The article is kinda vauge but AIUI they are talking about forcing american companies to store data about brazillians in brazil. Well that raises a few issues.
1: what are they going to do if some of those american companies tell them to go pound sand? Unless the company in question has a direct buisness presense in brazil it seems their choices are to either block connections (the "great firewall soloution) or lets things continue as they are. 2: If the NSA uses a national security letter to order the american parent company to get them some data on a brazillian what happens? Given a choice between breaking brazillian law and being punished for ignoring an american national security letter what do you expect an american company to do? 3: What happens when an amercian user and a brazillian user want/need to work together. For example suppose a brazillian user makes his calender accessible to his american friend. Should that calender be hosted in brazil? the USA? both? How will features like finding a slot where everything is free work if each country's citizens data has to remain within the country.
They are also taking about reducing reliance on the USA for connectivity to the rest of the world which just seems like good sense (afaict they already have one direct connection to europe but I doubt that is really enough) though it may well increase costs in the short term (US internet transit is CHEAP).
AIUI in the US only one of the major networks (t-mobile) gives users who bring their own phones a decent deal. Another (AT&T) doesn't forbid their use but structures their traffifs such that you effectively pay for a phone from them every couple of years whether you take them up on the offer or not. The remaining two use mostly "CDMA"* technology which doesn't use sims meaning you can only use phones the carrier will agree to activate. This may change with the introduction of LTE but AIUI that is not yet available in most places.
There are "virtual networks" out there too which can sometimes give better deals for those who want to bring their own phones to the AT&T network but AIUI they get lower priority and don't get to benefit from roaming arrangements AT&T has.
* Strictly speaking all modern mobile systems use CDMA modulation but there are still two sets of standards, those that grew out of the old CDMA standards and those that grew
AIUI the "Bureau of Engraving and Printing" (which is part of the treasury) physically prints it but the federal reserve issues it. Until it's issued it isn't legally money. In that sense saying that the federal reserve prints money is kind of like saying apple makes iphones.
However "printing" nowadays is mostly a metaphor anyway. Most of the money the federal reserve creates is just bits in a database but those bits in a database are legally equivilent to cash.
The treasury also creates a small ammount of money by issuing coins but that is even smaller than the ammount the federal reserve creates by issuing banknotes.
Part A has essentially happened. Everything by valve that's not way-too-old-to-port has been linuxified.
If anything the opposite seems true. The really old goldsrc stuff (HL, opposing force, blue shift, CS, TF:C) seems to have been ported and so does the first generation source stuff that was kept up to date (HL2, HL2:EP1, HL2:EP2, HL2:DM, HL2:LC, portal, TF2, CS:S). I also notice DOTA 2 and deathmatch classic are aslo available on linux but I don't know what engine branches they used.
The games based on more recent branches of the source engine (alien swarm, left 4 dead, portal 2, CS:GO) do not appear to have linux releases. Nor do the source engine conversions of the original half life and counterstrike.
Any one person's contributions to climate change are pretty tiny. Even rich gits who fly around in private planes. Even national presidents who fly arround in a jumbo jet bringing with them a massive entourage. The number of rich git's in private planes is also fairly small so the total impact of all rich gits flying arround in private planes is probablly not that high in the grand scheme of things.
However why would the commoners listen to a rich git who effectively says "do as I say not as I do" when it comes to climate change? While the impact of all commoners changing their ways would probablly be bigger than the impact of all the rich gits changing their ways the impact of one individual commoner changing their ways is going to be even tinier than the impact of one rich git changing their ways.
There are various ways a power supply* can respond to an overcurrent situation including
1: Overheat and destroy itself (possiblly even catching fire though that is thankfully rare). 2: Enter a current limit mode where the voltage drops off sharply as current increases. 3: Trip out and refuse to supply any power until reset somehow.
Now suppose I have a nominally 1A power supply. This power supply goes into current limit at just over 1A and shorts the D+ and D- lines as per the USB battery charging specification. When I plug a tablet into it the tablet tries to charge at 2A but 2A isn't available so the voltage drops until the tablet is only drawing just over 1A. The tablet charges successfully albiet slower than it would with a 2A charger..
On the other hand suppose I had a another nominally 1A power supply but that tripped at just over 1A. I connect a USB cable to it and short the D+ and D- lines. When I plug a tablet into it the tablet would try to draw 2A and trip the protection. The tablet therefore doesn't charge at all.
