still which is easier for a user who can turn the computer off if the button is not red?
The easiest thing for them to do is download a new binary, and run it.
even if it *is* super easy to apply a patch and recompile an open source program,
They would not need to, someone else (likely the person who coded the patch and sent it to the program maintainers) would do it for them and they would download a normal binary that just works.
No, but the cpu affinity thing is one of the things I was talking about, are you saying it is a _good_ thing that you have to force cpu affinity for a process for it not to crash and burn?
Kind of helps that they practically included a vm'd xp with it too, which brings back to my point about basically emulating the entire system and still hoping for the best.
Kludges upon kludges are bad, having an open source engine allows people to fix this.
i was meaning, sweden hasn't even charged him with anything. the extradition is over 'questioning' which he had already agreed to do but did not want the inconvenience of going back to sweden etc etc. Interpol warrants for 'questioning' are just bullshit.
If they think he's done something lay charges then get him there, otherwise they should sod off.
The last time brisbane was flooded completely was 1974, before that around 1893.
The media are motivated to sensationalise their reports for the ratings that this brings.
All the better for people to take this seriously as opposed to laughing at it and deciding to go 'swimming'
It is not a common occurrence to have half the damn state under water man. Yes some outback people are used to it and have prepared for it, but do you seriously think city dwellers that have not seen a flood in their lifetime have all their affairs in order for it to be not a problem?
Nobody will argue the climate is changing.. it always has, what you will get is people arguing about how much of that is caused by humans, which I think is fair debate.
A lot of greens seem to be very irrational to me, they are the sort of people who (rightly) dislike the burning of coal and yet also hate the idea of approving the construction of modern fast breeder nuclear reactors that recycle their own fuel until what is left is barely above background radiation (and ironically, releases less radiation into the environment in total, since burning coal tends to release some into the air.. I'd prefer a small amount in barrels then more in the air).
Take this flood for instance, one of almost the same level happened in 1974, and in 1893. By all accounts this is tends to be an every 50-100 year thing. Yet on the news there are so many people claiming straight up that this was mans doing without any evidence whatsoever.
The greens may have the right idea, but outright lying does not help their cause and makes people like me cynical towards them.
users just prefer the windows way: I've this program, I bought 20 years ago, and I can still launch it in windows *just fine*. no recompilations, no dependency, no libraries...
The problem is windows users CAN'T do that. Try playing thief the dark project on a modern windows machine, it will chuck shits, and no-one can fix the original engine because it is not open source.
the most messy thing users have to do on windows is to dig up an older direct x version (and there is the direct x 9 small updates flurry nightmare)
No, the most messy thing windows users have to do is emulate an entire system from the era and hope to god it doesn't need something like glide support for 3d acceleration which the VM won't do.
So you want release cycles of a few years then with the accompanying stagnation that bsd exhibits?
No thank you, if you've noticed linux supports a shitload more hardware than freebsd does.
As mentioned, if in mainline you don't really have to maintain your driver any more once it works, because if anyone changes anything that would break it, they are the ones that have to fix it. Non-mainline drivers aren't even worth considering because of the considerable drawbacks.
And don't get me started with solaris. If you have a M9000 that is quaded (big cpu board cut up logically). And you have a hardware failure, even though it is redundant hardware, zfs on solaris STILL gives you a non-fatal error, that crashes your server. What the hell is the point in buying a 2 million dollar plus machine if they can't get the drivers right when they are the ones both making both the hardware and software?
By having long release cycles you are sacrificing too much on the side of development for no benefit.
The ability to load binary kernel modules a decade old is of no real benefit, considering the serious risks you impose on yourself by doing so.
And if it's open source why not just get it in mainline and not care about it after it works? and it will stay working.
You don't sound like you are actually a programmer, so I can't see OS giving you much, if any, benefit, except for the fact that it's likely to be free-as-in-beer.
You'd be wrong, linux user of over ten years now and when things don't work I tend to 'make them' work. (latest was for a friend, ported bristol audio synth to mac os x, usually just little things).
