America's economy, basically sweat shops, Walmarts, and other places that pay minimum wage or less, depend on cheap labor. And you can't have cheap labor if that labor cannot take shortcuts so it can survive on a sub living wage.
So they drive unregistered, uninspected and uninsured cars. As soon as the powerful business interests realize this, they will scuttle that plan. The UK can get away with this because it sort of has public transportation. Not an option here in the US.
Real humans don't have such perfect symetry. It's true that better symetry is considered more beautiful but nobody has perfect symetry. And people who look too good, ie. too symetrical, do look sort of creepy.
I mean they're preventing me from implementing this new algorithm, RCU for preemptive user threads. IBM has way too many defensive patents in that area. It gives me an excuse for not doing it and anyway, we have too many algorithms already.
in objectieve terms. And that would explain why SCO wants all that info from IBM. And why that probably will not do SCO any good. It will take me a few minutes to write it up which is too long for Slashdot. I may post it later to GrokLaw which has a longer attention span than Slashdot.
that Sun wan't you to agree to in order to get access to their source. Doesn't look particularly viral. Assuming Sun's Unix license wasn't as viral as SCO thinks it is, Sun could transfer the non-derived parts of Solaris to GPL, Open Source, or whatever. And assuming Sun knew what parts were derived. Some parts would obviously appear not derived to anybody, except to SCO perhaps
All I have is a BA from a rather low grade college. I specialize in locking algorithms.
A reader lock-free algorithm now known as RCU. See HOS89 ref here
"A Distributed Solution to the Reader-Writer Problem" in SIGOPS Operating Systems Review, Apr 1990.
Two of the shortest reader-writer FIFO locks you will ever see (algorithm from above based on one) here
Lock-free reference counting using 2 different techniques, one for CAS2 and one for LL/SC that David Detlefs couldn't manage to do. Scan c.p.t for atomic_ptr.
A fast pathed counting semaphore that doesn't need futexes, i.e. will work on windows even.
A proxy garbage collection scheme implemented with atomic_ptr (less overhead than pure refcounted GC).
RCU for preemptive user threads currently being architected. This will work on any unix with the right/proc info (sorry windows) (I don't whether it will get implemented since it uses some IBM patented stuff).
Even more lesser hacks and unpublished stuff which I don't have time to go into here.
You have no idea what constitutes a patentable idea. And the number finite states for a computer greatly exceeds your capacity to imagine them. There are good reasons to not have software patents or to restrict them or otherwise curb patent abuse. You're not helping the cause with this kind of reasoning.
Actually, I wonder who's going to maintain all that code. All the out of work Solaris kernel developers? Hey, a new business strategy, "We can't afford to pay you, so we're laying you off and you should continue to work for free."
That could explain why the project I thought I was involved with, implementing RCU for preemptive user threads, doesn't seem to be going anywhere. There's no need for it.
That's only if the threads are completely independent. That's almost never the case. The problem is Amdahl's law has a bit of exponential bite to it. That's why you see all those scalabilty efforts do okay up to 8 or 16 way or so. When they start going to 32 or 64 or greater they start running into trouble. And that's with the highly tuned kernels.
Major applications like web servers, databases haven't even started on this. It's only in the discussion stages right now, if at all. I'm supposedly involved with such a project and as far as I know it's only a theoretical possiblity that someone might actually do something about it.
I use a slightly altered technique. Put the veggies in first and bring to a boil. Turn off heat and drop in ramen noodles and set set for 3 min and add in seasoning. Less pot watching when you do the heating this way.
If you want egg drop soup bring back to a near boil and stir in one raw egg making sure it gets cooked properly. What I do is actually mix the egg and some lemon juice together (shake in a small closed jar is easiest technique) and stir that into headed soup mixture. You get a creamier soup than the regular egg drop soup. Don't overheat though.
Nobody I know who has ever worked in a processed food plant will ever eat the stuff. When pressed for a reason they just say "You don't want to know". I suppose it's somewhat analogous to restaurants where you don't want to know what goes on in the kitchen. But with restaurants you can at least select on the external quality of the place. With processed food the same places that make the upscale stuff also make the cheap crap you wouldn't feed to your dog. Dogs will eat anything and come to think of it, so will geeks, so maybe this really isn't an issue.
If only the degradation was graceful. You hit the tipping point in system usage and with round robin scheduling suddenly the LRU page replacement algorithms don't work so well and Denning's law kicks in with a vengence.