Now supposed I had another nominally 1A power supply that overheats and destroys itself when attemting to supply more than 1A. I connect a USB cable to it and short the D+ and D- lines. When I plug my tablet in it would try to draw 2A and destroy the power supply.
That's like putting a restrictor plate on my car so that I can go faster then it allows me to go.
To take your car analogy imagine you had three cars. One doesn't have a rev limiter at all and you can destory the engine by over-revving it. One has a rev limiter that works in the conventional way limiting the engine speed to a predetermined level and the final one has a rev limiter that shuts down the whole car ifyou hit the limit.
* People reffer to them as "chargers" but the charge control circuitry is actually in the phone.
Afaict the problem is a practical latex document is likely to use a mixture of structural constants, "just put this where I damn well tell you to" constructs and constructs that while nominally strutural are being tweaked to make things look good on the printed page (for example moving a figure up or down in the text so it ends up on the page you want it on).
An automated conversion is likely to produce something that is just about readable but a high quality conversion is likely to require human judgement.
I don't know if you are being serious but AIUI at least one of the electrodes is a consumable. So to maintain crude batteries you need not just a supply of electrolyte (the wine) but also a supply of refined metals.
Which is why you take the eeprom off the original controller board and put it on the new board. In my experiance the EEPROM is in an 8 pin package with a 1.27mm pin spacing (either a SOIC or a similar sized leadless package). Pretty easy to pop off with hot air (and yes I have done this several times).
Of course many laptops do emergency hibernate when the battery status gets critical.
If emergency hibernate works it means you don't lose the work you had in progress when the battery fails but if it triggers too late then I could see it ending up with the drives power being cut while under heavy write load.
AIUI linux development works on the principle of "everything goes through linus". What level of attention linus gives it depends on how much he trusts the person sending it but every change to the main tree needs to be merged by him and then pushed to the public servers. I presume the requests to re-send pull requests were because he either wasn't making backups or his backups of the status of pull requests were out of sync with the backsup of the repo.
So it's like an employee workstation crash for the one employee that has the skills and authority to make your development "official".
He said the updates started at 3AM. Presumablly that is timed to be late enough that the night owls have gone home and early enough that under normal circumstances it would finish before the early birds get in.
The question would be why did it take 8 hours, was it something that MS screwed up, something his IT department screwed up or some combination of both.
The convertable tablet if well executed should cover both the usecase of the laptop and the usecase of the tablet leading to one less device to lug around.
Unfortunately so-far noones executed it well. Andriod isn't really suitable for laptops (notice how google's own laptops don't run it) because of a lack of multitasking/multiwindow. Windows RT has been gimped by the refusal to permit third party desktop apps*. Most x86 tablets have been too thick and heavy, the surface pro comes close but ideally it needs to get a bit thinner and lighter.
The article is postulating that haswell could change that and finally make a practical convertable tablet.
* Yes I know there is a hack to bypass this which may or may not continue to work in future. no I don't think it's reasonable for a customer to rely on such hacks.
It's pretty easy to go look at randomness and test it you know.
No it isn't. There are various tests you can do for particular types of pattern but beyond a certain size of seed data/key it becomes computationally infeasible to distinguish between a good CSPRNG/stream cypher and true randomness.
Supplying a stream of predictable but apparently data could cause the kernel to keep producing output on/dev/random even when it has no real entropy left with which to produce it. This could could theoretcally weaken keys generated on the system though I expect it would be very difficult to exploit this in practice as I would expect it would only make the difference between ludicrously strong and extremely strong.
If the water molecules themselves were radioactive then you would be right, however AIUI the majority of radioactivity in the wastewater from a contaminanted nuclear site like chernobyl or fukushima or even sellafield comes from disolved contaminents not from the water itself and those contaminants can be seperated from water.
you still have to store the crap you take out of the water and probablly come contaminated membranes but that is likely much easier than storing massive ammounts of water.
But in a theoretical pefrect wold of rainbow, unicorns, perfect crypto implementation and secure machine
And properly verified key management.
If the system works by having some authority tell clients both what network addresses they should connect to and which keys are and aren't valid for which other clients then the system is only as secure as that authority is.
The stock Intel CPU cooler can't come close to maintaining that limit.
I think you misintepreted it it "processor thermal soloution" is a fancy term for heatsink/fan combination (you see this with the xeons where the heatsink/fan units are sold seperately).
