yes. considering how ubiquitous the gtk libraries are becoming having an old gtk2 set and a new gtk3 set would help with both compatibility and continued development. This has been done before, though never as successfully as gcc2/egcs/gcc3, and the work involved in backporting is extensive, but the result is usually more stable, with important features being added and useless ones dropped because no-one wants to backport them. We've been hearing about all these freedesktop.org projects for years and few have actually gotten mainstream. The gnome guys seem conservative, putting appearance and percieved ease of use over functionality, which is great, how bout adding a set of functionality in parallel and integrating it into the mainline when you think its ready?
I do something similar, and browsing the page in html vs a menu is fine if its not too often, so i keep the common bm's in the browser. Nothing like having your bookmarks bookmarked. Also I tend to wipe systems a lot, so when my bookmark files get too big i start over from scratch with the most important ones.
I tend to follow a "if you don't miss it you didn't need it" policy for most data, keeping a reachable backup if i decide i need it later.
Time tends to sort things out better than anything.
You are on their property, and they have some cause. They can't search you, but in more expensive places they might just call the cops. YMMV. Less likely with the lame mag strips but real anti-theft devices ala dept stores they might take the chance.
If they try to search you anyway you can sue the hell out of them tho.
seriously, auto-insert for volumes? at least give us a control panel to turn it off. indexing service on by default, all those annoying "common tasks" panes in explorer especially the control panel, the fact that windows media player 9 or 10 always steals back file associations from mplayerc even if you tell it not to and remove the reg keys and delete the damn executables and files (oh they come back, like the damn poltergeist).
Before the stupid m$ rants come in, I use it for gaming, but really would it be that hard to make an "expert mode" for people who half half a f*ing clue what theyre doing? it's called windows xp professional, so you don't have to set it up by default for people without a clue.
Oh, and for home edition try making the default account non-superuser, just a thought.
Yes, but standards and competition are by nature mutually exclusive.
Standards are about things you can "use", ie practical things that people don't obsess about being the fastest and best and cheapest, they just freakin work. Competition is about having the best, the fastest, the cheapest, its a leapfrog game, but only crazy, dumbass early adopters (like me) buy half the crap that's "bleeding edge". Stuff like that doesn't work day-to-day as well as safe, boring, standards. For everyone who cares about the technology, about progress, yeah thats blasphemy, but about people who want to plug it in and have it work, about making it a commodity, thats a requirement. This is why we don't all drive rocket cars at 1200mph.
--
Technology always grows 10 times faster than the people it serves.
Yes but no. The BBC has 6 episode seasons, and that doesn't stop them from getting a fanbase. Quality over quantity does work, instead of 25 hours of crap people can put out 2 hours of gold and still support themselves. The problem isn't the work, or the medium, its the mechanism set up to manage the medium. 25 episode seasons are the de facto standard because they were made so.
There are things like Red vs Blue or flash animation sites that get by, there will probably be a few awesome hits that hit fad status for a while, and get dvd releases, but like any medium the successful or best is a very small fraction of the total content produced. I mean I can't stand most of whats on tv nowadays.
tyan 2-way boards do 8 sticks. you need 2 opteron mem controllers to do 8 sticks, and the timings suffer somewhat. a 4/8-way board might give you more. opterons are limited to 4 sticks per cpu for the most part, the price of onboard controllers.
you said amd, but intel can do 32gb with 1 cpu, and ddr2 scales better than ddr.
but it seems to me that there is a chink in their armor here. how does the government determine who is allowed to determine what is allowed to be viewed? employing thousands of people for the task of limiting the viewing capabilities of all the others doesn't seem very effective to me.
From a western perspective no, but considering their economic model man-hours are dirt cheap compared to raw or processed material. Also, part of their control comes from the fact that a vast fraction of their population is employed by the state. A great way to make people submit is to make them part of the problem in the first place.
Actually I don't want to come off as too condemning of the Chinese system, as they consider the welfare of the state to be critical to their survival, to the point that as long as the state functions, the manner in which it functions is not valid for questioning.