That being said, I'm a lazy person and while I can fix things myself it's far better if I can spend that time doing something productive instead.
What are the drawbacks of closed source, specifically?
The biggest problem for me is dependence, using closed source software you are being dependent on a vendor for future code updates, if the source code is open under a non-restrictive license you are free to continue development of it or if it's popular someone else likely will.
If the source is open you can get it going with gnu toolchain and from there start working on portability and overall general maintenance.
Point is you don't have to be a programmer to benefit, but if you are you have more options.
I'd recommend a different action than "getting the hell out". Get elected and change the laws.
Anyone who is ruthless enough to actually be elected normally does not deserve the position. To win you must be a master of both public speaking and doublespeak. After concessions are made to your ethics in order to gain the required popularity to win it is a slippery slope and by the time you get there (if you do) you become just as bad as those that were in power before you.
Not saying I have the solution to it, only that there are another set of problems to think about.
Weakie leaks has endangered many lives - espionage is not free speech.
The discussion of political implications of the leaks of internal government stuff is now espionage hey? We are talking about people discussing and being in favour of what wikileaks has done, ignoring whether you think what it has done is good or bad, you think that people agreeing with it politically should be enough to let the government have it's way with them too?
Wikileaks was also not involved in any espionage themselves, those that gave them the information may have but that is irrelevent to wiki leaks. Wikileaks merely published what they were given, as did the new york times and other publications. By your reasoning are they too guilty of espionage by covering some of the cables in articles?
Individuals are entitled to say as they wish to each other in their private lives, the moment that is stopped in the name of 'national security' when they are discussing politics is when you should get the hell out.
It would be easy to add in a customizable "category" field and let those in a category see only those in that category and any other categories you manually set and be able to push updates only to selected categories if you wished.
This is already implemented, you just add people t groups and then in your post click privacy settings and you can say which group(s) get the message
Google will become skynet, the information they possess combined with things such as google goggles and other experimental projects mean they are in the perfect position to get into the AI business.
They have linguistics almost downpat with google translate. They already know trends in the world through google trends and how people associate things.
Combine this with the extreme compute power available to them, googles presence nearly everywhere and I think all of their services combined into one entity could become... scary, over time.
So answer honestly Linux guys, how much money have you spent on the software that is currently installed on your PC?
Full ticket price? NWN set me back $80, quake 3 set me back $50, quake 4 set me back $60. Baldurs gate 2 set me back $80, mostly older titles I know.
Humble indie bundle I donated $25 to.
Thing is, if the game company tells me (ala ID software) that they will eventually open source the game engine code but not the art so you will still pay for it, I'm a hell of a lot more likely to pay for their stuff.
Closed source has far too many drawbacks to be worthwhile to me on most occasions, if the game engine is open it means so long as the game is semi popular it will always work on my platform of choice.
Money is not the issue, the issue is people (or at least me) don't want to fork out $60 for a game where you are screwed when the engine breaks because of lack of updates. Or in a couple years when the multiplayer servers get killed and you can't play online.
I've dropped about a grand on wii games simply because I'm guaranteed they will continue to work so long as they have a wii. Only way to ensure it will continue to work on the pc is to have the engine source open and have people actually interested in using it.
Added bonus is you get free labour for the port, and code improvements over time which the company did not have to pay a cent for.
and their unease with slowing down the development process. A properly thought out, long term stable ABI would take time to develop, and massive commitment.
Again, why should their development process suffer because you want something that provides no benefit and arguably actually hurts the kernel?
I have read the excuses made by the kernel developers and they just don't wash.
You have to have a legitimate need to even want a stable ABI, you have not presented one.
Depends, I mean for some algorithms lacking unsigned types in java the language is a hindrance, sure the hardware can do unsigned bytes but without the language exposing it what's the point.
Looking at the problems Linux has, how are the cons of an ABI not worth it? Can you even give one tangible pro for the status quo that an end-user would appreciate? "More frequent kernel releases" is so not an answer...