So what that is saying is that the temperature of the air the heatsink/fan takes from the case should be at or below 38C for the stock cooler to keep the processor within it's thermal specs (which according to intel for the i7-950 are a max Tcase of 67.9C.
Basically there are three main reasons to use a high end CPU cooler.
1: You will be operating the machine where ambient temperatures are high so you have less thermal margin for the cooler to work in 2: You want to make the machine quieter under load. A larger radiator area means that the fans can achieve acceptable cooling at much lower speeds 3: You want to seriously overclock. Serious overclocking can easilly double the power consumption of a CPU and that heat has to go somewhere.
If none of those apply to you then stick with the stock cooler.
At CD sampling rates (44.1kHz), you will have perfect reconstruction of any waveform that is bandwidth-limited at ~22 kHz
And a 15kHz sawtooth wave is certainly not a waveform that is "bandwidth limited at ~22kHz"
In real life the signals entering your system from the real world are NOT sharply bandwidth limited so you have to bandwidth limit them to avoid aliasing. There are two ways of doing this, the first is to use an aggressive a analogue filter and then sample at the rate you actually finally want. The other is to use a much less agressive analogy filter, sample at a higher sample rate and then filter and decimate the signal digitally.
Either way you will create artifacts. In particular there is a tradeoff that the more perfect a filter is in the frequency domain the more ringing it creates in the time domain.
Or if you don't care about data rate you could just stay in the higher sample rate throughout the system and avoid the need for agressive filters at all.
If your system had perfect filters then putting a 15kHz sawtooth wave in would result in a 15kHz sinewave out. Of course perfect filters can't exist so in reality you will also get a small ammount of alised signal in there too.
Does any of this matter to the human ear? that is for the audiophiles to argue over;)
Slashdot Mirror
The article is kinda vauge but AIUI they are talking about forcing american companies to store data about brazillians in brazil. Well that raises a few issues.
1: what are they going to do if some of those american companies tell them to go pound sand? Unless the company in question has a direct buisness presense in brazil it seems their choices are to either block connections (the "great firewall soloution) or lets things continue as they are.
2: If the NSA uses a national security letter to order the american parent company to get them some data on a brazillian what happens? Given a choice between breaking brazillian law and being punished for ignoring an american national security letter what do you expect an american company to do?
3: What happens when an amercian user and a brazillian user want/need to work together. For example suppose a brazillian user makes his calender accessible to his american friend. Should that calender be hosted in brazil? the USA? both? How will features like finding a slot where everything is free work if each country's citizens data has to remain within the country.
They are also taking about reducing reliance on the USA for connectivity to the rest of the world which just seems like good sense (afaict they already have one direct connection to europe but I doubt that is really enough) though it may well increase costs in the short term (US internet transit is CHEAP).
AIUI in the US only one of the major networks (t-mobile) gives users who bring their own phones a decent deal. Another (AT&T) doesn't forbid their use but structures their traffifs such that you effectively pay for a phone from them every couple of years whether you take them up on the offer or not. The remaining two use mostly "CDMA"* technology which doesn't use sims meaning you can only use phones the carrier will agree to activate. This may change with the introduction of LTE but AIUI that is not yet available in most places.
There are "virtual networks" out there too which can sometimes give better deals for those who want to bring their own phones to the AT&T network but AIUI they get lower priority and don't get to benefit from roaming arrangements AT&T has.
* Strictly speaking all modern mobile systems use CDMA modulation but there are still two sets of standards, those that grew out of the old CDMA standards and those that grew
AIUI the "Bureau of Engraving and Printing" (which is part of the treasury) physically prints it but the federal reserve issues it. Until it's issued it isn't legally money. In that sense saying that the federal reserve prints money is kind of like saying apple makes iphones.
However "printing" nowadays is mostly a metaphor anyway. Most of the money the federal reserve creates is just bits in a database but those bits in a database are legally equivilent to cash.
The treasury also creates a small ammount of money by issuing coins but that is even smaller than the ammount the federal reserve creates by issuing banknotes.
Host files have nothing to do with the kernel. Name resoloution on linux is handled by userland libraries.
Part A has essentially happened. Everything by valve that's not way-too-old-to-port has been linuxified.