The problem is system performance is queue based and the queue service times are different depending on whether processes/threads have to wait. If they start waiting then suspend/resume overhead gets added to the queue service times. You get a hysteresis effect on you performance curve. This means your system load has to be reduced dramatically before you see a return to normal performace levels.
It's not just overuse of memory that can cause this. Heavily threaded applications can suffer this because of queueing on locks for sharing. Use of lock-free algorithms helps because they don't use wait queues by definition.
Business is a weird little ducky. It's not like you can show empirically that businesses that invest more in X do better. A lot of funny stuff going on. If I had to pick one area where you could say businesses get the best ROI for their investment, I would have to say it's lobbying. If fact if there was a mutual fund based on lobbying, I invest heavily in it myself. Actually, that's not a joke. There is such a thing as lobbying for individuals. They get tax loopholes out of it.
From Duck Soup I think. Chico: (menacingly)I kill people for money. (looks at Harpo) I kill you for money. Harpo: (looks worried) Chico: (smiles). No, I no kill you for money. You my friend. I kill you for free. Harpo: (smiles in relief)
I'm sure killing spammers will be very economic as many people would be willing to do it for free.
whether that be the original authors in the case of copyright, public knowledge in the case of trade secrets, or prior art in the case of patents. The latter is more tenuous since we don't know what bogus patents may be laying in wait. Any of the non trivial parts of Linux should be documented. If you invented it, say so. If you got it from a textbook or paper, say so.
Great! Now us mainframe old timers can go and document everything that was invented on mainframes first. Or that everything was invented on mainframes first. Same difference.:)
since bartering income has to be declared as such on your tax return. But the cost of preparing your return is deductable but not in the same tax year usually. Otherwise it'd cancel out quite nicely for you but not the accountant.
I've done that with DOS 1.0. Present kernels are quite a bit larger than that now. I don't know how practical that would be unless you had a pretty good idea where to look. Sometimes that's possible but there wounldn't be any guarantee.
based on detectability of infringement of IP. Typically patenting something is only useful if usage of it is detectable. Normally you don't have access to source so you have to infer it from behavior and that behavior has to be fairly unique. For example, if all you have is O(log n) and there are ten other algorithms with O(log n) you will have a difficult time making a case for infringement. To say nothing of why you patented something that was not an improvement on existing art.
Not only does having the source make detection of infringement easier, it also makes patenting of otherwise completely indetectable algorithms feasible. Feasible that is if someone (SCO anyone?) can establish a precedent for making money by filing lawsuits against users of open source. This could get really bad given the USPTO's tendency to issue nonsense patents which will really lend themselves to being submarine patents due to their trivialness.
Slashdot Mirror
America's economy, basically sweat shops, Walmarts, and other places that pay minimum wage or less, depend on cheap labor. And you can't have cheap labor if that labor cannot take shortcuts so it can survive on a sub living wage. So they drive unregistered, uninspected and uninsured cars. As soon as the powerful business interests realize this, they will scuttle that plan. The UK can get away with this because it sort of has public transportation. Not an option here in the US.
Real humans don't have such perfect symetry. It's true that better symetry is considered more beautiful but nobody has perfect symetry. And people who look too good, ie. too symetrical, do look sort of creepy.
I mean they're preventing me from implementing this new algorithm, RCU for preemptive user threads. IBM has way too many defensive patents in that area. It gives me an excuse for not doing it and anyway, we have too many algorithms already.
Cisco certification, Oracle certification and a MSCE. You would almost think they were trying to replace 3 or 4 employess with one position.
in objectieve terms. And that would explain why SCO wants all that info from IBM. And why that probably will not do SCO any good. It will take me a few minutes to write it up which is too long for Slashdot. I may post it later to GrokLaw which has a longer attention span than Slashdot.
that Sun wan't you to agree to in order to get access to their source. Doesn't look particularly viral. Assuming Sun's Unix license wasn't as viral as SCO thinks it is, Sun could transfer the non-derived parts of Solaris to GPL, Open Source, or whatever. And assuming Sun knew what parts were derived. Some parts would obviously appear not derived to anybody, except to SCO perhaps
All I have is a BA from a rather low grade college. I specialize in locking algorithms.
Even more lesser hacks and unpublished stuff which I don't have time to go into here.
People keep adding nifty new features to browsers without thinking through the security implications. Any unwanted behavior is a security issue.
You have no idea what constitutes a patentable idea. And the number finite states for a computer greatly exceeds your capacity to imagine them. There are good reasons to not have software patents or to restrict them or otherwise curb patent abuse. You're not helping the cause with this kind of reasoning.
here. Not about Linux per se but I'm suprised ADTI hasn't picked up on it and tried to use it to bolster their case.
