All of your objections are against Linux distributions, not Linux. Some of your complaints are also highly subjective, or demonstrate a lack of understanding of why certain functionality is there or not. For someone aiming to replace Windows, for example, key combinations probably ought to match Windows. However, most of us couldn't care less. And most Windows users don't know many of those key combinations anyway... For distributions that care, making the relevant key combinations work as expected isn't a big deal.
As for partitioning, it depends on how it's done. Proper partitioning makes a huge difference - you don't have non-essential stuff filling your disk at the wrong moment, for example. A typical example is that you really don't want your database to grind to a halt because someone "forgot" to remove a bunch of temporary files from their home directory after doing maintenance etc.. Partitions are a tool, and you can do stupid things with partitioning like with any other tool. For a desktop partitioning is less important, yes, and you will find that several distributions use partitioning very sparingly for that reason.
As for getting rid of virtual memory... Are you insane? Most people are running systems with far too little RAM. Without virtual memory their systems would be useless. Besides, you CAN easily turn it off - just switch off swapping ("man swapoff" in your hated console). I've run systems with swap turned off, but only embedded systems where I knew the memory requirements exactly. You'd be surprised how badly many programs handle running out of memory. For a desktop I'd bet most people would much rather have things slow down than randomly stop working, crashing, or keep giving them errors about running out of memory.
Snapshots ought to be available easily, at any moment in time, without taking much space. ZFS does so, by only storing the changes and sharing the unmodified data. If you want to do so, you need an abstraction of the hardware. That is, crossing layers. Not to mention writeable snapshots.
No. You need an API that lets you ask for a snapshot, and that handles it on a block level. You can "cross layers" if you absolutely want to, but why? It doesn't buy you anything, and it creates interdependencies. Why should the snapshot functionality have to be copied or rewritten every time you create a new filesystem? It doesn't need to depend on filesystem organization.
Adding new drives without partitioning, slicing, formatting. Just adding to the existing pool. Inclusive striping being adapted automagically. This needs a cross-layer interface, right ?
Of course it needs a "cross-layer interface". The entire point of layering is to expose an interface to the layer above that provides services but hides HOW those services are implemented. Nobody is complaining about interfaces from one layer to the other, but about removing the barriers between layers and merging them together in the same codebase, which, again, creates dependencies and reduces genericity where there's no reason to.
The transactional filesystem guarantees uncorrupted data at power failures and OS crashes. If you do this across a pool of physical platters, you need operations across layers.
Or you create a generic API for the lower layers that expose the functionality that is needed, and gets that promoted into the generic part of the appropriate layers so that all filesystems that want to have the hooks needed to implement similiar functionality without having to duplicate lots of code.
There may be arguments for considering changes in where the boundaries between layers are drawn, or even if a specific layer is really needed. But layering is there in part because a system like Linux is a large ecosystem. We have dozens of filesystems. We have specialized replacements or additions to the lower layers (such as a choice of using software raid or not, or a block device that mirrors over a network or not etc.) etc. The number of possible combinations is huge. Cutting through the layers is a lazy way of adding extra features to a single filesystem instead of making parts of those features available to other parts of the ecosystem.
The excuses just don't cut it - the reason ZFS cuts through the layers isn't because it needs to, but because it was convenient for the people writing ZFS without thinking holistically. In other words, they were concerned with looking only at ZFS, rather than looking at how to make the featureset they are supporting the most useful.
That's fair enough - for Sun it makes sense. They have no real incentive for exposing all this functionality as separate interfaces for the different layers so it can be built on by others. But trying to pretend they're doing it that way because it's the right thing to do is just silly.
FUSE lets you do file systems in user space on Linux. It works fine, but it's definitively slower than the in kernel alternative, which is probably the biggest reason why it's not used by default. Since you can compile the filesystems as modules, the only thing it buys you is protection against crashes. But if the code managing your filesystem keeps crashing, you have a significant problem anyway. Not many people are in a situation where an occasional full crash is a huge deal, and not many people would accept running a filesystem that would keep crashing anyway, so it's a fairly special case that doesn't matter enough to most people.