If anything the opposite seems true. The really old goldsrc stuff (HL, opposing force, blue shift, CS, TF:C) seems to have been ported and so does the first generation source stuff that was kept up to date (HL2, HL2:EP1, HL2:EP2, HL2:DM, HL2:LC, portal, TF2, CS:S). I also notice DOTA 2 and deathmatch classic are aslo available on linux but I don't know what engine branches they used.
The games based on more recent branches of the source engine (alien swarm, left 4 dead, portal 2, CS:GO) do not appear to have linux releases. Nor do the source engine conversions of the original half life and counterstrike.
Any one person's contributions to climate change are pretty tiny. Even rich gits who fly around in private planes. Even national presidents who fly arround in a jumbo jet bringing with them a massive entourage. The number of rich git's in private planes is also fairly small so the total impact of all rich gits flying arround in private planes is probablly not that high in the grand scheme of things.
However why would the commoners listen to a rich git who effectively says "do as I say not as I do" when it comes to climate change? While the impact of all commoners changing their ways would probablly be bigger than the impact of all the rich gits changing their ways the impact of one individual commoner changing their ways is going to be even tinier than the impact of one rich git changing their ways.
There are various ways a power supply* can respond to an overcurrent situation including
1: Overheat and destroy itself (possiblly even catching fire though that is thankfully rare).
2: Enter a current limit mode where the voltage drops off sharply as current increases.
3: Trip out and refuse to supply any power until reset somehow.
Now suppose I have a nominally 1A power supply. This power supply goes into current limit at just over 1A and shorts the D+ and D- lines as per the USB battery charging specification. When I plug a tablet into it the tablet tries to charge at 2A but 2A isn't available so the voltage drops until the tablet is only drawing just over 1A. The tablet charges successfully albiet slower than it would with a 2A charger..
On the other hand suppose I had a another nominally 1A power supply but that tripped at just over 1A. I connect a USB cable to it and short the D+ and D- lines. When I plug a tablet into it the tablet would try to draw 2A and trip the protection. The tablet therefore doesn't charge at all.
Now supposed I had another nominally 1A power supply that overheats and destroys itself when attemting to supply more than 1A. I connect a USB cable to it and short the D+ and D- lines. When I plug my tablet in it would try to draw 2A and destroy the power supply.
That's like putting a restrictor plate on my car so that I can go faster then it allows me to go.
To take your car analogy imagine you had three cars. One doesn't have a rev limiter at all and you can destory the engine by over-revving it. One has a rev limiter that works in the conventional way limiting the engine speed to a predetermined level and the final one has a rev limiter that shuts down the whole car ifyou hit the limit.
* People reffer to them as "chargers" but the charge control circuitry is actually in the phone.
Afaict the problem is a practical latex document is likely to use a mixture of structural constants, "just put this where I damn well tell you to" constructs and constructs that while nominally strutural are being tweaked to make things look good on the printed page (for example moving a figure up or down in the text so it ends up on the page you want it on).
An automated conversion is likely to produce something that is just about readable but a high quality conversion is likely to require human judgement.
The "official" way is for the charger to short the data pins and then go into current limit if the device tries to draw more power than it can supply.
Which could cause problems for a charge only cable if it's used with a host port that shuts down rather than current limiting when overloaded.
I don't know if you are being serious but AIUI at least one of the electrodes is a consumable. So to maintain crude batteries you need not just a supply of electrolyte (the wine) but also a supply of refined metals.
Which is why you take the eeprom off the original controller board and put it on the new board. In my experiance the EEPROM is in an 8 pin package with a 1.27mm pin spacing (either a SOIC or a similar sized leadless package). Pretty easy to pop off with hot air (and yes I have done this several times).
Am I the only one who looks up the service manual for a laptop before buying it?
Of course many laptops do emergency hibernate when the battery status gets critical.
If emergency hibernate works it means you don't lose the work you had in progress when the battery fails but if it triggers too late then I could see it ending up with the drives power being cut while under heavy write load.
AIUI linux development works on the principle of "everything goes through linus". What level of attention linus gives it depends on how much he trusts the person sending it but every change to the main tree needs to be merged by him and then pushed to the public servers. I presume the requests to re-send pull requests were because he either wasn't making backups or his backups of the status of pull requests were out of sync with the backsup of the repo.
So it's like an employee workstation crash for the one employee that has the skills and authority to make your development "official".
2) Why are you running updates during work hours?
He said the updates started at 3AM. Presumablly that is timed to be late enough that the night owls have gone home and early enough that under normal circumstances it would finish before the early birds get in.
