I've got 8GB of RAM in the machine - RAM is so cheap now that it didn't seem worth skimping. It's a 1.6GHz AMD Fusion system. Over GigE, I was getting 40MB/s writes to the deduplicated filesystem, with the load on one core about 100%.
ZFS definitely likes RAM. I'm not sure what the minimum requirements are, but the general recommendation is 'as much as you can afford'. I think 8GB of SO-DIMMS for the mini-ITX board cost about £40, and maxed out its memory, so that was a pretty obvious choice. I'm not sure what happens when the deduplication tables don't fit into RAM, whether it degrades performance or degrades deduplication efficiency. Having 8GB means that a lot of the time it can satisfy reads from RAM.
I'm using it over WiFi 99% of the time, so I'm not too bothered about the performance: it can easily saturate the WiFi link without any problems. . The compression ratio is 1.11x. ZFS only shows the deduplication ratio for the entire pool, not for individual filesystems. That's currently 1.06x for my system, but that's with 1.43TB of data in total, only 266GB on the deduplicated filesystem, so that means that it's saving about a third. Roughly speaking, the extra space used by RAID-Z and the space saved by dedup seem to balance each other out, so (on my backup filesystem) I am using 1GB of hard disk space for every 1GB of data, and still have redundancy so one disk out of the three can fail without losing any data.
Time Machine on OS X does clever things like make a new copy of a 10GB file if 1MB of it has changed, and the deduplication on the NAS translates to a huge space saving for that. For things like DV footage, I don't bother with the dedup.
I have a NAS with a 1.6GHz AMD Fusion thingy, which should be in the same speed ballpark as an Atom. It happily got 40MB/s writing to the deduplicated filesystem with FreeBSD ZFS (with a kernel with all of the debug knobs turned on) over GigE. With a release kernel, I'd expect it to be a bit faster, but since I mainly use it over WiFi the bottleneck is generally somewhere else...
As I said in another post, ZFS development on FreeBSD is now funded by iXSystems. Given that most of their income is from selling large storage solutions built on top of FreeBSD and ZFS (often with a side order of FusionIO and other very expensive hardware things), they have a strong incentive to keep it stable and full of the features that their customers want.
Depends on your jurisdiction. In the UK, the limit is £5,000 (about $7800). In the USA, I believe it varies depending on the state. If it does go to a jury trial, it would cost PayPal a lot more than $2500 to fight it.
ZFS in FreeBSD 9 has deduplication support. I've been running the betas / release candidates on my NAS for a little while (everything important is backed up, so I thought I'd give it a test). ZFS development in FreeBSD is funded by iXSystems, who sell expensive NAS and SAN systems so they have an incentive to keep it improving.
I have a ZFS filesystem using compression and deduplication for my backups from my Mac laptop. I copy entire files to it, but it only stores the differences.
It depends on the app. If it's something like a game, where it's basically throw-away code that I hope works now and may not care much if it stops working in six months, then I agree. On the other hand, if it's something that I want to use long-term, then I do care. For example, I bought OmniGraffle for OS X for my PowerBook. It came as a PowerPC binary. OS X 10.7 doesn't include the PowerPC emulator, so if I want to open any of the OmniGraffle files then I have to either buy a new version, run OS X 10.6 in a VM, or run it on an old machine. In contrast, I can take pretty much any open source code I was running on the PowerBook and just recompile it. I can also usually take the open source applications I run on FreeBSD and compile them for the Mac, sometimes with minor tweaks.
I don't care about the fact that it's open source, but I do care about what the fact that it's open source means for long-term support. I'm happy to use some free and some proprietary software on a daily basis, but I won't depend on anything that isn't open source.
In any legal dispute, the person that you usually take to court is the person that you have a direct relationship with. In this case, the buyer gave the money to PayPal and PayPal then did not give it to the seller, having agreed to, or took back the money for spurious reasons. PayPal should therefore be taken to court.