Again what problems? think of it from a mainline developer point of view, you are suggesting the mainline kernel limit itself in its dealings in order to pander to those who will release closed source binaries (unsupported, major security risk and may not efficiently re-use existing code in the kernel) or release source but with source being in such a state that it is unfit to be included in mainline (i.e. poor quality)
I don't see any drawbacks to rejecting poor quality or code that cannot be reviewed.
The pros of the status quo are many, one of which that code re-use within the kernel is crazy and awesome. Writing a wifi driver? the mac80211 stack and most functions are already there you only have to write a small portion.
When a new subsystem that a lot of things will be using is created, all existing drivers get changed to use it, limiting the amount of code that does the same thing.
Another is that all code can be heavily reviewed, since the only reason to even want a stable ABI is for binary modules.
If you don't care for binary modules or out of tree drivers, what mainline are currently doing is the best option unquestionably and there are no drawbacks to it.
And strangely enough the kernel people tend to think for the long term benefit of the kernel, not short term, binary blobs come and go.
What it boils down to: Having a completely stable ABI/API inhibits kernel development by not allowing improvements if it would break it. And only benefits out of mainline kernel stuff.
The lesson? get your stuff in mainline. No problems then. Why should mainline limit their development for some person building an obscure home made module that they never intend to get into mainline?
The policy of a single monolithic kernel that alone provides hardware support is in itself harmful - it is, quite obviously, not scalable.
Bullshit, having all the drivers in kernel allows for serious code re-use. Things like mac80211 for wifi and other kernel modules for other things allow individual chipset drivers do as little as they really need to.
If everything was not in kernel when these common functions are put into their own module existing drivers would not be automatically updated to use/support them.
The linux kernel has modularity and code re-use to the extreme. Having everything separate makes code re-use suffer.
However, there is an awful lot of kernel code that hasn't been touched in years, that some people need to make their gizmos work, leaving them out of the loop for bug fixes and new feature support.
This differs from the closed source model of 'oh look, we've stopped shipping hardware, lets stop making new drivers so windows vista/7 will not work on perfectly good hardware.
And lets face it, linux backwards compatibility is far better than windows, hell MFM (pre-IDE) hard disks were only deprecated in 2.6.16
This is what made me quit using Linux. No stable branch with long term support and bug fixes means that you're always updating to the latest kernel for your bug fixes.
Unless you need some core functionality of the new kernel, why not just continue to run the first functioning kernel you find? as an end user machine assuming most servers are disabled and sane iptables settings by default security risk is of little concern to you.
If you are running a server RHEL etc are designing for your long term support needs. But the thing is of course a 'stable' kernel ABI wise is a rather serious hindrance and you don't get the shiny features etc, you really are better off just running newest kernel after first checking for the odd chance of regressions.
The damn thing of it is that the problems with this development model are obvious, but nobody wants to admit it.
Kernel community at large has went 'bah' to binary modules, only ones that matter are the ones in mainline kernel tree and they are looked after well and provide higher quality assurance than is possible with joe random giving you drivers.
Drivers are the main cause of BSOD's on windows these days, all because of hardware engineers thinking they can do software and failing, and no-one being able to inspect the ensuing mess.
The mainline kernel won't just accept any old driver into it, it take time to integrate code to ensure it's up to scratch. This is a _good_ thing. For code that is completely new this results in more testing and a better driver overall.
The only real lesson here for intel is give the kernel devs code before hardware ships if they want there to be a kernel that supports it released when the hardware is.
You mistake 'stable' for as a static ABI, in this sense it more refers to the software being stable when run.
The kernel api is relatively stable, with new key subsystem changes every now and then.
This is ideal for open source dev.
The only real lesson intel have to learn from this is release the code earlier so it can be integrated into mainline faster so distros will package it.
Slashdot Mirror
Yes, because the amount of iphones in the wild is a strong indicator of how many macs there are in the wild also....
The overwhelming majority of people don't own a mac, and those that do tend to be arts majors and not typically into programming.
To most people dropping a grand on a mac just to play with developing on their iphone is not a sane option.
still which is easier for a user who can turn the computer off if the button is not red?