Again, I don't want to come off as archtypal western *sshole about this, because my priorities and values are different.
qui custodiet ipsos custodes?
more watchers... no really. The system of management and people involved in these kind of projects is beyond our understanding. Imagine if the entire IRS was to work together on a project like a new finance tracking system for money laundering and tax-evasion. The personel efficiency of most Indian/Chinese/Japanese (and I can speak as one of the above) government projects would be obscene in American terms, simply put we do NOT know the true meaning of the word "bureaucracy".
Raw material and tools are expensive, labor and real estate is nearly free. Cong economics.
once wimax is released muni wifi will actually get out the gate, and hard. why? you can support a few square miles with a single base station comfortably, and farther if the user density is lower.
Want to support muni wifi? Support wimax.
p.s. someone nominate this guy biggest douche in the universe, billg can have a by-year for once.
GMO is neither good nor evil in itself, its a technology. Yeah GM'ing rice to make people more hungry or to grow faster w/o nutrients is vry bad, but making wheat more resistant to pests or less likely to get ergot might be a "good thing"(tm). Even modding a strain of wheat that grows better in regions with poorer soil or less water could save lives to people who can't grow their own food today. Not all the world is Europe and the US.
Judging a technology based on its worst possibly product is very "human", a word that just happens to not rhyme at all with "stupid".
n : active support; especially the act of pleading or arguing for something
Perhaps the word you mean is "Whoring"?
ntr.v. whored, whoring, whores
3. To compromise one's principles for personal gain.
Which sadly implies having had principles. I am for politcal advocacy but can you not agree that money distorts the process and interferes with the debate? That is a powerful corrupting factor to let lose on a complex democratic process. I also am not sure that mccain/feingold is a perfect solution, but honestly considering this country spends billions on an election year, and most if not all of that money is tagged with a subtext, ie "i'll help you get elected and you remember my interests", i really think taking some money out of politics is at least a start. Otherwise this country spends billions on an election year and all we get is more commercials, who's good does that serve?
I think independent (ie unsponsored) blogging should be free of FEC control, but if anything the "move on" vs "swiftboat vets..." crap shows how easy the black and white of political funding goes way gray fast.
As a centrist (look it up, there is such a thing), this bill is wrong from both sides. This is not a democratic or republican bill, its a political bill aimed at increasing the ability of the parties to rally their constituencies. Kinda like the US and USSR in the arms race, both sides think they can win with more money, more media, more influence.
This bill is in the interests of dems and reps, but not in the interests of normal human ppl who don't see every detail of the world as part of a huge ideological struggle. All 4 of us...
Perhaps... perhaps we shouldn't/. this as "good news". I know we all hate M$, and love linux, and hate sw patents... but maybe we should sit down a sec and think about what this bill is doing.
The Internet is "not" a public communications medium, so... it's a cheese bagel?
This is not about free speech, free speech is letting me say what I think w/o going to jail. This is about the net as a political medium.
That said, I am not against this bill, but the/. "hypocrisy is ok as long as it agrees with me" logic should take a back seat to common sense. Giving extremist political groups more room to shout their message for money is another thing we all think is "a bad thing(tm)" right?
I disagree with the supreme court ruling that says $ = speech, because that implies rich people have a louder voice than poor people, which seems not so good.
so basically they cut out a lot of signature DNA stuff to fit it on the screen... are we looking at "double platinum plus directors exclusive special edition" for 75.95 at best buy to get the complete movie ala LOTR?
Is this a new movie trick? Show half the movie in theatres to help with the dvd sales?
Dist chain for dvd's and audio cd's end up taking about 70% gross sales themselves. And thats if its successful. Assuming this scales well I'd expect to see a pro version of this complete with legal asstastics and ridiculous t&c's in the next year.
Retail is a messy incestuous world of middle-men and price-gouging, but don't worry, 10 years from now we'll be looking back at this wild-western hippy internet and wonder how we ever survived without pay-per-meg auto-spam, pop-up, and hack blocking and crappy family version rated PG searches. The people who know how to do things are always way in the minority compared to ppl who want it to "just work" and can't be bothered to learn, and that's how capitilism works.
Absolutely, and causes small companies (and lawyers) to drool uncontrollably. Also, the dirty secret of the democratic party is their dependence on the special interest groups of trial lawyers, which sucks.
Brought to you by, Sexual Harrass... err Don't Sue People Panda!