The question would be why did it take 8 hours, was it something that MS screwed up, something his IT department screwed up or some combination of both.
don't forget to keep a spare live CD handy in case your system becomes unbootable
I'd say being prepared for an unbootable system is a fairly normal part of preparing for a major version upgrade.
I'd also say that a major version upgrade is a very different thing from installing security updates.
The convertable tablet if well executed should cover both the usecase of the laptop and the usecase of the tablet leading to one less device to lug around.
Unfortunately so-far noones executed it well. Andriod isn't really suitable for laptops (notice how google's own laptops don't run it) because of a lack of multitasking/multiwindow. Windows RT has been gimped by the refusal to permit third party desktop apps*. Most x86 tablets have been too thick and heavy, the surface pro comes close but ideally it needs to get a bit thinner and lighter.
The article is postulating that haswell could change that and finally make a practical convertable tablet.
* Yes I know there is a hack to bypass this which may or may not continue to work in future. no I don't think it's reasonable for a customer to rely on such hacks.
It's pretty easy to go look at randomness and test it you know.
No it isn't. There are various tests you can do for particular types of pattern but beyond a certain size of seed data/key it becomes computationally infeasible to distinguish between a good CSPRNG/stream cypher and true randomness.
Supplying a stream of predictable but apparently data could cause the kernel to keep producing output on /dev/random even when it has no real entropy left with which to produce it. This could could theoretcally weaken keys generated on the system though I expect it would be very difficult to exploit this in practice as I would expect it would only make the difference between ludicrously strong and extremely strong.
If the water molecules themselves were radioactive then you would be right, however AIUI the majority of radioactivity in the wastewater from a contaminanted nuclear site like chernobyl or fukushima or even sellafield comes from disolved contaminents not from the water itself and those contaminants can be seperated from water.
you still have to store the crap you take out of the water and probablly come contaminated membranes but that is likely much easier than storing massive ammounts of water.
But in a theoretical pefrect wold of rainbow, unicorns, perfect crypto implementation and secure machine
And properly verified key management.
If the system works by having some authority tell clients both what network addresses they should connect to and which keys are and aren't valid for which other clients then the system is only as secure as that authority is.
The stock Intel CPU cooler can't come close to maintaining that limit.
I think you misintepreted it it "processor thermal soloution" is a fancy term for heatsink/fan combination (you see this with the xeons where the heatsink/fan units are sold seperately).
So what that is saying is that the temperature of the air the heatsink/fan takes from the case should be at or below 38C for the stock cooler to keep the processor within it's thermal specs (which according to intel for the i7-950 are a max Tcase of 67.9C.
Basically there are three main reasons to use a high end CPU cooler.
1: You will be operating the machine where ambient temperatures are high so you have less thermal margin for the cooler to work in
2: You want to make the machine quieter under load. A larger radiator area means that the fans can achieve acceptable cooling at much lower speeds
3: You want to seriously overclock. Serious overclocking can easilly double the power consumption of a CPU and that heat has to go somewhere.
If none of those apply to you then stick with the stock cooler.
Sorry I forgot to complete the last sentance.
That 192 kilobit per second is also a bitrate widely used for crappy compressed audio is merely a coincidence and unrelated to the topic at hand.
At CD sampling rates (44.1kHz), you will have perfect reconstruction of any waveform that is bandwidth-limited at ~22 kHz
And a 15kHz sawtooth wave is certainly not a waveform that is "bandwidth limited at ~22kHz"
In real life the signals entering your system from the real world are NOT sharply bandwidth limited so you have to bandwidth limit them to avoid aliasing. There are two ways of doing this, the first is to use an aggressive a analogue filter and then sample at the rate you actually finally want. The other is to use a much less agressive analogy filter, sample at a higher sample rate and then filter and decimate the signal digitally.
Either way you will create artifacts. In particular there is a tradeoff that the more perfect a filter is in the frequency domain the more ringing it creates in the time domain.
Or if you don't care about data rate you could just stay in the higher sample rate throughout the system and avoid the need for agressive filters at all.
If your system had perfect filters then putting a 15kHz sawtooth wave in would result in a 15kHz sinewave out. Of course perfect filters can't exist so in reality you will also get a small ammount of alised signal in there too.
Does any of this matter to the human ear? that is for the audiophiles to argue over ;)