It looks to me like it has two fans (presumablly intake and exhaust) with covers that have an opening on the bottom. That's probablly fine for keeping most weather out but not so good against high pressure jets.
The problem is if you put something that makes lots of heat in a sealed box not much large than the item and don't make any provision to get the heat out of the box it's just going to overheat. So in addition to the sealed box you are going to have to find/build a cooling system that can collect the heat from the components on the inside and radiate it to the outside without compromising the seal on the box and without being compromised itself by the conditions outside. I would expect some sort of liquid cooling to be appropriate.
Given the wide outside temperature range and the wide range of heat output of a moden PC I'd think you'd also have to design something in to stop it getting too cold, maybe a valve to divert the coolant to an internal radiator if the temperature dropped too low and an explicit heater as a last ditch measure.
AIUI the "clones" are totally different chips inside that present the same interface. So you can tell the difference by disolving the package and inspecting the die.
As a manufacturer of devices I can see a few defenses you can employ
1: avoid chips that hide their internal details from you. The FTDI clones are totally different internally from the genuine devices but because that is all hidden from it's difficult to tell the difference without decapping them. Something like a microcontroller is much harder to clone without the customer noticing. 2: avoid manufacturing in a country where faking things and substituting parts from non-approved sources are culturally accepted. 3: if you are big enough to justify it put in place a program of destructive testing of samples of incoming material. Especially if supply problems push you into buying from non-approved sources. 4: don't buy from non-approved sources just to save a few bucks.
Of course all these defenses will cost you money which is why many device builders don't do them.
Theres an expression "absense of evidence is not evidence of absense". This is especially true when dealing with small samples.
Not finding evidence of transmission during that period doesn't mean we should stop looking. The cost of following up all contacts of the trivial number of people with ebola outside the core outbreak countries is trivial compared to what the cost would be if one of those people acted as the seed for a significant outbreak.
This driver does not appear to make counterfeiting significantly more difficult, AIUI the counterfeit USB-Serial chips are microcontroller based and what the driver does is to rewrite one of it's IDs in a way that fails on the genuine chips but succeeds on the counterfiets. It will be trivial for the counterfieters to update their firmware to act like a genunine chip in this case.
What this driver did is make devices containing counterfeit chips mysteriously fail in the field long after deployment with no indication of why. Having a load of your devices suddenly and mysteriously fail in the field is not going to be good for your reputation.
Sure if you have any sense you try to keep counterfiet parts out of your supply chain but for the little guys that can be easier said than done. You are totally reliant on your distributors not to supply counterfeit parts and your manufacturing partners not to make unauthorised substitutions.
Politicians ultimately run our countries and therefore utlimately decide the strategy we take against such things.
Right now they are adopting a relatively light touch strategy, they are tracking known infected people and their contacts, setting up special precautions in hospitals and sending people to help treat the outbreak but they are not putting in place total travel bans or mandatory quarantines on people who have recently visited infected countries.
If the epidemic spreads to the west on a significant scale then people will blame those who chose to implement that light touch strategy.
Afaict in the core countries of the epedemic the problem is a lack of resources. Ideally you would use a new protective suit each time to minimise the risk of material transferring from outside to inside and you would work very slowly and carefully to make sure you didn't puncture the protective suit (remember medical treatment almost inevitablly involves sharp objects coming into contact with patients bodilly fluids but when you are short of time and materials you can't do that and still treat all your patients.
In cases like texas the problem seems to be that there is a delay between the infected patient showing up and the hospital realising what they are dealing with and how seriously they need to take things.
... or you could rely on that interference at the point of reception to form unique signals for every recevier by synthesizing waveforms from an array of transmitters, dynamically calculated based on the location of each transmitter + receiver + interference pattern. This would effectively give an unlimited number of signals within the same spectrum with no additional bandwidth (constrained only by the number of transmitters).>
MIMO techniques can be useful but they aren't as magic as you seem to think (and a nieve analysis would suggest)
In reality a couple of things limit your performance.