Yeah, cept we'd need an FTL mechanism. Actually we'd probably just want to leave the galaxy. We're in a relatively safe neighborhood, cosmoslogically speaking, few super-giant+ stars nearby, and those relatively young at that. To get much better we'd have to leave the more populous (with stars) segment of the galaxy and head out to the rim where the heavier elements are less common, ala oxygen and carbon. It could work, but would take tens of thousands of years even assuming the best slower than light speeds, even counting on finding a habitable system at the end. That and the fact that hydrogen starts becoming more like heavy microwave radiation at those speeds, and ir-vis light becomes supra-gamma.
All things considered a magnetic rad shield would probably be more achievable, and neccessary anyway due to the lack of ozone in the near future. This is rather possible using the same effect that stops the solar wind from ionizing life on earth, extending that effect so it dampens or more likely lenses high energy rads around the earth, kinda like the mythical cloaking device.
I think you were reading one of niven's known space series, in which the galactic core went supernova in a chain reaction which would wipe out life in the galaxy in 40k years or so.
The article said about 450m years ago no? Is it possible the recorded readings could also be effected by one of the "blackout" periods of the earths mag-field? There is a body of evidence that connects the shifts in earth's mag-field to the global extinction/rapid evolution pattern, but not enough to account for most of the recorded events.
Honestly if i was going to do it again I'd prolly go with a good nf4 board with an el asso wipo silicon image chip onboard, just do sw raid that way. nf4 is a great sata 'set, and you end up with 8 ports total, plus gig'e/dual gig'e. hell of a lot cheaper nowadays too.
nf4 is mostly supported under recent kernels, and you end up with a much better overall system also, especially if youre looking for an apache/mysql combo.
Also, adaptec 2610 6 pt hw card, about $200 on ebay. uses the common linux i20 adaptec driver. just make sure its returnable because a lot of the reboxed dells are either broken or very badly supported by nobody.
please don't do hw raid tho... that path leads to the dark side... and costs a f*ing lot too, when supported.
5x Maxtor 6 B200M0 sata 2 -- ~$120 in a supermicro 5 slot sata hotplug raid enclosure. -- ~$100 connected to a highpoint rocketraid 1820a 64/133 pcix 8-slot sata card. -- ~$150 2x xeon 2.0Ghz @ 533 fsb... -- def not priceless, but it has pcix and is spare *shrug integrated intel 1000MT gige controller on mb 500? watt ssi ps, forget the brand... -- ~$75
gentoo linux, kernel 2.6.9, highpoint card has a proprietary driver which sucks, and proprietary software raid module which gave me problems i'll explain later but would probably be workable if you are into that kind of thing.
linux md raid-5 across all 5. 800Gb resultant/dev/md0. formatted reiserfs 3.6.
my opinion: works well enough, performance is not amazing but is very solid under most/all loads (mostly fileserving on nfs/samba)
my biggest problem is that the hotplug enclosure tends to cause at least 1 drive to fail to start up properly on reboot, needing a removal and reinsertion and rmmod/insmod to get all the drives online, which is a royal pita. This also means I can't use the built-in software raid module, which doesn't bother me that much, except I lose the hotplugging (not something I'd use). Once I'm up and running tho it's rock-solid for as long as I need it.
md has been beautiful to me so far, too bad the drives support ncq and the card does not. Highly recommend something similar, unless you can get a better controller. I read lots of reviews and few hw sata raid cards came close to sw raid in linux, and the flexibility of md is a MUST-have. Don't think I'm hitting any bottlenecks yet. Really hard to beat the storage/price point right now, and at that size backups become non-trivial.
note: samba and nfs clients will probably require mild to severe tuning for best performance, ymmv. For mysql/apache loads this is possibly overkill, but md is very good with cache/paging if you have the ram for it. Most/all other linux hw raid controllers get killed on cache management alone from what ive read.
Also have a 4x 6y250m0 250Gb raid for backup purposes, but that is usually offline and uses a secondary raid controller.
Actually the internet uses implied addressing to simplify routing, ie 192.168.1.4 is supposedly closer to 192.168.2.15 than 4.2.1.5. After a point the routers just throw darts and hope the packets make it. The fact that there are so many darts and each dart gives feedback to sending the next one means the net is self-optimizing. In terms of an aggregate wireless domain, the subnets would be divided by space, not numbers or hops or routers. Using TCP logic on the internet barely works, using the same logic on a constantly changing mesh would be useless from an actual addressing standpoint.