Filing in a small claims court is usually very cheap and does not require a lawyer. The purpose of these courts is to allow low-value disputes to be resolved without involving the full legal process. File near you and PayPal has to send someone to your local court if they wish to defend it. If they don't defend then the judge or magistrate will rule based purely on your testimony.
Small claims courts do not usually expect either party to be a lawyer (taking a lawyer to a small claims court can often prejudice the judge or magistrate against you) and are not expected to have detailed legal knowledge. They are simply expected to state their grievance and allow the judge to decide what statue and common law is applicable. In this case, the buyer would state that, as a result of PayPal's actions, they do not have the violin worth $2,500, nor do they have the money, and so they have lost $2,500. The judge would then decide whether PayPal had acted correctly in this case.
Once you have a judgement, if PayPal refuses to pay then you can usually just hand it over to a collections agency. They will add something on top and require PayPal to pay the collections fee as well as the total amount of the judgement. If they still don't pay, then they will arrange to have PayPal assets confiscated and sold until the amount is reached.
Those of us who don't already live in the USA keep working on it. And we encourage the few of you who do to fix the laws in your country. Although not too loudly, because we're actually okay with the USA losing its competitive edge internationally...
Then it moves to a mirror in France and users are told to get it from there. Things like libavcodec violate a load of patents and can't legally be distributed in the USA, but they're fine pretty much everywhere else in the world.
It's also interesting for things that are only partly silk. Pashmina wool comes from the underside of the chin of the nepalese goat and is amazingly soft, but very fragile. You can't make things from it if you want to be able to wear them more than once, so you mix in some silk to add strength[1], but the more silk you add the more of the softness and warmth you lose. Stronger silk would mean that you could weave fabrics with a very small amount of silk and a lot of something softer.
[1] It amused me to see street sellers in NYC advertising pashmina shawls as '50% silk!' as if that was a good thing. The high quality ones are at most 20% silk. The silk is a lot cheaper than the wool.
1. Genetically modified spider escapes into the wild
They're genetically modifying silk worms, not spiders.
It mates with other sipders creating a legion of these spiders with super silk.
The silk is the same silk that spiders produce normally - that's the point.
New spider silk isn't as sticky as normal spider silk
Spiders produce two kinds of silk (massive oversimplification). The sticky stuff is relatively weak, the non-sticky stuff is used for structural parts of their webs. Go and poke a spiderweb sometime - you'll find some parts stick to you and tear easily, other parts don't stick and are tougher. Presumably the researchers are trying to make silk worms produce the non-sticky variety, as there is little call for silk that sticks to everything and tears very easily.
It's usually more expensive to build something that uses a small amount of power to achieve some desired effect than something that uses a lot of power to do the same thing. This means that, given two devices identical in every respect other than power consumption, users will just see the difference in the price tag and go for the less efficient one.
Additionally, the difference in power consumption needs to be quite big to make a difference. I worked out a while ago that I paid about £1 for 1W of always-on power for a year. So, if a device is using 5W in standby instead of 1W then it's costing me £4/year more. If it costs £20 more, then the saving probably isn't worth it because it's going to be five years before I see the money back and that's more than the typical operation lifespan of cheap consumer electronics (even if they're still working after that time, they've probably been superseded). Because of this, most devices don't bother advertising their power consumption.
Big appliances, like fridges, freezers and washing machines, can make a bigger difference. An efficient fridge can easily cost £10-20 less per year than a cheap one, and so you can make back the difference within a year. Things like this, however, are (by law) required to show a rated efficiency so that buyers can compare them. The minimum efficiency also goes up every year. I found my electricity usage dropped a lot after I moved from a rented flat, with landlord-supplied old and cheap appliances, to my own place where I bought new (but also quite cheap) appliances.
Active power consumption does, however standby consumption does not. And since most tablets don't maintain a connection to a cell tower when in standby mode, their standby consumption is even lower.