The easiest thing for them to do is download a new binary, and run it.
even if it *is* super easy to apply a patch and recompile an open source program,
They would not need to, someone else (likely the person who coded the patch and sent it to the program maintainers) would do it for them and they would download a normal binary that just works.
No, but the cpu affinity thing is one of the things I was talking about, are you saying it is a _good_ thing that you have to force cpu affinity for a process for it not to crash and burn?
Kind of helps that they practically included a vm'd xp with it too, which brings back to my point about basically emulating the entire system and still hoping for the best.
Kludges upon kludges are bad, having an open source engine allows people to fix this.
Just saying that proposing having long closed release cycles produces the epitome of quality is a bit of a farce.
Also, nothing to say about the rest of the points? (solaris business was a bit of a side-note)
i was meaning, sweden hasn't even charged him with anything. the extradition is over 'questioning' which he had already agreed to do but did not want the inconvenience of going back to sweden etc etc. Interpol warrants for 'questioning' are just bullshit.
If they think he's done something lay charges then get him there, otherwise they should sod off.
One problem: he has not been charged with anything.
The last time brisbane was flooded completely was 1974, before that around 1893.
The media are motivated to sensationalise their reports for the ratings that this brings.
All the better for people to take this seriously as opposed to laughing at it and deciding to go 'swimming'
It is not a common occurrence to have half the damn state under water man. Yes some outback people are used to it and have prepared for it, but do you seriously think city dwellers that have not seen a flood in their lifetime have all their affairs in order for it to be not a problem?
Nobody will argue the climate is changing.. it always has, what you will get is people arguing about how much of that is caused by humans, which I think is fair debate.
A lot of greens seem to be very irrational to me, they are the sort of people who (rightly) dislike the burning of coal and yet also hate the idea of approving the construction of modern fast breeder nuclear reactors that recycle their own fuel until what is left is barely above background radiation (and ironically, releases less radiation into the environment in total, since burning coal tends to release some into the air.. I'd prefer a small amount in barrels then more in the air).
Take this flood for instance, one of almost the same level happened in 1974, and in 1893. By all accounts this is tends to be an every 50-100 year thing. Yet on the news there are so many people claiming straight up that this was mans doing without any evidence whatsoever.
The greens may have the right idea, but outright lying does not help their cause and makes people like me cynical towards them.
users just prefer the windows way: I've this program, I bought 20 years ago, and I can still launch it in windows *just fine*. no recompilations, no dependency, no libraries...
The problem is windows users CAN'T do that. Try playing thief the dark project on a modern windows machine, it will chuck shits, and no-one can fix the original engine because it is not open source.
the most messy thing users have to do on windows is to dig up an older direct x version (and there is the direct x 9 small updates flurry nightmare)
No, the most messy thing windows users have to do is emulate an entire system from the era and hope to god it doesn't need something like glide support for 3d acceleration which the VM won't do.
It is theft in the same way as identity theft, yes you still have your identity, but so do they.
So you want release cycles of a few years then with the accompanying stagnation that bsd exhibits?
No thank you, if you've noticed linux supports a shitload more hardware than freebsd does.
As mentioned, if in mainline you don't really have to maintain your driver any more once it works, because if anyone changes anything that would break it, they are the ones that have to fix it. Non-mainline drivers aren't even worth considering because of the considerable drawbacks.
And don't get me started with solaris. If you have a M9000 that is quaded (big cpu board cut up logically). And you have a hardware failure, even though it is redundant hardware, zfs on solaris STILL gives you a non-fatal error, that crashes your server. What the hell is the point in buying a 2 million dollar plus machine if they can't get the drivers right when they are the ones both making both the hardware and software?
By having long release cycles you are sacrificing too much on the side of development for no benefit.
The ability to load binary kernel modules a decade old is of no real benefit, considering the serious risks you impose on yourself by doing so.
And if it's open source why not just get it in mainline and not care about it after it works? and it will stay working.
You don't sound like you are actually a programmer, so I can't see OS giving you much, if any, benefit, except for the fact that it's likely to be free-as-in-beer.