Instead of arbitrary subnets addresses, the subnets would be designated by relation, ie "1v2279x988bb9c12" would translate into something like "I'm the 0x12th slot in spatial domain "88bb9c12", with the rest of the address explaining how to get to that node, which frequency or protocol to use, and even possibly qos or traffic policies. Using 256-bits to do addressing is cheap nowadays. And if that mesh is beyond your near wireless domain, you can prefix 64 bits of wire routing to get to that domain.
Yes this is a rambling post, but by speccing out or at least considering extensible networking protocols ahead of time will save you 20x time trying to hack new functionality without breaking the old equipment.
And if you really want to make it easy, just give all those devices a crappy gps and have them use hashed position data as part of their uid. Leave enough data unencrypted to find the proper spatial domain, and the rest is just used for roaming or collision avoidance, or whatever.
a: i really doubt tcp would work in stock config, it does not likey high entropy networks, no matter what the RFC carrier pigeon says.
b: true, in random configs that is totally right, but at n much less than inf and especially with adaptive power and frequency modulation control and hopefully 2 freqency domains for near and far-range traffic the equation stays managable. The key is localized traffic using minimum power with high spatial rolloff.
c: no, but this is possible to do brute force by conservatively handling time and freqency domains, and long callout arbitration, etc. basically moving nodes will suck, but nodes with the oldest connections will be given precedence with freq/timeslots, as they are least likely to be moving. basically nodes that show up last get last dibs on bw, oh well.
mostly engineering problems, and honestly tcp needs to be updated or wrapped or something. stock tcp on fiber is not remotely funny, and gracefully handling small data losses would allow for vastly faster networks. likely a CDMA or TDMA approach for multi-link tcp would fix some of these problems and allow for higher bandwith by parallizing several slower streams, increasing agg bw and allowing for true multi-link/route connections.
it has years to go, but i doubt it would take decades. when pc's reach the point that they are by definition portable this becomes a neccessity.
two words:
winsock
of all the network stacks in all the world, MS had to go and badly rip-off that one.
offtopic question: shouldnt there be licensing issues there?
yes and no. This assumes all nodes are operating on or near the same spatial domain. In wireless communications bandwith is dependent on power, frequency, and snr/sensitivity, but there is also a counter-effect at higher freqencies of spatial separation, where enough space divides 2 nodes that they can be considered to be operating independent of each other. Future designs will have to work towards this tendency, but this will allow high short-range bandwith with much lower inter-node or inter-domain bandwith. Yes with n->inf the node bandwith = 0, but there are also values n under proper circumstances in which the inter-node bandwith is maximized. Likely this would be a way to build high-speed local networks while having slow broadband to get wan data, possibly caching it within the mesh.
or not...
Oh my lord, THOUSANDS of participants? How could we ever manage to route between 2^16 possible nodes. Even the numbers seem too big for my feeble vic-20 to comprehend. And even then we'd have to write out long node numbers like 192.168.
Best case bandwith will be increased, IF the bandwith is strongly localized, ie weak signals with quick dropoff for distance. Imagine 802.11g cept every participant who is reaching for a different node has a dedicated connection. Also with very low power and high dropoff transmissions you can reuse spectrum aggressively, and tricks like OFDM and MIMO really come into their own.
Christ, the internet does a halfass'd job of routing, and uses very little processing power relative to the task, and still manages to deliver good results as long there is more than 1 effective path (ie the route isn't deep inside a single ISP) to go by.
It's too early to say whether this will work, but why not just modify 802.11g, drop the power and create a differential transmission mechanism to increase the spatial SNR ratio? I mean/. has 20 articles a month showing how to send a signal 2 miles with tinfoil, this would be the opposite, with frequency shifting based on direction to boost usable bandwith per spectrum. I remember this being added to AEGIS baselines 5 years ago.
http://www.mobl.com/expansion/pci/index.html
cardbus to pci enclosures, everything you need.