I'd love if it was an easy choice - if it was easy to write filesystems for Linux so you had the option of either loading them as modules into the kernel or running them user space. The same would be great for other subsystems too. For some, the tradeoff in performance is acceptable if it buys any extra stability.
I doubt it will happen with the mainstream Linux kernel as long as Linus is in charge, though. A more likely scenario would be that someone retrofit Linux on top of one of the experimental microkernels (several projects do that already) and selectively move subsystems into separate processes (don't know if anyone is doing that yet).
I just recently found a long streak of blood thats about two years old on the side of one of our kitchen cabinets. It was partially obscured, and it's from my wife after she cut herself badly on a broken glass pane by slipping and pushing her hand through one of the glass panes in the kitchen door... At the time there was spatter on the kitchen walls and pools of blood in our hallway from the time it took before I managed to stop the bleed while waiting for the ambulance... It took weeks of regular cleaning to get most of the stains off, and we kept finding more for months, and now that streak two years later... Finding blood alone means nothing.
And in the Reiser case the police apparently still don't know for sure whose blood it is.
It's earlier been reported that Hans Reiser early started refusing to cooperate with the police. Something which I'd consider very reasonable if he had the slightest suspicion the police were seriously considering him a suspect. Anything he tells them would potentially come back to haunt him if they find inconsistencies or can turn around and fit what he tells them into a more believable scenario for a jury. If I'd been suspected of a crime or arrested - innocent or not - the first thing I'd do would be to shut my mouth and only open it when my lawyer tells me to.
As such, we have NO basis for saying anything about the car seat. We don't know when it was removed. We don't know if Hans Reiser knows where it is. We don't know if he has a plausible explanation for what happened to it.
As for the blood, we don't even know who the blood came from - to my knowledge the police have only stated they haven't been able to rule out that it is from Nina Reiser.
Lets rephrase that a bit: She was in the middle of a custody battle that she might worry about losing after various allegations of hers didn't seem to stick, and now he is in jail and the children are with her family in Russia and won't be coming back, even though they're wanted to testify in the case.
There's at the very least a chance she stage her own disappearance to get Hans in trouble and went back to Russia and got her kids brought there.
There's enough examples of serial killers confessing to many but not all of their murders through history for whatever reasons. In some cases they'll confess later, or they will try to use it as leverage to get something they want (like attention). In others they'll first confess when faced with evidence. For that matter, it is not unusual for serial killers to confess to murders they haven't committed too. In Scandinavia there was a case a few years back where a convicted serial killer kept confessing to more murders, some of which he clearly hadn't carried out, some which he had.
There's simply no basis for thinking a likely serial killer's claims to have killed or not killed a specific person are reliable.
In this case even less so, since he's had a lot of grievances against Hans Reiser and might very well have seen not confessing as an opportunity to get rid of Hans or just to have some fun with the police.
If she's alive, she's likely in Russia, where here kids and family is. I doubt she'd need to be very careful to hide in Russia - enough money to buy some fake papers would be enough.
However in this case the children were then subsequently sent to Russia, to her relatives. There was an ongoing custody battle and a nasty divorce. The defense is likely going to claim that it is possible she staged it to be able to get her children back to Russia where she can be reunited with them without any risk of Hans getting full or partial custody. That she just shortly before disappearing got a Russian passport for her second child is certainly going to be used.
In this case, there is no indication the children were "left behind". And mothers flee relationships often enough (as do men, of course), with or without their children that any claim that her disappearance alone is proof she's dead are unlikely to fly.
In any case, a lot of serial murderers don't confess to all their murders, and some even confess to murders they haven't committed, for a lot of different reasons. The defense will have an easy time killing any idea the prosecutor will try to put forward about the confession of these murders meaning that Sturgeon can somehow be trusted to tell the truth about Nina Reiser, as there are lots of examples to counteract any claim to "common sense".
But having broken up with a confessed serial killer who previously sued your ex-husband and allegedly (according to Reiser) has previously threatened both your husband, children and your husbands mother, certainly will create reasonable doubt unless they have some pretty damning evidence. The defense will likely claim Nina Reiser might have been killed as revenge for leaving him, and Reiser framed as a way of making good on his alleged threats to harm Hans, his children and mother (by getting Hans locked up and effectively leaving the children with no parents).