You'd be wrong, linux user of over ten years now and when things don't work I tend to 'make them' work. (latest was for a friend, ported bristol audio synth to mac os x, usually just little things).
That being said, I'm a lazy person and while I can fix things myself it's far better if I can spend that time doing something productive instead.
What are the drawbacks of closed source, specifically?
The biggest problem for me is dependence, using closed source software you are being dependent on a vendor for future code updates, if the source code is open under a non-restrictive license you are free to continue development of it or if it's popular someone else likely will.
If the source is open you can get it going with gnu toolchain and from there start working on portability and overall general maintenance.
Point is you don't have to be a programmer to benefit, but if you are you have more options.
I'd recommend a different action than "getting the hell out". Get elected and change the laws.
Anyone who is ruthless enough to actually be elected normally does not deserve the position. To win you must be a master of both public speaking and doublespeak. After concessions are made to your ethics in order to gain the required popularity to win it is a slippery slope and by the time you get there (if you do) you become just as bad as those that were in power before you.
Not saying I have the solution to it, only that there are another set of problems to think about.
Weakie leaks has endangered many lives - espionage is not free speech.
The discussion of political implications of the leaks of internal government stuff is now espionage hey? We are talking about people discussing and being in favour of what wikileaks has done, ignoring whether you think what it has done is good or bad, you think that people agreeing with it politically should be enough to let the government have it's way with them too?
Wikileaks was also not involved in any espionage themselves, those that gave them the information may have but that is irrelevent to wiki leaks. Wikileaks merely published what they were given, as did the new york times and other publications. By your reasoning are they too guilty of espionage by covering some of the cables in articles?
Individuals are entitled to say as they wish to each other in their private lives, the moment that is stopped in the name of 'national security' when they are discussing politics is when you should get the hell out.
To where is the only real question.
It would be easy to add in a customizable "category" field and let those in a category see only those in that category and any other categories you manually set and be able to push updates only to selected categories if you wished.
This is already implemented, you just add people t groups and then in your post click privacy settings and you can say which group(s) get the message
Google will become skynet, the information they possess combined with things such as google goggles and other experimental projects mean they are in the perfect position to get into the AI business.
They have linguistics almost downpat with google translate. They already know trends in the world through google trends and how people associate things.
Combine this with the extreme compute power available to them, googles presence nearly everywhere and I think all of their services combined into one entity could become... scary, over time.
The same argument about the network effect held true for them too, but it didn't stop people from migrating.
It doesn't stop people from migrating, but it provides a major incentive not to.
So answer honestly Linux guys, how much money have you spent on the software that is currently installed on your PC?
Full ticket price? NWN set me back $80, quake 3 set me back $50, quake 4 set me back $60. Baldurs gate 2 set me back $80, mostly older titles I know.
Humble indie bundle I donated $25 to.
Thing is, if the game company tells me (ala ID software) that they will eventually open source the game engine code but not the art so you will still pay for it, I'm a hell of a lot more likely to pay for their stuff.
Closed source has far too many drawbacks to be worthwhile to me on most occasions, if the game engine is open it means so long as the game is semi popular it will always work on my platform of choice.
Money is not the issue, the issue is people (or at least me) don't want to fork out $60 for a game where you are screwed when the engine breaks because of lack of updates. Or in a couple years when the multiplayer servers get killed and you can't play online.
I've dropped about a grand on wii games simply because I'm guaranteed they will continue to work so long as they have a wii. Only way to ensure it will continue to work on the pc is to have the engine source open and have people actually interested in using it.
Added bonus is you get free labour for the port, and code improvements over time which the company did not have to pay a cent for.
and their unease with slowing down the development process. A properly thought out, long term stable ABI would take time to develop, and massive commitment.
Again, why should their development process suffer because you want something that provides no benefit and arguably actually hurts the kernel?
I have read the excuses made by the kernel developers and they just don't wash.
You have to have a legitimate need to even want a stable ABI, you have not presented one.
The language itself has 0 to do with its speed.