1: the channel estimation is nessacerally imperfect 2: the calculations needed to seperate out the virtual channels also tend to amplify the noise.
This is especially true if the antennas are close together.
This is what Artemis is now testing with their pCell tech, using a data-center of waveform-calculating servers + cheap low-power transmitters about the size and cost of a off-the-shelf wireless router:
Yeah distributed mimo has the potential for significant gains. It also has a significant cost as it needs very high bandwidth fixed line connections from the "waveform calculating servers" to the transcivers.
Be careful not to take marketing material too literally.
Another line of thought especially with more serious crimes is removal from society, if a criminal is locked up then it's very difficult for them to commit crimes against ordinary citizens.
Traditionally mobile phone networks worked as follows
SMS (at least with GSM) goes on the control channel, so if the phone can associate with the network it can almost certainly get a SMS through. Voice calls need data channels but those data channels are circuit switched and have priority over packet data. So provided you are outdoors (indoors multipath can screw stuff up) you can usually get an intelligable voice call through.. Packet data gets the capacity left over after circuit switched stuff (mostly voice, circuit switched data also exists in theory but is rarely used in practice) has taken the slots it needs. If a cell has capacity problems then it is packet data service that will bear the brunt of the problem. On the plus side web browsing, email etc are much more tolerant of jitter/loss than voice calls are so they sometimes work acceptablly indoors when voice call quality becomes unacceptablly poor.
By trying to run VOIP over packet data services you are giving yourself the worst of both worlds, the low priority of packet data services with the jitter/loss intolerance of voice communication.
Now with LTE they are moving away from circuit switching but I would still expect voice calls to have priority over generic packet data service. Also many areas don't have LTE coverage.
Your proposal may work if you live your whole life in an area with good uncongested LTE coverage. If you want to remain in communication in areas that don't have those things then forget about it.
I've taken two debian jessie systems successfully from systemd to sysvinit without running into the problems the author of that bug mentions.
There is probablly a valid issue underlying that bug but one should always be aware when reading bugs that they are often not as general as their author's make out.
Debian's offering of Gnome 3 and Systemd are not comparable. Gnome 3 is only the default desktop for people who just want to click through the installer.
It's also what everyone previously using gnome2 on squeeze would get on upgrade to wheezy or above.
A fork of gnome 2 did eventually make it back into debian but it wasn't in wheezy and you still have to manually remove all the gnome bits and replace them with their forks.
Even if one hangs on to sysvinit as one's init system, programs like Gimp on Debian now end up pulling in systemd libs.
Pulling in libs related to things you don't use isn't really anything new.
The bigger question IMO is to what extent will systems that don't use systemd as init be supported going forward? Will users of other init systems be treated as second-class citizens. When the technical comitee chose the default init system they refused to rule on this issue and it looks like the current GR is what this is about.
While an electric vehicle won't be zero vibration it will likely be much better than a vehicle based around a reciprocating internal combustion engine for serveral reasons.
1: IC engine vehicles need to "idle". That is run the engine with no load as part of the startup process and during short stops. Mechanisms running unloaded are much worse from a vibration point of view than the same ones running under load. Electric cars can simply turn the main motors off when stopped (accessories can use their own motors which are sized to be appropriate to the load in question). 2: a reciprocating engine by it's very nature has objects of substantial mass changing position (rather than simply rotating). They try to balance this of course but said balancing is often of limited effectiveness. An electric motor (or a gas turbine but those didn't catch on in cars for other reasons) by contrast only has rotating parts. 3: while it's true that electric motors do have an optimal speed they have a much wider band of usable performance than an internal combustion engine. This means that it is practical for electric cars to use a fixed-ratio transmission.
What's especially annoying is when you have some components designed in metric and some designed in imperial. So whicever you chose a proportion of components don't line up nicely with your grid.
Milk in the UK is a bit strange. milkmen supermarkets and conviniance stores associated with supermarket chains sell it in round multiples of 1 pint. Other shops sell it in round multiples of 500ml.