Unless there are clear physical evidence linking Hans to it that can't have been tampered with by a psychopath or sociopath who had a relationship with Nina and a close business relationship with Hans and might thus have known a lot of details about their lives and who might even had opportunities to copy house keys etc., there's a lot of room for a hell of a lot of doubt.
Sean Sturgeon and Hans Reiser had a business relationship from 1999 to 2002. The BDSM stuff was brought up by Reiser in court proceedings as part of a lawsuit by Sean Sturgeon regarding a loan that Reiser allegedly didn't pay back. Reiser alleged that Sturgeon and Nina Reiser were having an affair at the time, and that the money Namesys loaned were mostly spent on Nina Reiser with Sturgeons full knowledge. He further alleged that Sturgeon threatened him and claimed he would hurt both Reiser, Reisers children and mother if he didn't get the money back.
All of these allegations came before Nina Reiser disappeared, and are well documented (lots of press + the court proceedings themselves). If anything, I'd expect the defense team to bring all that up, and present all that as motive - either for murder (Nina Reiser had a new boyfriend) or as a setup to frame Hans Reiser.
If you'd RTFA, you might also have noticed that testimony in one of the preliminary hearings stated that Nina Reiser broke off the relationship with Sturgeon because she was unhappy with his BDSM tendencies. If that's the case, you'd think she'd have brought up things like that in the rather nasty divorce proceedings if Hans Reiser was into it too.
All of these allegations came before Nina Reiser disappeared, and are well documented (lots of press + the court proceedings themselves). If anything, I'd expect the defense team to bring all that up, and present all that as motive - either for murder (Nina Reiser had a new boyfriend) or as a setup to frame Hans Reiser.
They all do that... Years ago the company I worked for rented some space from a colo facility that also hosted Yahoo and Verisign, and they went into great detail about the exact security precautions those companies had asked for.
If I ever decide to go for a life of crime, targeting colo facilities would be high on my list - they love showing off their customers and describe in detail how their infrastructure is designed and what customers are where, including showing where the cable gates are etc. Makes their high fences and guards kind of useless when their sales guys can happily be talked into giving a grand tour to a hacker with some nice wireless gadgets to connect in the right places.
Birth rates are already well below maintenance levels in most industrialized countries, and even China is set to see it's population peak soon due to the one child policy. The solution to the problem of too high growth is helping developing countries out of poverty.
We're maybe as little as a century away from actually seeing the worlds population shrinking unless we start increasing lifespans a lot faster than we have.
Limiting fraud to within a minute after getting a user to enter their details on a fake form is huge though. I've run billing system handling a million dollars a month in payments, and we had people from all over the world hammering us with US credit card numbers, apparently mainly to try to test whether the card numbers were still valid (as the service we offered would be useless to them but was cheap enough to be worthwhile to test with). A fairly high percentage of those cards were already reported stolen/lost, so the numbers were clearly not new. Whenever someone would manage to get through our fraud prevention systems and we got chargeback documentation there were often already several days worth of fraudulent transactions reported for the card.
Yes it's a good thing, but RTFA. Microsoft was one of a group of companies that wanted SCOTUS to make this decision. Obvious patents cost Microsoft far more in litigation than what they can expect to lose in licensing revenue.
Or he is hinting at an application for proper 3D. I don't go to the theatre often, but I'm perfectly aware how different the experience is. I've also seen enough filmed plays to know that getting proper 3D effects would make it a hell of a lot better. For one, being able to change focus and concentrate on different parts of the scene does make a lot of difference. So while a lot of movies wouldn't gain all that much from proper 3D, there are a lot of things that would.
For an individual, require purchase with a credit card, and send the letter to the billing address of the credit card, which can be verified during credit card verification.
That's true for only a handful or so countries in the world, and even for most of those the address verification services are woefully incomplete (that is, you have to expect them to return no result or wrong result for a fairly large percentage of users).