1 caveat: some (newer) bus-mastering pci cards dont take too kindly to multiple bridging as used in these types of boxes. other than that they should cover you fine
Slashdot Mirror
yes.
considering how ubiquitous the gtk libraries are becoming having an old gtk2 set and a new gtk3 set would help with both compatibility and continued development. This has been done before, though never as successfully as gcc2/egcs/gcc3, and the work involved in backporting is extensive, but the result is usually more stable, with important features being added and useless ones dropped because no-one wants to backport them. We've been hearing about all these freedesktop.org projects for years and few have actually gotten mainstream. The gnome guys seem conservative, putting appearance and percieved ease of use over functionality, which is great, how bout adding a set of functionality in parallel and integrating it into the mainline when you think its ready?
just a thought
I do something similar, and browsing the page in html vs a menu is fine if its not too often, so i keep the common bm's in the browser. Nothing like having your bookmarks bookmarked. Also I tend to wipe systems a lot, so when my bookmark files get too big i start over from scratch with the most important ones.
I tend to follow a "if you don't miss it you didn't need it" policy for most data, keeping a reachable backup if i decide i need it later.
Time tends to sort things out better than anything.
gray area.
You are on their property, and they have some cause. They can't search you, but in more expensive places they might just call the cops. YMMV. Less likely with the lame mag strips but real anti-theft devices ala dept stores they might take the chance.
If they try to search you anyway you can sue the hell out of them tho.
all of it.
not thrilled about dos either.
seriously, auto-insert for volumes? at least give us a control panel to turn it off. indexing service on by default, all those annoying "common tasks" panes in explorer especially the control panel, the fact that windows media player 9 or 10 always steals back file associations from mplayerc even if you tell it not to and remove the reg keys and delete the damn executables and files (oh they come back, like the damn poltergeist).
Before the stupid m$ rants come in, I use it for gaming, but really would it be that hard to make an "expert mode" for people who half half a f*ing clue what theyre doing? it's called windows xp professional, so you don't have to set it up by default for people without a clue.
Oh, and for home edition try making the default account non-superuser, just a thought.
christ.
Yes, but standards and competition are by nature mutually exclusive.
Standards are about things you can "use", ie practical things that people don't obsess about being the fastest and best and cheapest, they just freakin work. Competition is about having the best, the fastest, the cheapest, its a leapfrog game, but only crazy, dumbass early adopters (like me) buy half the crap that's "bleeding edge". Stuff like that doesn't work day-to-day as well as safe, boring, standards. For everyone who cares about the technology, about progress, yeah thats blasphemy, but about people who want to plug it in and have it work, about making it a commodity, thats a requirement. This is why we don't all drive rocket cars at 1200mph.
--
Technology always grows 10 times faster than the people it serves.
Yes but no. The BBC has 6 episode seasons, and that doesn't stop them from getting a fanbase. Quality over quantity does work, instead of 25 hours of crap people can put out 2 hours of gold and still support themselves. The problem isn't the work, or the medium, its the mechanism set up to manage the medium. 25 episode seasons are the de facto standard because they were made so.
There are things like Red vs Blue or flash animation sites that get by, there will probably be a few awesome hits that hit fad status for a while, and get dvd releases, but like any medium the successful or best is a very small fraction of the total content produced. I mean I can't stand most of whats on tv nowadays.
http://www.newegg.com/app/ViewProductDesc.asp?desc ription=13-151-134&depa=0
tyan 2-way boards do 8 sticks. you need 2 opteron mem controllers to do 8 sticks, and the timings suffer somewhat. a 4/8-way board might give you more. opterons are limited to 4 sticks per cpu for the most part, the price of onboard controllers.
you said amd, but intel can do 32gb with 1 cpu, and ddr2 scales better than ddr.
Actually I don't want to come off as too condemning of the Chinese system, as they consider the welfare of the state to be critical to their survival, to the point that as long as the state functions, the manner in which it functions is not valid for questioning.
Again, I don't want to come off as archtypal western *sshole about this, because my priorities and values are different. more watchers... no really. The system of management and people involved in these kind of projects is beyond our understanding. Imagine if the entire IRS was to work together on a project like a new finance tracking system for money laundering and tax-evasion. The personel efficiency of most Indian/Chinese/Japanese (and I can speak as one of the above) government projects would be obscene in American terms, simply put we do NOT know the true meaning of the word "bureaucracy".