How do you know they are using "crap lead free solder". Doesn't ROHS have exemptions for this sort of stuff.
Also even if they are using lead free solder I don't think the situation is anywhere near as bad as you imply. I have plenty of stuff that still works which is 5-10 years old. Capacitor failure still seems to be a bigger problem than solder failure.
The problem is testing "every possible op code" is insufficiant, you would have to test every possible opcode/operand/register state combination since the condition for "evil behviour" may test on a tight combination of those. Doing so is compututationally infeasible.
Often manufacturers seem to do a really shitty job of marking their chips requiring you to hold them to the light in just the right way to read the bloody things.
I suspect you wouldn't even need to erase the markings to relabel those, just print the new markings in a way that was actually visible under normal conditions.
The scientists were well aware of the small quakes. The prosecution alledged that they should have known that this meant an elevated risk of a big quake and that their downplaying of said risks was sufficiently negligent to ammount to manslauter. The court agreed with the prosecution and convicted them but the sentances were apparently suspended until appeal.
Did the court convict because they took a detatched view and found the people truely negligent or was their verdict colord by rage and the need to find a scapegoat. That is why we have appeals.
Unfortunately I can't seem to find any information on whether the appeal was successful, a failure or still in-progress.
I don't mind paying $20 for a light bulb I'll probably never have to replace again.
With CFLs we were promised 10 times the life of incandescents, some live up to that but many don't even from supposedly reputable brands. Fully enclosed fitings with the lampholder above the bulb seem to be particular death to CFLs.
Yes in theory an LED bulb can last even longer than a CFL but i'd be very reluctant to pay such high bulb prices on a relatively unproven and easy to screw up (AIUI high power LEDS need careful cooling design and drive circuit failure is also an issue as it is with CFLs) tech.
Slashdot Mirror
It looks to me like it has two fans (presumablly intake and exhaust) with covers that have an opening on the bottom. That's probablly fine for keeping most weather out but not so good against high pressure jets.
That enclosure does not appear to be sealed, this is good from a cooling point of view but may be problematic if the environment is too wet.
Theres more to it than that.
The problem is if you put something that makes lots of heat in a sealed box not much large than the item and don't make any provision to get the heat out of the box it's just going to overheat. So in addition to the sealed box you are going to have to find/build a cooling system that can collect the heat from the components on the inside and radiate it to the outside without compromising the seal on the box and without being compromised itself by the conditions outside. I would expect some sort of liquid cooling to be appropriate.
Given the wide outside temperature range and the wide range of heat output of a moden PC I'd think you'd also have to design something in to stop it getting too cold, maybe a valve to divert the coolant to an internal radiator if the temperature dropped too low and an explicit heater as a last ditch measure.
AIUI the "clones" are totally different chips inside that present the same interface. So you can tell the difference by disolving the package and inspecting the die.
As a manufacturer of devices I can see a few defenses you can employ
1: avoid chips that hide their internal details from you. The FTDI clones are totally different internally from the genuine devices but because that is all hidden from it's difficult to tell the difference without decapping them. Something like a microcontroller is much harder to clone without the customer noticing.
2: avoid manufacturing in a country where faking things and substituting parts from non-approved sources are culturally accepted.
3: if you are big enough to justify it put in place a program of destructive testing of samples of incoming material. Especially if supply problems push you into buying from non-approved sources.
4: don't buy from non-approved sources just to save a few bucks.
Of course all these defenses will cost you money which is why many device builders don't do them.
Theres an expression "absense of evidence is not evidence of absense". This is especially true when dealing with small samples.
Not finding evidence of transmission during that period doesn't mean we should stop looking. The cost of following up all contacts of the trivial number of people with ebola outside the core outbreak countries is trivial compared to what the cost would be if one of those people acted as the seed for a significant outbreak.
This driver does not appear to make counterfeiting significantly more difficult, AIUI the counterfeit USB-Serial chips are microcontroller based and what the driver does is to rewrite one of it's IDs in a way that fails on the genuine chips but succeeds on the counterfiets. It will be trivial for the counterfieters to update their firmware to act like a genunine chip in this case.