True, although not totally irrelevant to the question of government. If laws contained clear statements of intent in the comments then it would be much easier to judge whether they were working and either repeal or amend the ones that weren't...
I use a simple metric for judging code quality: how much of it do I have to read to understand what a single line is doing and how far away from the line is the furthest line that I have to read? In good code, I can usually fit everything I need in one editor window. In bad code, I need half a dozen files open and the distance is thousands of lines.
The why matters a lot. Good comments should be things like/* I chose this algorithm because I expect the data to meet these criteria. */ When you come to the code ten years after this was written and see that, in fact, the data don't match those criteria at all, you can replace it with a different algorithm. Or you can find that the data still do and so the algorithm makes sense even though its worst-case performance in the general case is terrible.
The sanity check is also useful. If the comments say the code does one thing, yet the code does something else, you've identified a bug. It's a lot better to use a proper specification language for the project than rely on this, but it's also at least an order of magnitude more time consuming...
Just because you're not still together doesn't mean you're not still friends. I still talk to some ex-girlfriends quite regularly, and I'd count them as friends, even close friends. You look for different qualities in a friend and a partner, and just because you turn out not to be compatible as a couple doesn't mean that you have to never talk to each other again. If you were together for a while, then you probably know each other better than most of your other friends know you, and once you're no longer together you can be a lot more objective about offering advice.
Just of the top of my head I can think of books like "Little Women" or "Jane Eyre" that happen about 150 years ago, where women are already able to work and support themselves, even if society is still not accepting it as "normal".
It's worth noting that most of these books were written by and about people at the top end of society. Being in trade was also unacceptable for men of this social rank. Having a profession was just about acceptable, but it was a sign of being a younger son (i.e. poor, relatively speaking). Among the rest of the population, women needed to work - even if it only meant helping their husband or father in his job - in between running the house, caring for children, and so on. Lots of servants were women, as were pretty much all teachers (that were not members of the clergy).
It was only the upper classes, where this was not true. For a long time, a big part of being a gentleman meant deriving your income through rents on land. If a man needed to work for a living, then he was not a member of this class. If a woman needed to work for a living then it meant that her father was not, or that he'd either mismanaged his holdings to the extent that he'd failed to care for her - neither was a good thing. Worse, it might mean that she'd married someone who wasn't a gentleman, and was so far from the station that not even his income from working in a profession was enough...
And if you don't restrict yourself to officially recognised ones, things like a first driving lesson and first pint of beer fill the same social role in a lot of western cultures.
Where did we go wrong that we ended up with software so fragile that you can't safely open just any document?
When we decided that, instead of web sites providing data, they should provide complex executable code. From the late '90s, there were two groups pushing the web in different directions. The group led by Tim Berners-Lee and the W3C wanted to define rich semantic markup languages so that different services could provide data that could then be interpreted in different ways by the client. The group backed by Google and others wanted to use the web as a deployment mechanism for huge blobs of mixed code and data that would be executed on the client and display the data as the author, not the reader, wanted.
The second approach intrinsically provides a much larger attack surface. Guess which one won.
Actually Steve Jobs too, since Wozniak was the brains behind the actual technology.
That's not totally true at Apple, and it certainly wasn't at NeXT. Jobs was very good at identifying trends. NeXT, for example, shipped the first workstation with a general purpose DMA controller because Jobs looked at how mainframes got better performance than workstations and said 'copy that idea' to his hardware people. The CPU wasn't stellar, but the fact that the network, disk, video and sound systems could all access memory without needing the CPU to get involved meant that a NeXT workstation with a 33MHz 68040 felt a lot faster than comparable machines of the same era.
Slashdot Mirror
I've got 8GB of RAM in the machine - RAM is so cheap now that it didn't seem worth skimping. It's a 1.6GHz AMD Fusion system. Over GigE, I was getting 40MB/s writes to the deduplicated filesystem, with the load on one core about 100%.