Raw material and tools are expensive, labor and real estate is nearly free. Cong economics.
once wimax is released muni wifi will actually get out the gate, and hard. why? you can support a few square miles with a single base station comfortably, and farther if the user density is lower.
Want to support muni wifi? Support wimax.
p.s. someone nominate this guy biggest douche in the universe, billg can have a by-year for once.
Wow, broad generalisms, nifty.
GMO is neither good nor evil in itself, its a technology. Yeah GM'ing rice to make people more hungry or to grow faster w/o nutrients is vry bad, but making wheat more resistant to pests or less likely to get ergot might be a "good thing"(tm). Even modding a strain of wheat that grows better in regions with poorer soil or less water could save lives to people who can't grow their own food today. Not all the world is Europe and the US.
Judging a technology based on its worst possibly product is very "human", a word that just happens to not rhyme at all with "stupid".
Damn the internet because of child porn?
Judge the act, not the idea.
advocacy
n : active support; especially the act of pleading or arguing for something
Perhaps the word you mean is "Whoring"?
ntr.v. whored, whoring, whores
3. To compromise one's principles for personal gain.
Which sadly implies having had principles.
I am for politcal advocacy but can you not agree that money distorts the process and interferes with the debate? That is a powerful corrupting factor to let lose on a complex democratic process.
I also am not sure that mccain/feingold is a perfect solution, but honestly considering this country spends billions on an election year, and most if not all of that money is tagged with a subtext, ie "i'll help you get elected and you remember my interests", i really think taking some money out of politics is at least a start. Otherwise this country spends billions on an election year and all we get is more commercials, who's good does that serve?
I think independent (ie unsponsored) blogging should be free of FEC control, but if anything the "move on" vs "swiftboat vets..." crap shows how easy the black and white of political funding goes way gray fast.
Summation:
Money + agenda = bad.
As a centrist (look it up, there is such a thing), this bill is wrong from both sides. This is not a democratic or republican bill, its a political bill aimed at increasing the ability of the parties to rally their constituencies. Kinda like the US and USSR in the arms race, both sides think they can win with more money, more media, more influence.
This bill is in the interests of dems and reps, but not in the interests of normal human ppl who don't see every detail of the world as part of a huge ideological struggle. All 4 of us...
Perhaps... perhaps we shouldn't /. this as "good news". I know we all hate M$, and love linux, and hate sw patents... but maybe we should sit down a sec and think about what this bill is doing.
/. "hypocrisy is ok as long as it agrees with me" logic should take a back seat to common sense. Giving extremist political groups more room to shout their message for money is another thing we all think is "a bad thing(tm)" right?
The Internet is "not" a public communications medium, so... it's a cheese bagel?
This is not about free speech, free speech is letting me say what I think w/o going to jail. This is about the net as a political medium.
That said, I am not against this bill, but the
I disagree with the supreme court ruling that says $ = speech, because that implies rich people have a louder voice than poor people, which seems not so good.
so basically they cut out a lot of signature DNA stuff to fit it on the screen... are we looking at "double platinum plus directors exclusive special edition" for 75.95 at best buy to get the complete movie ala LOTR?
Is this a new movie trick? Show half the movie in theatres to help with the dvd sales?
Dist chain for dvd's and audio cd's end up taking about 70% gross sales themselves. And thats if its successful. Assuming this scales well I'd expect to see a pro version of this complete with legal asstastics and ridiculous t&c's in the next year. Retail is a messy incestuous world of middle-men and price-gouging, but don't worry, 10 years from now we'll be looking back at this wild-western hippy internet and wonder how we ever survived without pay-per-meg auto-spam, pop-up, and hack blocking and crappy family version rated PG searches. The people who know how to do things are always way in the minority compared to ppl who want it to "just work" and can't be bothered to learn, and that's how capitilism works.
Absolutely, and causes small companies (and lawyers) to drool uncontrollably. Also, the dirty secret of the democratic party is their dependence on the special interest groups of trial lawyers, which sucks.
Brought to you by, Sexual Harrass... err Don't Sue People Panda!