It's more like 100,000 to 10,000,000 depending on how expensive chips you use, and it's per erase unit, and it's not writes but erases. The latter two points are important, because it means that even if erase units are large, you can still write to the erase unit multiple times (until you've filled it), so you can split it into convenient block sizes, and you'd typically either use a flash translation layer or a flash specific filesystem that avoids erasing the unit until there are no free blocks elsewhere to minimize writes. In practice that means you have to have either an extremely high write load or an almost full drive with a medium level of writes to run into problems. A simple way of increasing the reliability then is for the flash drive to have some extra "hidden" capacity and present a smaller logical drive and remap the erase erase units, combined with write leveling (occasionally moving data around even if it hasn't been modified, to move frequently rewritten blocks around the drive).
To your suggestion of execution out of flash: It is possible for some types of flash units, but you still need RAM for working memory - flash typically let you change a bit from one to zero or from zero to one (depending on type of flash) on a bit, byte or word basis, but only let you change it the other way in "erase units" that can range from hundreds of bytes to hundreds of KB at the time depending on the unit. This means that applications not specifically written to use memory that way won't be able to use it as a replacement for normal RAM for data that is both read and written.
What makes it impractical for most modern systems to execute out of flash apart from performance (flash is far slower than DRAM still) is things like relocation. On a limited function embedded device that's often an ok tradeoff (you can often KNOW that the total set of applications that will ever be loaded at the same time won't take up the full address space, so you can use absolute addresses.
That plants can use it, doesn't mean it will - especially when the CO2 levels have massively increased. But if we're going down that route, then taking the CO2 out of the atmosphere makes no difference either, as it can be mixed into various compounds that will eventually decompose or get burned or whatever and release oxygen again. It's not like this takes CO2 permanently out of use - in fact, one of the main problems with projects such as these have been that we don't know any really good ways of securely storing CO2 for long enough periods to avoid it from getting back into the atmosphere.
The article addressed several reasons for that, one being that it means that you are able to extract the CO2 where you want it, i.e. where you want to use it or sequester it. Another being that improving efficiency of a lot of current industry and vehicles may be more expensive than removing the CO2 with these devices afterwards, and devices like this can be run anywhere, making agreements between developing and developed countries far easier to achieve as we're no longer tied to getting these countries to improve their industries. You could even create an added economic incentive for building and operating these devices, by for example granting the operator a 1/2 ton CO2 quota to sell on the open market for every ton of CO2 they remove from the air.
English and German are both Germanic languages. If anything, the largest influence goes the other way around - the Saxon invasions was a major factor in shaping English. Just look at Old English (Anglo Saxon) texts, and chances are you'll find more resemblance to modern German than to modern English, largely because English since has taken up so many other words (after the Norman invasion, for example).
Slashdot Mirror
As for partitioning, it depends on how it's done. Proper partitioning makes a huge difference - you don't have non-essential stuff filling your disk at the wrong moment, for example. A typical example is that you really don't want your database to grind to a halt because someone "forgot" to remove a bunch of temporary files from their home directory after doing maintenance etc.. Partitions are a tool, and you can do stupid things with partitioning like with any other tool. For a desktop partitioning is less important, yes, and you will find that several distributions use partitioning very sparingly for that reason.
As for getting rid of virtual memory... Are you insane? Most people are running systems with far too little RAM. Without virtual memory their systems would be useless. Besides, you CAN easily turn it off - just switch off swapping ("man swapoff" in your hated console). I've run systems with swap turned off, but only embedded systems where I knew the memory requirements exactly. You'd be surprised how badly many programs handle running out of memory. For a desktop I'd bet most people would much rather have things slow down than randomly stop working, crashing, or keep giving them errors about running out of memory.
No. You need an API that lets you ask for a snapshot, and that handles it on a block level. You can "cross layers" if you absolutely want to, but why? It doesn't buy you anything, and it creates interdependencies. Why should the snapshot functionality have to be copied or rewritten every time you create a new filesystem? It doesn't need to depend on filesystem organization.
Adding new drives without partitioning, slicing, formatting. Just adding to the existing pool. Inclusive striping being adapted automagically. This needs a cross-layer interface, right ?
