INS has a lot of latitude in how they interpret and enforce existing laws. They have a number of places they look to for guidance, and the president is likely the largest. Trump is looking for statistics that show he is 'making a difference'. Cost to the economy doesn't appear to be a factor at all, only appearances.
One example I've seen recently is that my wife's company had a Mexican field support engineer denied entry recently, despite having a visa that was valid for his purpose (training). It sounds like he probably described the purpose for his visit incorrectly (along the lines of "I'm going to do some work for my US employer", exactly as he has for many entries over more than a decade, but which is not generally allowed on this visa), so the denial is not totally unreasonable, yet it points to immigration people who are currently looking for reasons to deny entry rather than looking for reasons to allow entry.
And appealing these decisions is difficult to the point of being almost impossible. This engineer may not ever be allowed into the US again under any VISA.
Thanks, very interesting. Once someone mentioned 'thermal neutrons' it made more sense. Whether or not this kind of thing actually turns out to be real, it is good to see scientists working on high-risk research.
I read the linked article, and he refers to "ultra-low-momentum neutrons with a huge capture cross-section". Has anyone dug deeper into this? How do you create these neutrons? Or is it just pseudo-science?
Seems like if you could actually do this, there are a lot of crazy things you could do. Such as creating unstable isotopes that decay into something else. Either the decay by-product could be valuable, or the energy released...
I mostly agree with you, but this is not the point. Unfortunately, the flip side of the amounts 'not being worth suing over' reduces the motivation of project originators to actually do what they say they will do. There is a real risk that Kickstarter will be overwhelmed by projects that are deceptive or fraudulent, like the Bonaverde Coffee Changers project.
It seems like Kickstarter will, in the longer run, be hurt by fraudulent projects and that they should have a more proactive approach to avoid being taken over by criminals. Perhaps they are waiting to see if this becomes a wide-spread problem before doing more. The fact that they have made changes to their rules indicates that some issues have been identified.
Of course, in the BV case it appears there may be applicable consumer protection laws, and possibly even criminal laws.
BTW, I was the Anonymous Coward poster above, I did not notice I was not logged in before posting.
This specific example is not useful, but for other units it would make tons of sense and be more useful than the Google results if it produces an answer to a unit multiplication question rather than giving a link to a site that will do it. Often with extra features come extra ways to get silly answers, this is one I would be willing to live with.
as a 'loser in 2007'. This was in their January 2007 Edition. Kind of interesting that Germany has already pulled out.
See:
http://www.spectrum.ieee.org/jan07/4842
There might be grounding problems with this setup, since each power supply will have a separate ground. The grounds for the power supplies will probably end up connected through the SATA or IDE cables, which could lead to major 'ground bounce' and kill signal integrity.
If he has a multiple power supply configuration, perhaps this is his 'dirty power'. Might be solved by creating a separate very low resistance connection (perhaps a low gauge copper wire) between the grounds on each power supply. Or perhaps the dual power supply cases come with a solution to this? I'm not familiar with these.
As someone else pointed out, heat could also be a problem. Overheated components can cause all kinds of weird problems.
Note that higher-end hardware RAID controllers use non-volatile RAM (usually battery powered) for the write-back cache. This prevents the protocol violation since it can survive a power problem so long as power is restored before the NV cache battery goes out. There are also redundant solutions with two RAID controllers that mirror their battery-backed cache to provide even higher reliability. The theoretical MTBF for a properly configured system with one of these is high enough that software bugs are much more likely to cause data loss than any combination of hardware failure.
Regarding: "The uptime isn't the reason for using RAID at home. Data integrity is."
and
"However, these days disks are big enough that a RAID 0 configuration is reasonable, and that's what I have now"
Perhaps you meant RAID 1? I agree that mirroring is often enough (I run this for my main PC drive), unless you want really large amounts of redundant storage (> 250GB or so with current drive prices).
If you really mean RAID 0, you need to know that it is not redundant and you will lose all of your data if any drive fails. RAID 0 is mainly a performance configuration. It can be combined with redundant RAID types (i.e. RAID 0+1/0+5) to get redundancy.