Yeah, cept we'd need an FTL mechanism. Actually we'd probably just want to leave the galaxy. We're in a relatively safe neighborhood, cosmoslogically speaking, few super-giant+ stars nearby, and those relatively young at that. To get much better we'd have to leave the more populous (with stars) segment of the galaxy and head out to the rim where the heavier elements are less common, ala oxygen and carbon. It could work, but would take tens of thousands of years even assuming the best slower than light speeds, even counting on finding a habitable system at the end. That and the fact that hydrogen starts becoming more like heavy microwave radiation at those speeds, and ir-vis light becomes supra-gamma.
All things considered a magnetic rad shield would probably be more achievable, and neccessary anyway due to the lack of ozone in the near future. This is rather possible using the same effect that stops the solar wind from ionizing life on earth, extending that effect so it dampens or more likely lenses high energy rads around the earth, kinda like the mythical cloaking device.
I think you were reading one of niven's known space series, in which the galactic core went supernova in a chain reaction which would wipe out life in the galaxy in 40k years or so.
The article said about 450m years ago no? Is it possible the recorded readings could also be effected by one of the "blackout" periods of the earths mag-field? There is a body of evidence that connects the shifts in earth's mag-field to the global extinction/rapid evolution pattern, but not enough to account for most of the recorded events.
just my 2c
Honestly if i was going to do it again I'd prolly go with a good nf4 board with an el asso wipo silicon image chip onboard, just do sw raid that way. nf4 is a great sata 'set, and you end up with 8 ports total, plus gig'e/dual gig'e. hell of a lot cheaper nowadays too.
nf4 is mostly supported under recent kernels, and you end up with a much better overall system also, especially if youre looking for an apache/mysql combo.
Also, adaptec 2610 6 pt hw card, about $200 on ebay. uses the common linux i20 adaptec driver. just make sure its returnable because a lot of the reboxed dells are either broken or very badly supported by nobody.
please don't do hw raid tho... that path leads to the dark side... and costs a f*ing lot too, when supported.
5x Maxtor 6 B200M0 sata 2 -- ~$120 ... -- def not priceless, but it has pcix and is spare *shrug
/dev/md0. formatted reiserfs 3.6.
in a supermicro 5 slot sata hotplug raid enclosure. -- ~$100
connected to a highpoint rocketraid 1820a 64/133 pcix 8-slot sata card. -- ~$150
2x xeon 2.0Ghz @ 533 fsb
integrated intel 1000MT gige controller on mb
500? watt ssi ps, forget the brand... -- ~$75
gentoo linux, kernel 2.6.9, highpoint card has a proprietary driver which sucks, and proprietary software raid module which gave me problems i'll explain later but would probably be workable if you are into that kind of thing.
linux md raid-5 across all 5. 800Gb resultant
my opinion:
works well enough, performance is not amazing but is very solid under most/all loads (mostly fileserving on nfs/samba)
my biggest problem is that the hotplug enclosure tends to cause at least 1 drive to fail to start up properly on reboot, needing a removal and reinsertion and rmmod/insmod to get all the drives online, which is a royal pita. This also means I can't use the built-in software raid module, which doesn't bother me that much, except I lose the hotplugging (not something I'd use). Once I'm up and running tho it's rock-solid for as long as I need it.
md has been beautiful to me so far, too bad the drives support ncq and the card does not. Highly recommend something similar, unless you can get a better controller. I read lots of reviews and few hw sata raid cards came close to sw raid in linux, and the flexibility of md is a MUST-have. Don't think I'm hitting any bottlenecks yet. Really hard to beat the storage/price point right now, and at that size backups become non-trivial.
note: samba and nfs clients will probably require mild to severe tuning for best performance, ymmv. For mysql/apache loads this is possibly overkill, but md is very good with cache/paging if you have the ram for it. Most/all other linux hw raid controllers get killed on cache management alone from what ive read.
Also have a 4x 6y250m0 250Gb raid for backup purposes, but that is usually offline and uses a secondary raid controller.
pricing is half-guess/half-bs
Actually the internet uses implied addressing to simplify routing, ie 192.168.1.4 is supposedly closer to 192.168.2.15 than 4.2.1.5. After a point the routers just throw darts and hope the packets make it. The fact that there are so many darts and each dart gives feedback to sending the next one means the net is self-optimizing. In terms of an aggregate wireless domain, the subnets would be divided by space, not numbers or hops or routers. Using TCP logic on the internet barely works, using the same logic on a constantly changing mesh would be useless from an actual addressing standpoint.
