Not the grandparent, but this kind of setup is pretty trivial with ZFS on FreeBSD. Generally if you've got a small SSD (in comparison to your total pool size), it's a good idea to split it between L2ARC (read cache) and log device (write cache). The log device wants to be big enough to handle 10 seconds of writes to the disks. If that 20TB array is made of 10 2TB disks that can each handle 100MB/s sustained writes then about 10GB should be enough. The rest can be L2ARC. It's worth remembering that it does take RAM to store the metadata for the L2ARC, so don't skimp on RAM too much in this kind of setting.
That depends a bit on the filesystem. Having the journal on an SSD can be a big speed win for things involving lots of writes, as it lets your disk do what it does best: long sequences of contiguous writes without having to go back and write the journal again. A log device in ZFS does this and doesn't have to be very big (the maximum amount of data that you can write to the array in 10 seconds is enough) and lets the OS return from fsync very quickly (because once the data is on the SSD, it's safe).
It used to be fairly common for the high-end SCSI drives to say 'RAID optimized' on them, meaning that they'd do less sector remapping than consumer drives. Normally, if a drive detects a sector failing error checking, it will relocate the data elsewhere (if it can) and remap a spare sector from a small reserved section on the drive. This cripples RAID performance, because something that looks like it should be a contiguous write is on two of the drives but isn't on the third (for example), and all three have to finish the write and sync before you can do the next one. It's better to have the drive just report that the sector has failed and let the RAID controller deal with it, by also avoiding that sector on other disks.
There are also a lot of companies that collect the old drives and will, for a small (but often quite reasonable) fee, transfer data from an old tape to a new one. I did some consulting for one such company years ago: a big part of their business was still scanning bits of paper and putting them on microfiche, but they had a regular contract with a bank that would send them a tape from their mainframe every day and have their old minicomputer transcribe it to microfilm for archiving. They'd keep an eye out on eBay for various kinds of tape drives as they started to become cheap and then buy a small stash of them, so when someone came to them and said 'we have backups on this kind of tape and all the drives have died, please help!' they could find the right kind of drive, copy the tape, and put the data on DVD (or some newer kind of tape).
KHTML came from KDE. WebKit is an annoying name because it's used to mean both the Objective-C HTML framework (BSD licensed, purely Apple code) and the overall project that contains the WebKit framework and a load of other stuff. When most people say WebKit, they actually mean WebCore, which evolved from KHTML but has little KHTML code left. There's also JavaScriptCore, which is mostly Apple and includes a bytecode interpreter, a simple JIT, a CPS-based optimising JIT, and an LLVM-based even-more-optimising JIT. Chrome only ever used WebCore, along with their own JavaScript implementation.
That doesn't necessarily have to be for tracking, it can simply be that they want to encourage people to buy things. There are three steps of resistance that you need to overcome: creating an account, associating a credit card with it, deciding that a particular program is worth the money. If you require creating an account for free things then you're one step closer to getting people to hand over money for other things.
The why is because females can get married, get pregnant, quite their jobs after stealing 3 months of maternity leave with no plans to return, divorce the father and get child support for 18 years and alimony for the rest of her life.
All before the age of 17? Women in your country sound like they have very impressive time management.
Same here. I bought three 2TB drives for the NAS that I built just before the floods. I thought that I'd replace them with 4TB drives once the 4TB ones were as cheap as the 2TB ones that I'd bought. 4TB ones are still more expensive, but the 2TB disks are older than I'm comfortable with (and getting full - ZFS performance really starts to degrade at 90% capacity). The WD drives seem to be reasonably well regarded, but I don't know whether the Red ones are actually worth more to me than the Green. Most of the time I'm accessing the NAS over WiFi, so performance isn't an issue at all, but it's on 24/7.
It also looks like the 6TB drives are getting close to 4TB in terms of price per GB, so I'm wondering if it's worth skipping 4TB and buying 6TB drives now. Going from about 4TB to 12TB would give me lots of headroom, if it looked like I could trust the disks to last that long.
The big problem is that rotting trees produce a lot of methane, which is a much worse greenhouse gas than carbon dioxide. Burning the trees cleanly is a lot better for the atmosphere than letting them decompose. The good thing about trees is that they live a long time and absorb a lot of carbon in the meantime. Fast-growing evergreens are a good cheap way of getting carbon out of the atmosphere, as long as you do something sensible with the wood. If you keep replanting them, as long as you do so in ground that didn't previously contain much vegetation, you can even burn them for heat periodically and have a net reduction in carbon.