What this driver did is make devices containing counterfeit chips mysteriously fail in the field long after deployment with no indication of why. Having a load of your devices suddenly and mysteriously fail in the field is not going to be good for your reputation.
Sure if you have any sense you try to keep counterfiet parts out of your supply chain but for the little guys that can be easier said than done. You are totally reliant on your distributors not to supply counterfeit parts and your manufacturing partners not to make unauthorised substitutions.
Politicians ultimately run our countries and therefore utlimately decide the strategy we take against such things.
Right now they are adopting a relatively light touch strategy, they are tracking known infected people and their contacts, setting up special precautions in hospitals and sending people to help treat the outbreak but they are not putting in place total travel bans or mandatory quarantines on people who have recently visited infected countries.
If the epidemic spreads to the west on a significant scale then people will blame those who chose to implement that light touch strategy.
Afaict in the core countries of the epedemic the problem is a lack of resources. Ideally you would use a new protective suit each time to minimise the risk of material transferring from outside to inside and you would work very slowly and carefully to make sure you didn't puncture the protective suit (remember medical treatment almost inevitablly involves sharp objects coming into contact with patients bodilly fluids but when you are short of time and materials you can't do that and still treat all your patients.
In cases like texas the problem seems to be that there is a delay between the infected patient showing up and the hospital realising what they are dealing with and how seriously they need to take things.
... or you could rely on that interference at the point of reception to form unique signals for every recevier by synthesizing waveforms from an array of transmitters, dynamically calculated based on the location of each transmitter + receiver + interference pattern. This would effectively give an unlimited number of signals within the same spectrum with no additional bandwidth (constrained only by the number of transmitters).>
MIMO techniques can be useful but they aren't as magic as you seem to think (and a nieve analysis would suggest)
In reality a couple of things limit your performance.
1: the channel estimation is nessacerally imperfect
2: the calculations needed to seperate out the virtual channels also tend to amplify the noise.
This is especially true if the antennas are close together.
This is what Artemis is now testing with their pCell tech, using a data-center of waveform-calculating servers + cheap low-power transmitters about the size and cost of a off-the-shelf wireless router:
Yeah distributed mimo has the potential for significant gains. It also has a significant cost as it needs very high bandwidth fixed line connections from the "waveform calculating servers" to the transcivers.
Be careful not to take marketing material too literally.
Another line of thought especially with more serious crimes is removal from society, if a criminal is locked up then it's very difficult for them to commit crimes against ordinary citizens.
Traditionally mobile phone networks worked as follows
SMS (at least with GSM) goes on the control channel, so if the phone can associate with the network it can almost certainly get a SMS through.
Voice calls need data channels but those data channels are circuit switched and have priority over packet data. So provided you are outdoors (indoors multipath can screw stuff up) you can usually get an intelligable voice call through..
Packet data gets the capacity left over after circuit switched stuff (mostly voice, circuit switched data also exists in theory but is rarely used in practice) has taken the slots it needs. If a cell has capacity problems then it is packet data service that will bear the brunt of the problem. On the plus side web browsing, email etc are much more tolerant of jitter/loss than voice calls are so they sometimes work acceptablly indoors when voice call quality becomes unacceptablly poor.
By trying to run VOIP over packet data services you are giving yourself the worst of both worlds, the low priority of packet data services with the jitter/loss intolerance of voice communication.
Now with LTE they are moving away from circuit switching but I would still expect voice calls to have priority over generic packet data service. Also many areas don't have LTE coverage.
Your proposal may work if you live your whole life in an area with good uncongested LTE coverage. If you want to remain in communication in areas that don't have those things then forget about it.
I've taken two debian jessie systems successfully from systemd to sysvinit without running into the problems the author of that bug mentions.
There is probablly a valid issue underlying that bug but one should always be aware when reading bugs that they are often not as general as their author's make out.