Perhaps I am misinterpreting your post, but it sounds like you may have a serious misconception that could lead to a real problem if any drive fails.
Note also that even if you have to power down to swap a RAID drive that fails, you still have the reliability. You just don't have the uptime, which is not important for many home users but critical for many companies.
I'm curious about why they mention floating point performance at all since it would seem that integer performance would be far more useful for just about anything Echelon needs to do.
Anyone know if there is a reason for the floating point reference other than just as a 'gee whiz' number?
All RAIDs have problems if a drive returns the wrong data. This is rare since the drives store ECC information for all data, but it does happen.
The only systems that protect against this (that I know of) put end-to-end CRCs on the data. This means a CRC is sent with the data, then read and checked with the data. This is usually implemented with 520 byte blocks and SCSI (which allows it), and is rather unusual. Very expensive equipment for this, such as from EMC or Hitachi.
RAID 5 has the same problem with losing data. It could detect the problem by reading the parity data and checking, but this would slow things down so much that most implementations don't even provide the option (a few do). RAID 1 can similarly detect the error by reading both drives. However, neither can tell which version is correct when they detect a problem.
Overall RAID 1 tends to deal better with errors both by having less of a performance hit and less chance of a implementation bug causing you to lost data (which often goes unnoticed since it is somewhere you don't end up seeing the error).
RAID-5s without NV cache are either unsafe (they don't protect their writes) or slow (they do their writes synchronously). This is why, for cheaper solutions, I would personally feel much more comfortable with RAID 1.
This is simply not true. There are trade-offs for all RAID implementations, and RAID 5 has significant disadvantages such as very poor write performance especially if you don't have non-volatile write cache. Read performance can also be less than RAID 1 for decent RAID 1 implementations (which are unfortunately rare).
Also, RAID-5 rebuilds are relatively slow and compute intensive (XOR computations). Low-end RAID-5 implementations can even have major data security problems due to the difficulty of getting good performance without sacrificing data safety. Software RAID-5s in particular are often not 100% safe.
As others have also said, I would strongly recommend RAID 1 for an individual situation unless you need a large (> 1TB) amount of space.
This is what I use on my personal PC, though I am considering setting up a RAID 5 server with a few TB of space. I have not, however, found a really good and inexpensive RAID-5 controller yet.
I am looking into Linux software RAID, but am not sure whether I can get reasonable write performance. I'm in particular not sure if it has support for NV write caching or what kind of NV to use if it does. Perhaps a solid state disk?
This is the right idea, and something I would love to be able to do with the computers on my home net. GB is now so cheap that it would be very practical to set up a home SAN.
iSCSI targets for Linux are already available, and could be used to share out parts of a RAID partition lots of options here.
It would be nice if we had NIC boot prom support for iSCSI. I wonder if the EtherBoot team has any plans to incorporate iSCSI? The adventurous could try to set up a Linux image to boot via tftp that then mounts its filesystems via iSCSI. Probably would require some kernel changes...
A simple iSCSI initiator is not that much code, less than an NFS client, so it could be put into the boot prom. iSCSI can be extremely cheap if done right, no need for a different protocol.
By presenting the code only in closed court, SCO can reduce the chance of the public in-depth analysis that has destroyed their previous "evidence". I suspect the closed court is mainly intended to improve their chances of making money from this lawsuit and reduce the damage when their claims are shown to be false.
Actually, there is no plan to convert carbon dioxide to hydrogen with any kind of organism. That would require nuclear transmutation, which so far as I know has never been done in a biological organism.
I think the plans are for a bacteria or virus that traps carbon dioxide or uses photosynthesis to convert it to carbon and oxygen (lots of stuff that already does this, of course). There are also known bacteria that can split water into hydrogen and oxygen. Sorry, I have not tracked down any references...
The problem with 'eating up all the CO2' is certainly a consideration, but likely not as big a deal as it may sound since the organism would be designed (like existing ones) to reach a balance at an appropriate level, possibly related to an inability to survive with less than a certain amount of CO2 in its environment.