Because it's not born out internationally. You can see this at conferences, where universities from some countries (Israel, Iran and Korea, for example) have either an even or female-dominated mixture, but universities from the UK and US are predominantly male. Our gender balance in international students is quite close to even. There's nothing innate about girls' brains that puts them off, something cultural is. And it's not an acceptable conclusion to me that we're losing out on around 30% of the top applicants because something at school is putting them off.
If you want to fix the imbalance in STEM, fix this one first. There was a study a few years ago that showed that female primary school teachers who were insecure about maths were the biggest reason that girls were put off mathematical subjects. Girls, on average, develop empathy at a younger age and if they have a maths teacher who is not confident, then they pick up on it. If it's a male teacher, it doesn't have any effect. If it's a female teacher, then they learn that maths is hard for girls. Boys at that age tend to be totally oblivious to whether the teacher is unsure of the subject.
There are two ways to fix this: more men, or more women with a decent education in maths, in primary education. Given the shortage of the latter, more men seems an easy solution. Shame that we've decided that any man who wants to be a primary school teacher is pedophile...
I've seen zero indication it's for any reason beyond a lack of interest on the part of females
That's certainly true by the time that you get to university level, but the important question is why? One of my hats is to be responsible for computer science admissions at an all-women Cambridge college. From what we see from international applicants, it's pretty clear that there are cultural factors putting off women in the UK and US from the subject. We're losing out on some of the top talent because something is putting them off even considering the subject by the time they're 14-16 years old (applications are at 17, but A-level selection is at 15-16 and that's strongly influenced by GCSE choices at 13-14).
Depending on your locale, the purchase might be covered by distance selling regulations. In the UK, you have a few days in which you can cancel the order for any reason. Cancel the order citing their poor security practices as the reason, keep a copy of any correspondence, and forward it to your credit card company if they try to charge you anything.
If you can agree to contractual terms by clicking through some agreement, you can agree to "waive" your DNT setting
In the US and UK, the requirement for a contract to be enforceable in court is that the side wishing to enforce it must demonstrate that a meeting of minds has occurred. It's far from a binary decision. Some things, such as witnessed signatures at the bottom with each page initialed, have large amounts of case law backing them up, so you need a very strong argument if you want to discount them. For click-through licenses, there's a lot less case law and everything on the opposing side helps. If you can demonstrate that you have actively opted out of tracking and then been presented with a click-through license that, buried somewhere in legalese, there is a permission to track, it's easier to argue that the contract is invalid.
Either way, I am not sure what court is going to protect you from malicious actors that would not follow DNT.
The various European data protection offices would be a good bet.
We should be working on stopping the ability to track, not about making statements of intent for possible future litigation in a court of law.
Making it impossible to track means making clients indistinguishable, which is very hard. Making tracking without consent illegal is much easier, because the companies that you really worry about doing the tracking are the ones with large and expensive data centres where they can process the data, and these are nice big targets.
It's not the Cold War anymore. You don't have to pretend that any country that you don't like is communist. The hereditary dictatorship in North Korea is about as far as you can get from communism and stopped pretending to be communist some time ago. It still claims to be democratic though, so if you're going to object to political philosophies based on the buzzwords that dictators use, you should probably be complaining about democracy, not communism...
No. The nVidia drivers share around 90% of their code between all platforms (Windows, Linux, FreeBSD, Solaris) and the open source ones all use the Gallium framework, which is designed for portability from the ground up.
Modern GPU drivers require a set of services from the kernel, mostly related to memory management. They need to be able to get access to the device's I/O range in the physical address map and they need the kernel to grant access to texture memory in both main memory and the device. That's about all that they need from the kernel.
At the top, they need a state tracker that manages 3D API state (which is fairly minimal on modern APIs, as they aim to be stateless for performance reasons) and that translates the shader programs into some intermediate representation.
The majority of the device-specific driver code lives between these two layers, which are usually handled by abstraction layers so that they can be plugged into different APIs. You use the same Gallium driver with an OpenGL 2, OpenGL 3, OpenVG or Direct3D state tracker.
The only person who should be curating personal photos in Facebook is the profile owner.
You mean the person who clicked through the ToS that grant Facebook a perpetual, commercial, sublicenceable, license to use the photos however they wish? Including (as they've done in the last) licensing them to third parties to use in adverts?
The purpose of DNT was to demonstrate, in a measurable way, that people did not wish to be tracked. It was not intended as an enforcement mechanism, but as a statement of intent. It makes it very hard to argue in court that your click-through ToS permits tracking (or constitutes a meeting of minds at all), when the user has explicitly requested not to be tracked.