ZFS definitely likes RAM. I'm not sure what the minimum requirements are, but the general recommendation is 'as much as you can afford'. I think 8GB of SO-DIMMS for the mini-ITX board cost about £40, and maxed out its memory, so that was a pretty obvious choice. I'm not sure what happens when the deduplication tables don't fit into RAM, whether it degrades performance or degrades deduplication efficiency. Having 8GB means that a lot of the time it can satisfy reads from RAM.
I'm using it over WiFi 99% of the time, so I'm not too bothered about the performance: it can easily saturate the WiFi link without any problems. . The compression ratio is 1.11x. ZFS only shows the deduplication ratio for the entire pool, not for individual filesystems. That's currently 1.06x for my system, but that's with 1.43TB of data in total, only 266GB on the deduplicated filesystem, so that means that it's saving about a third. Roughly speaking, the extra space used by RAID-Z and the space saved by dedup seem to balance each other out, so (on my backup filesystem) I am using 1GB of hard disk space for every 1GB of data, and still have redundancy so one disk out of the three can fail without losing any data.
Time Machine on OS X does clever things like make a new copy of a 10GB file if 1MB of it has changed, and the deduplication on the NAS translates to a huge space saving for that. For things like DV footage, I don't bother with the dedup.
I have a NAS with a 1.6GHz AMD Fusion thingy, which should be in the same speed ballpark as an Atom. It happily got 40MB/s writing to the deduplicated filesystem with FreeBSD ZFS (with a kernel with all of the debug knobs turned on) over GigE. With a release kernel, I'd expect it to be a bit faster, but since I mainly use it over WiFi the bottleneck is generally somewhere else...
As I said in another post, ZFS development on FreeBSD is now funded by iXSystems. Given that most of their income is from selling large storage solutions built on top of FreeBSD and ZFS (often with a side order of FusionIO and other very expensive hardware things), they have a strong incentive to keep it stable and full of the features that their customers want.
Depends on your jurisdiction. In the UK, the limit is £5,000 (about $7800). In the USA, I believe it varies depending on the state. If it does go to a jury trial, it would cost PayPal a lot more than $2500 to fight it.
ZFS in FreeBSD 9 has deduplication support. I've been running the betas / release candidates on my NAS for a little while (everything important is backed up, so I thought I'd give it a test). ZFS development in FreeBSD is funded by iXSystems, who sell expensive NAS and SAN systems so they have an incentive to keep it improving.
I have a ZFS filesystem using compression and deduplication for my backups from my Mac laptop. I copy entire files to it, but it only stores the differences.
Ah, I didn't check, but the point of pride was what made me think it was a direct quote.
It depends on the app. If it's something like a game, where it's basically throw-away code that I hope works now and may not care much if it stops working in six months, then I agree. On the other hand, if it's something that I want to use long-term, then I do care. For example, I bought OmniGraffle for OS X for my PowerBook. It came as a PowerPC binary. OS X 10.7 doesn't include the PowerPC emulator, so if I want to open any of the OmniGraffle files then I have to either buy a new version, run OS X 10.6 in a VM, or run it on an old machine. In contrast, I can take pretty much any open source code I was running on the PowerBook and just recompile it. I can also usually take the open source applications I run on FreeBSD and compile them for the Mac, sometimes with minor tweaks.
I don't care about the fact that it's open source, but I do care about what the fact that it's open source means for long-term support. I'm happy to use some free and some proprietary software on a daily basis, but I won't depend on anything that isn't open source.
I am not a lawyer, but a few rules of thumb:
In any legal dispute, the person that you usually take to court is the person that you have a direct relationship with. In this case, the buyer gave the money to PayPal and PayPal then did not give it to the seller, having agreed to, or took back the money for spurious reasons. PayPal should therefore be taken to court.