Slashdot Mirror
INS has a lot of latitude in how they interpret and enforce existing laws. They have a number of places they look to for guidance, and the president is likely the largest. Trump is looking for statistics that show he is 'making a difference'. Cost to the economy doesn't appear to be a factor at all, only appearances.
One example I've seen recently is that my wife's company had a Mexican field support engineer denied entry recently, despite having a visa that was valid for his purpose (training). It sounds like he probably described the purpose for his visit incorrectly (along the lines of "I'm going to do some work for my US employer", exactly as he has for many entries over more than a decade, but which is not generally allowed on this visa), so the denial is not totally unreasonable, yet it points to immigration people who are currently looking for reasons to deny entry rather than looking for reasons to allow entry.
And appealing these decisions is difficult to the point of being almost impossible. This engineer may not ever be allowed into the US again under any VISA.
Thanks, very interesting.
Once someone mentioned 'thermal neutrons' it made more sense. Whether or not this kind of thing actually turns out to be real, it is good to see scientists working on high-risk research.
I read the linked article, and he refers to "ultra-low-momentum neutrons with a huge capture cross-section". Has anyone dug deeper into this? How do you create these neutrons? Or is it just pseudo-science?
Seems like if you could actually do this, there are a lot of crazy things you could do. Such as creating unstable isotopes that decay into something else. Either the decay by-product could be valuable, or the energy released...
Alchemy, anyone?
I mostly agree with you, but this is not the point. Unfortunately, the flip side of the amounts 'not being worth suing over' reduces the motivation of project originators to actually do what they say they will do. There is a real risk that Kickstarter will be overwhelmed by projects that are deceptive or fraudulent, like the Bonaverde Coffee Changers project.
It seems like Kickstarter will, in the longer run, be hurt by fraudulent projects and that they should have a more proactive approach to avoid being taken over by criminals. Perhaps they are waiting to see if this becomes a wide-spread problem before doing more. The fact that they have made changes to their rules indicates that some issues have been identified.
Of course, in the BV case it appears there may be applicable consumer protection laws, and possibly even criminal laws.
BTW, I was the Anonymous Coward poster above, I did not notice I was not logged in before posting.
This specific example is not useful, but for other units it would make tons of sense and be more useful than the Google results if it produces an answer to a unit multiplication question rather than giving a link to a site that will do it. Often with extra features come extra ways to get silly answers, this is one I would be willing to live with.
as a 'loser in 2007'. This was in their January 2007 Edition. Kind of interesting that Germany has already pulled out. See: http://www.spectrum.ieee.org/jan07/4842
There might be grounding problems with this setup, since each power supply will have a separate ground. The grounds for the power supplies will probably end up connected through the SATA or IDE cables, which could lead to major 'ground bounce' and kill signal integrity.
If he has a multiple power supply configuration, perhaps this is his 'dirty power'. Might be solved by creating a separate very low resistance connection (perhaps a low gauge copper wire) between the grounds on each power supply. Or perhaps the dual power supply cases come with a solution to this? I'm not familiar with these.
As someone else pointed out, heat could also be a problem. Overheated components can cause all kinds of weird problems.
Note that higher-end hardware RAID controllers use non-volatile RAM (usually battery powered) for the write-back cache. This prevents the protocol violation since it can survive a power problem so long as power is restored before the NV cache battery goes out. There are also redundant solutions with two RAID controllers that mirror their battery-backed cache to provide even higher reliability. The theoretical MTBF for a properly configured system with one of these is high enough that software bugs are much more likely to cause data loss than any combination of hardware failure.
Regarding: "The uptime isn't the reason for using RAID at home. Data integrity is."
and
"However, these days disks are big enough that a RAID 0 configuration is reasonable, and that's what I have now"
Perhaps you meant RAID 1? I agree that mirroring is often enough (I run this for my main PC drive), unless you want really large amounts of redundant storage (> 250GB or so with current drive prices).
If you really mean RAID 0, you need to know that it is not redundant and you will lose all of your data if any drive fails. RAID 0 is mainly a performance configuration. It can be combined with redundant RAID types (i.e. RAID 0+1/0+5) to get redundancy.