It depends a lot on the category of goods. Amazon was successful in part because their recommendation system did exactly what you and the grandparent are complaining about: it recommended things that were very similar to the thing that you'd just bought. This works well for books, music, and films / TV shows, because if you like one thing in one of these categories then you'll probably like other similar things in the same category. At the simplest level, if you just bought season 1 of a show, there's a good chance that you'll buy season 2. It doesn't work so well for things like cars or computers: if you've bought one laptop, then there's a very low chance that you'll want to buy a similar laptop next week.
For me, the quality of ads (meaning the probability that I'd actually click on them) went down a lot when Google started targeting ads at me, rather than at the content of the page that I was viewing. You don't need all of the stalker-like behaviour on ad networks to classify web pages, match them with relevant adverts, and show non-tracking ads.
I'm a bit surprised that there isn't a startup doing tracking-free ads. I bet a lot of people who use AdBlock would be willing to put in an exemption for a company that did not track and ran plain text only ads (you know, like the ads Google used to run, back when we all liked the relevant and non-annoying Google ads).
I always understood that the point of DNT was simply to advertise intent, so that in any future discussions, in or out of court, the tracking companies would not be able to claim any form of implicit consent. It doesn't matter that it's optional or unenforceable on a technical level, it matters that you can't track people who set the DNT header and then say 'well, they didn't object at the time...' when hit by a class-action lawsuit.
Always Europe at one end, Asia or the USA at the other end. I've been on one or two full flights from the USA, but I've also been on one where everyone in economy plus had a row of 3 seats to themselves, though economy was packed. Flying ANA to Japan there were quite a few empty seats.
I've flown first class before, but the value proposition isn't really there. Given the choice between flying first, or flying economy and keeping the price difference, I'd pick the latter (I'll happily fly first when someone else is paying and I don't have the choice of taking the money though). Economy (well, Economy Plus, but it's United, so Economy on any other airline) on the 787 was the first time I've been sufficiently comfortable in an economy seat to get productive work done - usually I just sleep or zone out and watch bad movies. The interesting thing was that the first and business sections didn't seem any different from the 777, only the cheap seats improved.
If this works the same way that it does in Europe, then even after you've gone through this you get a code valid for 30 days that you can give to another operator to port your number. This gives them a little window to try to change your mind and is a fairly good way to protest.
Slashdot Mirror
Not the grandparent, but this kind of setup is pretty trivial with ZFS on FreeBSD. Generally if you've got a small SSD (in comparison to your total pool size), it's a good idea to split it between L2ARC (read cache) and log device (write cache). The log device wants to be big enough to handle 10 seconds of writes to the disks. If that 20TB array is made of 10 2TB disks that can each handle 100MB/s sustained writes then about 10GB should be enough. The rest can be L2ARC. It's worth remembering that it does take RAM to store the metadata for the L2ARC, so don't skimp on RAM too much in this kind of setting.
That depends a bit on the filesystem. Having the journal on an SSD can be a big speed win for things involving lots of writes, as it lets your disk do what it does best: long sequences of contiguous writes without having to go back and write the journal again. A log device in ZFS does this and doesn't have to be very big (the maximum amount of data that you can write to the array in 10 seconds is enough) and lets the OS return from fsync very quickly (because once the data is on the SSD, it's safe).
It used to be fairly common for the high-end SCSI drives to say 'RAID optimized' on them, meaning that they'd do less sector remapping than consumer drives. Normally, if a drive detects a sector failing error checking, it will relocate the data elsewhere (if it can) and remap a spare sector from a small reserved section on the drive. This cripples RAID performance, because something that looks like it should be a contiguous write is on two of the drives but isn't on the third (for example), and all three have to finish the write and sync before you can do the next one. It's better to have the drive just report that the sector has failed and let the RAID controller deal with it, by also avoiding that sector on other disks.
There are also a lot of companies that collect the old drives and will, for a small (but often quite reasonable) fee, transfer data from an old tape to a new one. I did some consulting for one such company years ago: a big part of their business was still scanning bits of paper and putting them on microfiche, but they had a regular contract with a bank that would send them a tape from their mainframe every day and have their old minicomputer transcribe it to microfilm for archiving. They'd keep an eye out on eBay for various kinds of tape drives as they started to become cheap and then buy a small stash of them, so when someone came to them and said 'we have backups on this kind of tape and all the drives have died, please help!' they could find the right kind of drive, copy the tape, and put the data on DVD (or some newer kind of tape).