Of course it needs a "cross-layer interface". The entire point of layering is to expose an interface to the layer above that provides services but hides HOW those services are implemented. Nobody is complaining about interfaces from one layer to the other, but about removing the barriers between layers and merging them together in the same codebase, which, again, creates dependencies and reduces genericity where there's no reason to.
The transactional filesystem guarantees uncorrupted data at power failures and OS crashes. If you do this across a pool of physical platters, you need operations across layers.
Or you create a generic API for the lower layers that expose the functionality that is needed, and gets that promoted into the generic part of the appropriate layers so that all filesystems that want to have the hooks needed to implement similiar functionality without having to duplicate lots of code.
There may be arguments for considering changes in where the boundaries between layers are drawn, or even if a specific layer is really needed. But layering is there in part because a system like Linux is a large ecosystem. We have dozens of filesystems. We have specialized replacements or additions to the lower layers (such as a choice of using software raid or not, or a block device that mirrors over a network or not etc.) etc. The number of possible combinations is huge. Cutting through the layers is a lazy way of adding extra features to a single filesystem instead of making parts of those features available to other parts of the ecosystem.
The excuses just don't cut it - the reason ZFS cuts through the layers isn't because it needs to, but because it was convenient for the people writing ZFS without thinking holistically. In other words, they were concerned with looking only at ZFS, rather than looking at how to make the featureset they are supporting the most useful.
That's fair enough - for Sun it makes sense. They have no real incentive for exposing all this functionality as separate interfaces for the different layers so it can be built on by others. But trying to pretend they're doing it that way because it's the right thing to do is just silly.
I'd love if it was an easy choice - if it was easy to write filesystems for Linux so you had the option of either loading them as modules into the kernel or running them user space. The same would be great for other subsystems too. For some, the tradeoff in performance is acceptable if it buys any extra stability.
I doubt it will happen with the mainstream Linux kernel as long as Linus is in charge, though. A more likely scenario would be that someone retrofit Linux on top of one of the experimental microkernels (several projects do that already) and selectively move subsystems into separate processes (don't know if anyone is doing that yet).
And in the Reiser case the police apparently still don't know for sure whose blood it is.
As such, we have NO basis for saying anything about the car seat. We don't know when it was removed. We don't know if Hans Reiser knows where it is. We don't know if he has a plausible explanation for what happened to it.
As for the blood, we don't even know who the blood came from - to my knowledge the police have only stated they haven't been able to rule out that it is from Nina Reiser.
There's at the very least a chance she stage her own disappearance to get Hans in trouble and went back to Russia and got her kids brought there.
There's simply no basis for thinking a likely serial killer's claims to have killed or not killed a specific person are reliable.
In this case even less so, since he's had a lot of grievances against Hans Reiser and might very well have seen not confessing as an opportunity to get rid of Hans or just to have some fun with the police.
If she's alive, she's likely in Russia, where here kids and family is. I doubt she'd need to be very careful to hide in Russia - enough money to buy some fake papers would be enough.
In this case, there is no indication the children were "left behind". And mothers flee relationships often enough (as do men, of course), with or without their children that any claim that her disappearance alone is proof she's dead are unlikely to fly.
In any case, a lot of serial murderers don't confess to all their murders, and some even confess to murders they haven't committed, for a lot of different reasons. The defense will have an easy time killing any idea the prosecutor will try to put forward about the confession of these murders meaning that Sturgeon can somehow be trusted to tell the truth about Nina Reiser, as there are lots of examples to counteract any claim to "common sense".
Unless there are clear physical evidence linking Hans to it that can't have been tampered with by a psychopath or sociopath who had a relationship with Nina and a close business relationship with Hans and might thus have known a lot of details about their lives and who might even had opportunities to copy house keys etc., there's a lot of room for a hell of a lot of doubt.
All of these allegations came before Nina Reiser disappeared, and are well documented (lots of press + the court proceedings themselves). If anything, I'd expect the defense team to bring all that up, and present all that as motive - either for murder (Nina Reiser had a new boyfriend) or as a setup to frame Hans Reiser.
If you'd RTFA, you might also have noticed that testimony in one of the preliminary hearings stated that Nina Reiser broke off the relationship with Sturgeon because she was unhappy with his BDSM tendencies. If that's the case, you'd think she'd have brought up things like that in the rather nasty divorce proceedings if Hans Reiser was into it too.