Instead of arbitrary subnets addresses, the subnets would be designated by relation, ie "1v2279x988bb9c12" would translate into something like "I'm the 0x12th slot in spatial domain "88bb9c12", with the rest of the address explaining how to get to that node, which frequency or protocol to use, and even possibly qos or traffic policies. Using 256-bits to do addressing is cheap nowadays. And if that mesh is beyond your near wireless domain, you can prefix 64 bits of wire routing to get to that domain.
Yes this is a rambling post, but by speccing out or at least considering extensible networking protocols ahead of time will save you 20x time trying to hack new functionality without breaking the old equipment.
And if you really want to make it easy, just give all those devices a crappy gps and have them use hashed position data as part of their uid. Leave enough data unencrypted to find the proper spatial domain, and the rest is just used for roaming or collision avoidance, or whatever.
--human
lots of points.
a: i really doubt tcp would work in stock config, it does not likey high entropy networks, no matter what the RFC carrier pigeon says.
b: true, in random configs that is totally right, but at n much less than inf and especially with adaptive power and frequency modulation control and hopefully 2 freqency domains for near and far-range traffic the equation stays managable. The key is localized traffic using minimum power with high spatial rolloff.
c: no, but this is possible to do brute force by conservatively handling time and freqency domains, and long callout arbitration, etc. basically moving nodes will suck, but nodes with the oldest connections will be given precedence with freq/timeslots, as they are least likely to be moving. basically nodes that show up last get last dibs on bw, oh well.
mostly engineering problems, and honestly tcp needs to be updated or wrapped or something. stock tcp on fiber is not remotely funny, and gracefully handling small data losses would allow for vastly faster networks. likely a CDMA or TDMA approach for multi-link tcp would fix some of these problems and allow for higher bandwith by parallizing several slower streams, increasing agg bw and allowing for true multi-link/route connections.
it has years to go, but i doubt it would take decades. when pc's reach the point that they are by definition portable this becomes a neccessity.
two words: winsock of all the network stacks in all the world, MS had to go and badly rip-off that one. offtopic question: shouldnt there be licensing issues there?
yes and no. This assumes all nodes are operating on or near the same spatial domain. In wireless communications bandwith is dependent on power, frequency, and snr/sensitivity, but there is also a counter-effect at higher freqencies of spatial separation, where enough space divides 2 nodes that they can be considered to be operating independent of each other. Future designs will have to work towards this tendency, but this will allow high short-range bandwith with much lower inter-node or inter-domain bandwith. Yes with n->inf the node bandwith = 0, but there are also values n under proper circumstances in which the inter-node bandwith is maximized. Likely this would be a way to build high-speed local networks while having slow broadband to get wan data, possibly caching it within the mesh. or not...
Oh my lord, THOUSANDS of participants? How could we ever manage to route between 2^16 possible nodes. Even the numbers seem too big for my feeble vic-20 to comprehend. And even then we'd have to write out long node numbers like 192.168.
/. has 20 articles a month showing how to send a signal 2 miles with tinfoil, this would be the opposite, with frequency shifting based on direction to boost usable bandwith per spectrum. I remember this being added to AEGIS baselines 5 years ago.
Best case bandwith will be increased, IF the bandwith is strongly localized, ie weak signals with quick dropoff for distance. Imagine 802.11g cept every participant who is reaching for a different node has a dedicated connection. Also with very low power and high dropoff transmissions you can reuse spectrum aggressively, and tricks like OFDM and MIMO really come into their own.
Christ, the internet does a halfass'd job of routing, and uses very little processing power relative to the task, and still manages to deliver good results as long there is more than 1 effective path (ie the route isn't deep inside a single ISP) to go by.
It's too early to say whether this will work, but why not just modify 802.11g, drop the power and create a differential transmission mechanism to increase the spatial SNR ratio? I mean
my 2c
http://www.mobl.com/expansion/pci/index.html cardbus to pci enclosures, everything you need. 1 caveat: some (newer) bus-mastering pci cards dont take too kindly to multiple bridging as used in these types of boxes. other than that they should cover you fine