EA is pretty evil, but they're also pretty easy to avoid. Few people have jobs that require using or interacting with EA products.
KHTML came from KDE. WebKit is an annoying name because it's used to mean both the Objective-C HTML framework (BSD licensed, purely Apple code) and the overall project that contains the WebKit framework and a load of other stuff. When most people say WebKit, they actually mean WebCore, which evolved from KHTML but has little KHTML code left. There's also JavaScriptCore, which is mostly Apple and includes a bytecode interpreter, a simple JIT, a CPS-based optimising JIT, and an LLVM-based even-more-optimising JIT. Chrome only ever used WebCore, along with their own JavaScript implementation.
That doesn't necessarily have to be for tracking, it can simply be that they want to encourage people to buy things. There are three steps of resistance that you need to overcome: creating an account, associating a credit card with it, deciding that a particular program is worth the money. If you require creating an account for free things then you're one step closer to getting people to hand over money for other things.
The why is because females can get married, get pregnant, quite their jobs after stealing 3 months of maternity leave with no plans to return, divorce the father and get child support for 18 years and alimony for the rest of her life.
All before the age of 17? Women in your country sound like they have very impressive time management.
It also looks like the 6TB drives are getting close to 4TB in terms of price per GB, so I'm wondering if it's worth skipping 4TB and buying 6TB drives now. Going from about 4TB to 12TB would give me lots of headroom, if it looked like I could trust the disks to last that long.
The big problem is that rotting trees produce a lot of methane, which is a much worse greenhouse gas than carbon dioxide. Burning the trees cleanly is a lot better for the atmosphere than letting them decompose. The good thing about trees is that they live a long time and absorb a lot of carbon in the meantime. Fast-growing evergreens are a good cheap way of getting carbon out of the atmosphere, as long as you do something sensible with the wood. If you keep replanting them, as long as you do so in ground that didn't previously contain much vegetation, you can even burn them for heat periodically and have a net reduction in carbon.
Because it's not born out internationally. You can see this at conferences, where universities from some countries (Israel, Iran and Korea, for example) have either an even or female-dominated mixture, but universities from the UK and US are predominantly male. Our gender balance in international students is quite close to even. There's nothing innate about girls' brains that puts them off, something cultural is. And it's not an acceptable conclusion to me that we're losing out on around 30% of the top applicants because something at school is putting them off.
If you want to fix the imbalance in STEM, fix this one first. There was a study a few years ago that showed that female primary school teachers who were insecure about maths were the biggest reason that girls were put off mathematical subjects. Girls, on average, develop empathy at a younger age and if they have a maths teacher who is not confident, then they pick up on it. If it's a male teacher, it doesn't have any effect. If it's a female teacher, then they learn that maths is hard for girls. Boys at that age tend to be totally oblivious to whether the teacher is unsure of the subject.
There are two ways to fix this: more men, or more women with a decent education in maths, in primary education. Given the shortage of the latter, more men seems an easy solution. Shame that we've decided that any man who wants to be a primary school teacher is pedophile...
I've seen zero indication it's for any reason beyond a lack of interest on the part of females
That's certainly true by the time that you get to university level, but the important question is why? One of my hats is to be responsible for computer science admissions at an all-women Cambridge college. From what we see from international applicants, it's pretty clear that there are cultural factors putting off women in the UK and US from the subject. We're losing out on some of the top talent because something is putting them off even considering the subject by the time they're 14-16 years old (applications are at 17, but A-level selection is at 15-16 and that's strongly influenced by GCSE choices at 13-14).
Depending on your locale, the purchase might be covered by distance selling regulations. In the UK, you have a few days in which you can cancel the order for any reason. Cancel the order citing their poor security practices as the reason, keep a copy of any correspondence, and forward it to your credit card company if they try to charge you anything.
If you can agree to contractual terms by clicking through some agreement, you can agree to "waive" your DNT setting
In the US and UK, the requirement for a contract to be enforceable in court is that the side wishing to enforce it must demonstrate that a meeting of minds has occurred. It's far from a binary decision. Some things, such as witnessed signatures at the bottom with each page initialed, have large amounts of case law backing them up, so you need a very strong argument if you want to discount them. For click-through licenses, there's a lot less case law and everything on the opposing side helps. If you can demonstrate that you have actively opted out of tracking and then been presented with a click-through license that, buried somewhere in legalese, there is a permission to track, it's easier to argue that the contract is invalid.