Filing in a small claims court is usually very cheap and does not require a lawyer. The purpose of these courts is to allow low-value disputes to be resolved without involving the full legal process. File near you and PayPal has to send someone to your local court if they wish to defend it. If they don't defend then the judge or magistrate will rule based purely on your testimony.
Small claims courts do not usually expect either party to be a lawyer (taking a lawyer to a small claims court can often prejudice the judge or magistrate against you) and are not expected to have detailed legal knowledge. They are simply expected to state their grievance and allow the judge to decide what statue and common law is applicable. In this case, the buyer would state that, as a result of PayPal's actions, they do not have the violin worth $2,500, nor do they have the money, and so they have lost $2,500. The judge would then decide whether PayPal had acted correctly in this case.
Once you have a judgement, if PayPal refuses to pay then you can usually just hand it over to a collections agency. They will add something on top and require PayPal to pay the collections fee as well as the total amount of the judgement. If they still don't pay, then they will arrange to have PayPal assets confiscated and sold until the amount is reached.
The entire post (aside from the parenthetical comment at the end) was a direct quote from American Gods.
Those of us who don't already live in the USA keep working on it. And we encourage the few of you who do to fix the laws in your country. Although not too loudly, because we're actually okay with the USA losing its competitive edge internationally...
Then it moves to a mirror in France and users are told to get it from there. Things like libavcodec violate a load of patents and can't legally be distributed in the USA, but they're fine pretty much everywhere else in the world.
Mod parent up. Sad to see my post with its oversimplification warning sitting at +5 while this one with a much better explanation is at 0.
It's also interesting for things that are only partly silk. Pashmina wool comes from the underside of the chin of the nepalese goat and is amazingly soft, but very fragile. You can't make things from it if you want to be able to wear them more than once, so you mix in some silk to add strength[1], but the more silk you add the more of the softness and warmth you lose. Stronger silk would mean that you could weave fabrics with a very small amount of silk and a lot of something softer.
[1] It amused me to see street sellers in NYC advertising pashmina shawls as '50% silk!' as if that was a good thing. The high quality ones are at most 20% silk. The silk is a lot cheaper than the wool.
1. Genetically modified spider escapes into the wild
They're genetically modifying silk worms, not spiders.
It mates with other sipders creating a legion of these spiders with super silk.
The silk is the same silk that spiders produce normally - that's the point.
New spider silk isn't as sticky as normal spider silk
Spiders produce two kinds of silk (massive oversimplification). The sticky stuff is relatively weak, the non-sticky stuff is used for structural parts of their webs. Go and poke a spiderweb sometime - you'll find some parts stick to you and tear easily, other parts don't stick and are tougher. Presumably the researchers are trying to make silk worms produce the non-sticky variety, as there is little call for silk that sticks to everything and tears very easily.
It's usually more expensive to build something that uses a small amount of power to achieve some desired effect than something that uses a lot of power to do the same thing. This means that, given two devices identical in every respect other than power consumption, users will just see the difference in the price tag and go for the less efficient one.
Additionally, the difference in power consumption needs to be quite big to make a difference. I worked out a while ago that I paid about £1 for 1W of always-on power for a year. So, if a device is using 5W in standby instead of 1W then it's costing me £4/year more. If it costs £20 more, then the saving probably isn't worth it because it's going to be five years before I see the money back and that's more than the typical operation lifespan of cheap consumer electronics (even if they're still working after that time, they've probably been superseded). Because of this, most devices don't bother advertising their power consumption.
Big appliances, like fridges, freezers and washing machines, can make a bigger difference. An efficient fridge can easily cost £10-20 less per year than a cheap one, and so you can make back the difference within a year. Things like this, however, are (by law) required to show a rated efficiency so that buyers can compare them. The minimum efficiency also goes up every year. I found my electricity usage dropped a lot after I moved from a rented flat, with landlord-supplied old and cheap appliances, to my own place where I bought new (but also quite cheap) appliances.
