full circle is that when speed for basic operations was an issue then Linux wins. But in todays world we have plenty of fast CPUs. What we lack is stability and the ability to architect robust systems in a scalable manner. For applications that require every ounce of speed, well only those that actually interact with the OS will notice much differnce between Micro-kernel and linux. So this is not to say that the whole computer is slow under micro kernel.
It seems to me the whole issue boils down to memory isolation. If you always have to pass messages to communicate you have good isolation but costly syncronization of data/state and hence potential performance hits. And vica versa: Linux is prone to instability and security breaches from every non-iolated portion of it.
As I understand it, as a novice, the only way to communincate or syncronize data is via copies of data passed via something analogous to a socket. A Socket is a serial interface. If you think about this for a moment, you realize this could be thought of as one byte of shared memory. Thus a copy operation is in effect the iteration of this one byte over the data to share. At any one moment you can only syncronize that one byte.
But this suggests it's own solution. Why not share pages of memory in parallel between processes. This is short of full access to all of the state of another process. But it would allow locking and syncronization processes on entire system states and the rapid passing of data without copies.
Then it would seem like the isolation of mickrokernels would be fully gained without the complications that arrise in multi processing, or compartmentalization.
Or is there a bigger picture I'm missing.
On the one hand there is clear neccessity for the governement to establish a foresnic identity system. Finger prints, photographs, age, weight, height, eye color, build, race and gender are all legitimate and well established metric the government collects and wisely uses in our collective best interest.
One the other hand, DNA is quite different. You can learn from DNA things the govenrment is not entitled to know. Your lineage, your health prospects, your allegries, and any number of personal attributes. From blood you can learn even more. e.g. are you HiV positive.
So saying DNA and bllod are one more in a long line of useful tools is not a gimme. We have to think it through.
It is quite clear that infinite knowledge of people is not neccessarily in societies best interest. Or at least our society does not agree that it is. And crime deterence is not the sole purpose of governement. protection of privacy and civil lberties needs to be considered. For example, even prisons and navy ships, the most well watched populations on the planet, do not fully prevent crime. And we certainly would not be willing to subject ourselves to that kind of scrutiny just to reduce crime. So there must be a trade between security and liberty and risk. One should not just blindly always trade liberty for security becuase the trade off is without limit.
Yet coming back to DNA. unlike everything except finger prints, it's something that ubquitously taints crime scenes, and it's utility is thus so much above any othe rmetric it's foolish not to atleast consider a DNA databse of former felons and possibly even citizens at large. One solution to this might be DNA hashing. perhaps there is a way to hash a DNA sequence in a manner that would be sufficient to establish presence at a crime scene. Or maybe atleast probable cause for further testing of a particular individual without actually having the governement retain DNA samples of innocent people.
An approach to this would be to identify a long list of biological diversity markers then weed out all the ones know to be associated with any health condition. Then hash these in a way that preserves just enough features to establish likely identity between two samples without revelaing any further details. The govenrment would be required to destroy the original samples and to delete any of the pre-hash specific information. This would have to be done in a manner we can trust them to actually execute this policy. I think this could be done and just to make the point, here's how. Have all testing done in labs in non-networked computers with small hard disks. This would be a physical layer to prevent overt records retention. One could of course imagine ways this could be subverted on a case by case basis but it would impede wholsale collection.
So if I ship my DVDs in a netflix enevelope wil they opne them. If they play them to see what they contain are they violating thr DRM? Do they he probable cause to perform this search?
Having lately been coding Interprocess communication structures, I think Linus nails the issue. The solution to this it seems to me is very simple in concept. We need something akin to a parallel socket connection. As it is stands copying memory, as one does with serial scoket communication, is the bottleneck in any message passing communication between processes.
Now one way to look at this is that serial communicaiton, like copying, is effectively like have one byte of shared memory between two processes. To transfer a data straucture you effectively have to serialize it across this one byte. You do have locking bewteen the processes, only its just on the one byte (the last byte copied or the last byte serialized). So it's not locking on the data structure which is the problem.
So the solution to this is to have a page of shared memory. Lots of bytes. and locking on this page.
One still has message passing and isolated processes since they are only sharing that page not their stacks or registers, or anything else.
I like to call this a parallel socket. But whatever it's called, if you implemented it, it would bridge the world of micro and monolithic kernels. And it would allow you to acrue all the advantages of micro kernels with only a slight weakinging of their isolation.
I know you were joking but you have a serious point. iTunes succeeded because of the quality of service. You could easily find massive (massive!) numbers of songs, with precise documentation, preview them at high quality, and then buy and download them in a snap without a single surpise. Outside of concentrated areas like univeristies, The filesharing community could not match that quality of service. The reason for this is probably due to two things first the catalog is so large, and the immatrutity of the bit torrent like networks at the time.
With movies and the more mature bit torrent this all changes. First there are not that many profitable movies out at any given time. Older movies don't retain their value like older songs do so the catalog of interest to the studios at any given time is smaller and more vulnerable. Because the number of movies of interest to viewers is so concentrated it makes it very easy to have a critical mass of bit torrent.
By critical mass I mean a community of peers sufficient to saturate your inbound or outbound transfer rates. Once you saturate those, then the studios themselves cannot compete on the basis of "ease of download" since they cannot possibly offer a superior network. COntrast this to audio recorndings where it was quite difficult to get a high download rate or even locate peers for every song of interest.
Moreover, right now the quality of movies being sent out is pretty much near maximal. That is when MPEG2 movies are being sent they generally sent as 4 gig movies due to the desire to place them on single layer DVD. The demand for 9 gig movies is almost non existant, not because of the donwload speed but because of this stoarage issue. The Studio's will face the same constraint so it's hard to see how they can offer any better quality than ripped movies can.
thus unlike itunes. It's hard to understand how the movie studio's will have an edge here.
I've never understood the attraction of Vim, maybe someone could explain. It seems like a throwback to keyboard command line editors with it's modal editing.
For my needs I either want a nice gui, in which case I will use kwrite, or bbedit, or some IDE.
or I want something simple from the commandline, in which case pico is almost useful, though I prefer emacs for that. I am not an emacs power user. All I can do is do primive searches, cut and paste. But that's really all I need for quick command line edits.
The other reason I like emacs and it's non-modal behaviour is that on a mac, those simple key bindings are available in every cocoa test window.
So why is Vim so popular?
and no can tell you are a dog. Woof.
If you never read A canticle for Leibowitz, well you need to, it's part of any liberal education. In any case what is the most enduring instituion bar none. Religion. Start a religious order that protects the sites.
Okay I agree, sorry for being touchy. So many other critical responses made me read your post wrong.