Perhaps I am misinterpreting your post, but it sounds like you may have a serious misconception that could lead to a real problem if any drive fails.
Note also that even if you have to power down to swap a RAID drive that fails, you still have the reliability. You just don't have the uptime, which is not important for many home users but critical for many companies.
I'm curious about why they mention floating point performance at all since it would seem that integer performance would be far more useful for just about anything Echelon needs to do.
Anyone know if there is a reason for the floating point reference other than just as a 'gee whiz' number?
All RAIDs have problems if a drive returns the wrong data. This is rare since the drives store ECC information for all data, but it does happen.
The only systems that protect against this (that I know of) put end-to-end CRCs on the data. This means a CRC is sent with the data, then read and checked with the data. This is usually implemented with 520 byte blocks and SCSI (which allows it), and is rather unusual. Very expensive equipment for this, such as from EMC or Hitachi.
RAID 5 has the same problem with losing data. It could detect the problem by reading the parity data and checking, but this would slow things down so much that most implementations don't even provide the option (a few do). RAID 1 can similarly detect the error by reading both drives. However, neither can tell which version is correct when they detect a problem.
Overall RAID 1 tends to deal better with errors both by having less of a performance hit and less chance of a implementation bug causing you to lost data (which often goes unnoticed since it is somewhere you don't end up seeing the error).
RAID-5s without NV cache are either unsafe (they don't protect their writes) or slow (they do their writes synchronously). This is why, for cheaper solutions, I would personally feel much more comfortable with RAID 1.
This is simply not true. There are trade-offs for all RAID implementations, and RAID 5 has significant disadvantages such as very poor write performance especially if you don't have non-volatile write cache. Read performance can also be less than RAID 1 for decent RAID 1 implementations (which are unfortunately rare).
Also, RAID-5 rebuilds are relatively slow and compute intensive (XOR computations). Low-end RAID-5 implementations can even have major data security problems due to the difficulty of getting good performance without sacrificing data safety. Software RAID-5s in particular are often not 100% safe.
As others have also said, I would strongly recommend RAID 1 for an individual situation unless you need a large (> 1TB) amount of space.
This is what I use on my personal PC, though I am considering setting up a RAID 5 server with a few TB of space. I have not, however, found a really good and inexpensive RAID-5 controller yet.
I am looking into Linux software RAID, but am not sure whether I can get reasonable write performance. I'm in particular not sure if it has support for NV write caching or what kind of NV to use if it does. Perhaps a solid state disk?
This is the right idea, and something I would love to be able to do with the computers on my home net. GB is now so cheap that it would be very practical to set up a home SAN.
iSCSI targets for Linux are already available, and could be used to share out parts of a RAID partition lots of options here.
It would be nice if we had NIC boot prom support for iSCSI. I wonder if the EtherBoot team has any plans to incorporate iSCSI? The adventurous could try to set up a Linux image to boot via tftp that then mounts its filesystems via iSCSI. Probably would require some kernel changes...
A simple iSCSI initiator is not that much code, less than an NFS client, so it could be put into the boot prom. iSCSI can be extremely cheap if done right, no need for a different protocol.
By presenting the code only in closed court, SCO can reduce the chance of the public in-depth analysis that has destroyed their previous "evidence". I suspect the closed court is mainly intended to improve their chances of making money from this lawsuit and reduce the damage when their claims are shown to be false.
Actually, there is no plan to convert carbon dioxide to hydrogen with any kind of organism. That would require nuclear transmutation, which so far as I know has never been done in a biological organism.
I think the plans are for a bacteria or virus that traps carbon dioxide or uses photosynthesis to convert it to carbon and oxygen (lots of stuff that already does this, of course). There are also known bacteria that can split water into hydrogen and oxygen. Sorry, I have not tracked down any references...
The problem with 'eating up all the CO2' is certainly a consideration, but likely not as big a deal as it may sound since the organism would be designed (like existing ones) to reach a balance at an appropriate level, possibly related to an inability to survive with less than a certain amount of CO2 in its environment.