Debian's offering of Gnome 3 and Systemd are not comparable. Gnome 3 is only the default desktop for people who just want to click through the installer.
It's also what everyone previously using gnome2 on squeeze would get on upgrade to wheezy or above.
A fork of gnome 2 did eventually make it back into debian but it wasn't in wheezy and you still have to manually remove all the gnome bits and replace them with their forks.
Even if one hangs on to sysvinit as one's init system, programs like Gimp on Debian now end up pulling in systemd libs.
Pulling in libs related to things you don't use isn't really anything new.
The bigger question IMO is to what extent will systems that don't use systemd as init be supported going forward? Will users of other init systems be treated as second-class citizens. When the technical comitee chose the default init system they refused to rule on this issue and it looks like the current GR is what this is about.
While an electric vehicle won't be zero vibration it will likely be much better than a vehicle based around a reciprocating internal combustion engine for serveral reasons.
1: IC engine vehicles need to "idle". That is run the engine with no load as part of the startup process and during short stops. Mechanisms running unloaded are much worse from a vibration point of view than the same ones running under load. Electric cars can simply turn the main motors off when stopped (accessories can use their own motors which are sized to be appropriate to the load in question).
2: a reciprocating engine by it's very nature has objects of substantial mass changing position (rather than simply rotating). They try to balance this of course but said balancing is often of limited effectiveness. An electric motor (or a gas turbine but those didn't catch on in cars for other reasons) by contrast only has rotating parts.
3: while it's true that electric motors do have an optimal speed they have a much wider band of usable performance than an internal combustion engine. This means that it is practical for electric cars to use a fixed-ratio transmission.
You can do that if you aren't bothered about it being a rough approximation.
What's especially annoying is when you have some components designed in metric and some designed in imperial. So whicever you chose a proportion of components don't line up nicely with your grid.
the "second" is an SI unit :D
It is but as far as I can tell noone talks about kiloseconds or megaseconds. They talk about minuites, hours, days etc.
Milk in the UK is a bit strange. milkmen supermarkets and conviniance stores associated with supermarket chains sell it in round multiples of 1 pint. Other shops sell it in round multiples of 500ml.
How do you know they are using "crap lead free solder". Doesn't ROHS have exemptions for this sort of stuff.
Also even if they are using lead free solder I don't think the situation is anywhere near as bad as you imply. I have plenty of stuff that still works which is 5-10 years old. Capacitor failure still seems to be a bigger problem than solder failure.
The problem is testing "every possible op code" is insufficiant, you would have to test every possible opcode/operand/register state combination since the condition for "evil behviour" may test on a tight combination of those. Doing so is compututationally infeasible.
Often manufacturers seem to do a really shitty job of marking their chips requiring you to hold them to the light in just the right way to read the bloody things.
I suspect you wouldn't even need to erase the markings to relabel those, just print the new markings in a way that was actually visible under normal conditions.
The scientists were well aware of the small quakes. The prosecution alledged that they should have known that this meant an elevated risk of a big quake and that their downplaying of said risks was sufficiently negligent to ammount to manslauter. The court agreed with the prosecution and convicted them but the sentances were apparently suspended until appeal.
http://rt.com/news/italy-jail-...
Did the court convict because they took a detatched view and found the people truely negligent or was their verdict colord by rage and the need to find a scapegoat. That is why we have appeals.
Unfortunately I can't seem to find any information on whether the appeal was successful, a failure or still in-progress.
I don't mind paying $20 for a light bulb I'll probably never have to replace again.
With CFLs we were promised 10 times the life of incandescents, some live up to that but many don't even from supposedly reputable brands. Fully enclosed fitings with the lampholder above the bulb seem to be particular death to CFLs.
Yes in theory an LED bulb can last even longer than a CFL but i'd be very reluctant to pay such high bulb prices on a relatively unproven and easy to screw up (AIUI high power LEDS need careful cooling design and drive circuit failure is also an issue as it is with CFLs) tech.
I thought some were UV LEDs with a phosphor coating which gives better color spectrum but lower efficiency.