Depends, I mean for some algorithms lacking unsigned types in java the language is a hindrance, sure the hardware can do unsigned bytes but without the language exposing it what's the point.
Looking at the problems Linux has, how are the cons of an ABI not worth it? Can you even give one tangible pro for the status quo that an end-user would appreciate? "More frequent kernel releases" is so not an answer...
Again what problems? think of it from a mainline developer point of view, you are suggesting the mainline kernel limit itself in its dealings in order to pander to those who will release closed source binaries (unsupported, major security risk and may not efficiently re-use existing code in the kernel) or release source but with source being in such a state that it is unfit to be included in mainline (i.e. poor quality)
I don't see any drawbacks to rejecting poor quality or code that cannot be reviewed.
The pros of the status quo are many, one of which that code re-use within the kernel is crazy and awesome. Writing a wifi driver? the mac80211 stack and most functions are already there you only have to write a small portion.
When a new subsystem that a lot of things will be using is created, all existing drivers get changed to use it, limiting the amount of code that does the same thing.
Another is that all code can be heavily reviewed, since the only reason to even want a stable ABI is for binary modules.
If you don't care for binary modules or out of tree drivers, what mainline are currently doing is the best option unquestionably and there are no drawbacks to it.
And strangely enough the kernel people tend to think for the long term benefit of the kernel, not short term, binary blobs come and go.
What it boils down to: Having a completely stable ABI/API inhibits kernel development by not allowing improvements if it would break it. And only benefits out of mainline kernel stuff.
The lesson? get your stuff in mainline. No problems then. Why should mainline limit their development for some person building an obscure home made module that they never intend to get into mainline?
The policy of a single monolithic kernel that alone provides hardware support is in itself harmful - it is, quite obviously, not scalable.
Bullshit, having all the drivers in kernel allows for serious code re-use. Things like mac80211 for wifi and other kernel modules for other things allow individual chipset drivers do as little as they really need to.
If everything was not in kernel when these common functions are put into their own module existing drivers would not be automatically updated to use/support them.
The linux kernel has modularity and code re-use to the extreme. Having everything separate makes code re-use suffer.
However, there is an awful lot of kernel code that hasn't been touched in years, that some people need to make their gizmos work, leaving them out of the loop for bug fixes and new feature support.
This differs from the closed source model of 'oh look, we've stopped shipping hardware, lets stop making new drivers so windows vista/7 will not work on perfectly good hardware.
And lets face it, linux backwards compatibility is far better than windows, hell MFM (pre-IDE) hard disks were only deprecated in 2.6.16
This is what made me quit using Linux. No stable branch with long term support and bug fixes means that you're always updating to the latest kernel for your bug fixes.
Unless you need some core functionality of the new kernel, why not just continue to run the first functioning kernel you find? as an end user machine assuming most servers are disabled and sane iptables settings by default security risk is of little concern to you.
If you are running a server RHEL etc are designing for your long term support needs. But the thing is of course a 'stable' kernel ABI wise is a rather serious hindrance and you don't get the shiny features etc, you really are better off just running newest kernel after first checking for the odd chance of regressions.
The damn thing of it is that the problems with this development model are obvious, but nobody wants to admit it.
Kernel community at large has went 'bah' to binary modules, only ones that matter are the ones in mainline kernel tree and they are looked after well and provide higher quality assurance than is possible with joe random giving you drivers.
Drivers are the main cause of BSOD's on windows these days, all because of hardware engineers thinking they can do software and failing, and no-one being able to inspect the ensuing mess.
The mainline kernel won't just accept any old driver into it, it take time to integrate code to ensure it's up to scratch. This is a _good_ thing. For code that is completely new this results in more testing and a better driver overall.
The only real lesson here for intel is give the kernel devs code before hardware ships if they want there to be a kernel that supports it released when the hardware is.
You mistake 'stable' for as a static ABI, in this sense it more refers to the software being stable when run.
The kernel api is relatively stable, with new key subsystem changes every now and then.
This is ideal for open source dev.
The only real lesson intel have to learn from this is release the code earlier so it can be integrated into mainline faster so distros will package it.