Either way, I am not sure what court is going to protect you from malicious actors that would not follow DNT.
The various European data protection offices would be a good bet.
We should be working on stopping the ability to track, not about making statements of intent for possible future litigation in a court of law.
Making it impossible to track means making clients indistinguishable, which is very hard. Making tracking without consent illegal is much easier, because the companies that you really worry about doing the tracking are the ones with large and expensive data centres where they can process the data, and these are nice big targets.
It's not the Cold War anymore. You don't have to pretend that any country that you don't like is communist. The hereditary dictatorship in North Korea is about as far as you can get from communism and stopped pretending to be communist some time ago. It still claims to be democratic though, so if you're going to object to political philosophies based on the buzzwords that dictators use, you should probably be complaining about democracy, not communism...
No. The nVidia drivers share around 90% of their code between all platforms (Windows, Linux, FreeBSD, Solaris) and the open source ones all use the Gallium framework, which is designed for portability from the ground up.
Modern GPU drivers require a set of services from the kernel, mostly related to memory management. They need to be able to get access to the device's I/O range in the physical address map and they need the kernel to grant access to texture memory in both main memory and the device. That's about all that they need from the kernel.
At the top, they need a state tracker that manages 3D API state (which is fairly minimal on modern APIs, as they aim to be stateless for performance reasons) and that translates the shader programs into some intermediate representation.
The majority of the device-specific driver code lives between these two layers, which are usually handled by abstraction layers so that they can be plugged into different APIs. You use the same Gallium driver with an OpenGL 2, OpenGL 3, OpenVG or Direct3D state tracker.
The only person who should be curating personal photos in Facebook is the profile owner.
You mean the person who clicked through the ToS that grant Facebook a perpetual, commercial, sublicenceable, license to use the photos however they wish? Including (as they've done in the last) licensing them to third parties to use in adverts?
The purpose of DNT was to demonstrate, in a measurable way, that people did not wish to be tracked. It was not intended as an enforcement mechanism, but as a statement of intent. It makes it very hard to argue in court that your click-through ToS permits tracking (or constitutes a meeting of minds at all), when the user has explicitly requested not to be tracked.
It depends a lot on the category of goods. Amazon was successful in part because their recommendation system did exactly what you and the grandparent are complaining about: it recommended things that were very similar to the thing that you'd just bought. This works well for books, music, and films / TV shows, because if you like one thing in one of these categories then you'll probably like other similar things in the same category. At the simplest level, if you just bought season 1 of a show, there's a good chance that you'll buy season 2. It doesn't work so well for things like cars or computers: if you've bought one laptop, then there's a very low chance that you'll want to buy a similar laptop next week.
For me, the quality of ads (meaning the probability that I'd actually click on them) went down a lot when Google started targeting ads at me, rather than at the content of the page that I was viewing. You don't need all of the stalker-like behaviour on ad networks to classify web pages, match them with relevant adverts, and show non-tracking ads.
I'm a bit surprised that there isn't a startup doing tracking-free ads. I bet a lot of people who use AdBlock would be willing to put in an exemption for a company that did not track and ran plain text only ads (you know, like the ads Google used to run, back when we all liked the relevant and non-annoying Google ads).
I always understood that the point of DNT was simply to advertise intent, so that in any future discussions, in or out of court, the tracking companies would not be able to claim any form of implicit consent. It doesn't matter that it's optional or unenforceable on a technical level, it matters that you can't track people who set the DNT header and then say 'well, they didn't object at the time...' when hit by a class-action lawsuit.
Always Europe at one end, Asia or the USA at the other end. I've been on one or two full flights from the USA, but I've also been on one where everyone in economy plus had a row of 3 seats to themselves, though economy was packed. Flying ANA to Japan there were quite a few empty seats.
I've flown first class before, but the value proposition isn't really there. Given the choice between flying first, or flying economy and keeping the price difference, I'd pick the latter (I'll happily fly first when someone else is paying and I don't have the choice of taking the money though). Economy (well, Economy Plus, but it's United, so Economy on any other airline) on the 787 was the first time I've been sufficiently comfortable in an economy seat to get productive work done - usually I just sleep or zone out and watch bad movies. The interesting thing was that the first and business sections didn't seem any different from the 777, only the cheap seats improved.
If this works the same way that it does in Europe, then even after you've gone through this you get a code valid for 30 days that you can give to another operator to port your number. This gives them a little window to try to change your mind and is a fairly good way to protest.