All of these allegations came before Nina Reiser disappeared, and are well documented (lots of press + the court proceedings themselves). If anything, I'd expect the defense team to bring all that up, and present all that as motive - either for murder (Nina Reiser had a new boyfriend) or as a setup to frame Hans Reiser.
If I ever decide to go for a life of crime, targeting colo facilities would be high on my list - they love showing off their customers and describe in detail how their infrastructure is designed and what customers are where, including showing where the cable gates are etc. Makes their high fences and guards kind of useless when their sales guys can happily be talked into giving a grand tour to a hacker with some nice wireless gadgets to connect in the right places.
We're maybe as little as a century away from actually seeing the worlds population shrinking unless we start increasing lifespans a lot faster than we have.
Yes it's a good thing, but RTFA. Microsoft was one of a group of companies that wanted SCOTUS to make this decision. Obvious patents cost Microsoft far more in litigation than what they can expect to lose in licensing revenue.
Or he is hinting at an application for proper 3D. I don't go to the theatre often, but I'm perfectly aware how different the experience is. I've also seen enough filmed plays to know that getting proper 3D effects would make it a hell of a lot better. For one, being able to change focus and concentrate on different parts of the scene does make a lot of difference. So while a lot of movies wouldn't gain all that much from proper 3D, there are a lot of things that would.
That's true for only a handful or so countries in the world, and even for most of those the address verification services are woefully incomplete (that is, you have to expect them to return no result or wrong result for a fairly large percentage of users).
It's more like 100,000 to 10,000,000 depending on how expensive chips you use, and it's per erase unit, and it's not writes but erases. The latter two points are important, because it means that even if erase units are large, you can still write to the erase unit multiple times (until you've filled it), so you can split it into convenient block sizes, and you'd typically either use a flash translation layer or a flash specific filesystem that avoids erasing the unit until there are no free blocks elsewhere to minimize writes. In practice that means you have to have either an extremely high write load or an almost full drive with a medium level of writes to run into problems. A simple way of increasing the reliability then is for the flash drive to have some extra "hidden" capacity and present a smaller logical drive and remap the erase erase units, combined with write leveling (occasionally moving data around even if it hasn't been modified, to move frequently rewritten blocks around the drive).
What makes it impractical for most modern systems to execute out of flash apart from performance (flash is far slower than DRAM still) is things like relocation. On a limited function embedded device that's often an ok tradeoff (you can often KNOW that the total set of applications that will ever be loaded at the same time won't take up the full address space, so you can use absolute addresses.
It's a financial system. 12GB of financial data is quite a bit - it could very well be worldwide.
That plants can use it, doesn't mean it will - especially when the CO2 levels have massively increased. But if we're going down that route, then taking the CO2 out of the atmosphere makes no difference either, as it can be mixed into various compounds that will eventually decompose or get burned or whatever and release oxygen again. It's not like this takes CO2 permanently out of use - in fact, one of the main problems with projects such as these have been that we don't know any really good ways of securely storing CO2 for long enough periods to avoid it from getting back into the atmosphere.
The article addressed several reasons for that, one being that it means that you are able to extract the CO2 where you want it, i.e. where you want to use it or sequester it. Another being that improving efficiency of a lot of current industry and vehicles may be more expensive than removing the CO2 with these devices afterwards, and devices like this can be run anywhere, making agreements between developing and developed countries far easier to achieve as we're no longer tied to getting these countries to improve their industries. You could even create an added economic incentive for building and operating these devices, by for example granting the operator a 1/2 ton CO2 quota to sell on the open market for every ton of CO2 they remove from the air.
No, you "removed" the (usable, free) oxygen the moment you burned something where oxygen was part of the reaction.
Subsequently removing the CO2 makes no difference to that.
English and German are both Germanic languages. If anything, the largest influence goes the other way around - the Saxon invasions was a major factor in shaping English. Just look at Old English (Anglo Saxon) texts, and chances are you'll find more resemblance to modern German than to modern English, largely because English since has taken up so many other words (after the Norman invasion, for example).