Active power consumption does, however standby consumption does not. And since most tablets don't maintain a connection to a cell tower when in standby mode, their standby consumption is even lower.
True, although not totally irrelevant to the question of government. If laws contained clear statements of intent in the comments then it would be much easier to judge whether they were working and either repeal or amend the ones that weren't...
The communist utopia isn't a falsehood. It's just that humans aren't going to be allowed in...
I use a simple metric for judging code quality: how much of it do I have to read to understand what a single line is doing and how far away from the line is the furthest line that I have to read? In good code, I can usually fit everything I need in one editor window. In bad code, I need half a dozen files open and the distance is thousands of lines.
The why matters a lot. Good comments should be things like /* I chose this algorithm because I expect the data to meet these criteria. */ When you come to the code ten years after this was written and see that, in fact, the data don't match those criteria at all, you can replace it with a different algorithm. Or you can find that the data still do and so the algorithm makes sense even though its worst-case performance in the general case is terrible.
The sanity check is also useful. If the comments say the code does one thing, yet the code does something else, you've identified a bug. It's a lot better to use a proper specification language for the project than rely on this, but it's also at least an order of magnitude more time consuming...
Just because you're not still together doesn't mean you're not still friends. I still talk to some ex-girlfriends quite regularly, and I'd count them as friends, even close friends. You look for different qualities in a friend and a partner, and just because you turn out not to be compatible as a couple doesn't mean that you have to never talk to each other again. If you were together for a while, then you probably know each other better than most of your other friends know you, and once you're no longer together you can be a lot more objective about offering advice.
Just of the top of my head I can think of books like "Little Women" or "Jane Eyre" that happen about 150 years ago, where women are already able to work and support themselves, even if society is still not accepting it as "normal".
It's worth noting that most of these books were written by and about people at the top end of society. Being in trade was also unacceptable for men of this social rank. Having a profession was just about acceptable, but it was a sign of being a younger son (i.e. poor, relatively speaking). Among the rest of the population, women needed to work - even if it only meant helping their husband or father in his job - in between running the house, caring for children, and so on. Lots of servants were women, as were pretty much all teachers (that were not members of the clergy).
It was only the upper classes, where this was not true. For a long time, a big part of being a gentleman meant deriving your income through rents on land. If a man needed to work for a living, then he was not a member of this class. If a woman needed to work for a living then it meant that her father was not, or that he'd either mismanaged his holdings to the extent that he'd failed to care for her - neither was a good thing. Worse, it might mean that she'd married someone who wasn't a gentleman, and was so far from the station that not even his income from working in a profession was enough...
And if you don't restrict yourself to officially recognised ones, things like a first driving lesson and first pint of beer fill the same social role in a lot of western cultures.
Where did we go wrong that we ended up with software so fragile that you can't safely open just any document?
When we decided that, instead of web sites providing data, they should provide complex executable code. From the late '90s, there were two groups pushing the web in different directions. The group led by Tim Berners-Lee and the W3C wanted to define rich semantic markup languages so that different services could provide data that could then be interpreted in different ways by the client. The group backed by Google and others wanted to use the web as a deployment mechanism for huge blobs of mixed code and data that would be executed on the client and display the data as the author, not the reader, wanted.
The second approach intrinsically provides a much larger attack surface. Guess which one won.
Actually Steve Jobs too, since Wozniak was the brains behind the actual technology.
That's not totally true at Apple, and it certainly wasn't at NeXT. Jobs was very good at identifying trends. NeXT, for example, shipped the first workstation with a general purpose DMA controller because Jobs looked at how mainframes got better performance than workstations and said 'copy that idea' to his hardware people. The CPU wasn't stellar, but the fact that the network, disk, video and sound systems could all access memory without needing the CPU to get involved meant that a NeXT workstation with a 33MHz 68040 felt a lot faster than comparable machines of the same era.