Actually, I wonder who's going to maintain all that code. All the out of work Solaris kernel developers? Hey, a new business strategy, "We can't afford to pay you, so we're laying you off and you should continue to work for free."
That could explain why the project I thought I was involved with, implementing RCU for preemptive user threads, doesn't seem to be going anywhere. There's no need for it.
That's only if the threads are completely independent. That's almost never the case. The problem is Amdahl's law has a bit of exponential bite to it. That's why you see all those scalabilty efforts do okay up to 8 or 16 way or so. When they start going to 32 or 64 or greater they start running into trouble. And that's with the highly tuned kernels. Major applications like web servers, databases haven't even started on this. It's only in the discussion stages right now, if at all. I'm supposedly involved with such a project and as far as I know it's only a theoretical possiblity that someone might actually do something about it.
when yout go the multi-threaded, CMT, etc.. route then Amdahl's law trumps Moore's law.
If you want egg drop soup bring back to a near boil and stir in one raw egg making sure it gets cooked properly. What I do is actually mix the egg and some lemon juice together (shake in a small closed jar is easiest technique) and stir that into headed soup mixture. You get a creamier soup than the regular egg drop soup. Don't overheat though.
Nobody I know who has ever worked in a processed food plant will ever eat the stuff. When pressed for a reason they just say "You don't want to know". I suppose it's somewhat analogous to restaurants where you don't want to know what goes on in the kitchen. But with restaurants you can at least select on the external quality of the place. With processed food the same places that make the upscale stuff also make the cheap crap you wouldn't feed to your dog. Dogs will eat anything and come to think of it, so will geeks, so maybe this really isn't an issue.
If only the degradation was graceful. You hit the tipping point in system usage and with round robin scheduling suddenly the LRU page replacement algorithms don't work so well and Denning's law kicks in with a vengence.
The problem is system performance is queue based and the queue service times are different depending on whether processes/threads have to wait. If they start waiting then suspend/resume overhead gets added to the queue service times. You get a hysteresis effect on you performance curve. This means your system load has to be reduced dramatically before you see a return to normal performace levels.
It's not just overuse of memory that can cause this. Heavily threaded applications can suffer this because of queueing on locks for sharing. Use of lock-free algorithms helps because they don't use wait queues by definition.
Business is a weird little ducky. It's not like you can show empirically that businesses that invest more in X do better. A lot of funny stuff going on. If I had to pick one area where you could say businesses get the best ROI for their investment, I would have to say it's lobbying. If fact if there was a mutual fund based on lobbying, I invest heavily in it myself. Actually, that's not a joke. There is such a thing as lobbying for individuals. They get tax loopholes out of it.
Chico: (menacingly)I kill people for money. (looks at Harpo) I kill you for money.
Harpo: (looks worried)
Chico: (smiles). No, I no kill you for money. You my friend. I kill you for free.
Harpo: (smiles in relief)
I'm sure killing spammers will be very economic as many people would be willing to do it for free.
as in shared virtual memory. Big problem is locking and the memory coherency that implies.
whether that be the original authors in the case of copyright, public knowledge in the case of trade secrets, or prior art in the case of patents. The latter is more tenuous since we don't know what bogus patents may be laying in wait. Any of the non trivial parts of Linux should be documented. If you invented it, say so. If you got it from a textbook or paper, say so.
Great! Now us mainframe old timers can go and document everything that was invented on mainframes first. Or that everything was invented on mainframes first. Same difference. :)
since bartering income has to be declared as such on your tax return. But the cost of preparing your return is deductable but not in the same tax year usually. Otherwise it'd cancel out quite nicely for you but not the accountant.
I've done that with DOS 1.0. Present kernels are quite a bit larger than that now. I don't know how practical that would be unless you had a pretty good idea where to look. Sometimes that's possible but there wounldn't be any guarantee.
based on detectability of infringement of IP. Typically patenting something is only useful if usage of it is detectable. Normally you don't have access to source so you have to infer it from behavior and that behavior has to be fairly unique. For example, if all you have is O(log n) and there are ten other algorithms with O(log n) you will have a difficult time making a case for infringement. To say nothing of why you patented something that was not an improvement on existing art.
Not only does having the source make detection of infringement easier, it also makes patenting of otherwise completely indetectable algorithms feasible. Feasible that is if someone (SCO anyone?) can establish a precedent for making money by filing lawsuits against users of open source. This could get really bad given the USPTO's tendency to issue nonsense patents which will really lend themselves to being submarine patents due to their trivialness.