There's no reason that, even at a base level, you can't write scripts that do the same as it does - for all functions, but also for parts of the functions
Actually there is. Systemd does lots of things that run at very low levels and thus performance matters. Processes that run multiple times per second aren't scripted in Unix. Unix has always accepted that you use a high level scripting language to tie together low level binary function, not that you script the whole system. What you are asking for is a design like what you see in LISP machines where the scripting languages are designed to degrade into very fast code so that it is safe to script almost the entire OS. Readying your criticism of systemd it would apply equally well to the kernel or to many of the standard libraries on Linux.
The kind of operating system you really want doesn't exist anymore on hardware What you really want is a LISP machine. Quite literally the essential binary part of the operating system was a few hundred assembly language functions and everything else scripted. I think where you do see this is in cloud computing. Many of the web based systems and IaaS manipulations are scriptable easily changeable and modular.
Honestly, I'm waiting for the crash-and-burn moment at which someone steps up, gives us the same features, using predictable, modular code or even scripts, and we can put in the bits we like and leave out the bits we don't like and replace any bit and NOBODY will know or care that we've done that.
Systemd itself allows for that, but not system admins. It is modular for developers. So what you would need is say 6-15 guys who change systemd over into a meta process management system with a scriptable configuration system. The question is whether you would use such a thing? And I don't think most of the system admins who object to systemd would. Where I do see potential users is in the embedded space as systemd makes it way into complex embedded systems. Once it is developed for embedded Linux though, it could be used by other distributions. I'd guess you are talking the years 2025-35 is when that becomes available in a mainstream way.
Or even better to have desktop / workstation distributions that are tweaked that for that use. So for example mainly support server applications in developer modes rather than being tweaked for large numbers of users, or have much the ability to shift security levels up and down easily to see if what security feature is introducing a bug in interprocess communications. Caldera, Mandrake, Corel Linux / Xandros...
In complex projects there is a lag. Software isn't done by magic. Once software people decide to start moving towards a standard the actual use of standard in release product lag. Ubuntu announced it couldn't continue with upstart and needed to switch away two years ago. That doesn't cause the repositories to instantly change.
You had just claimed they didn't exist, that it was "utter bullshit" that "process management was part of most of the commercial Unixes and improved them considerably". That they didn't exist.
As for things like daemontools working well... service A depends on B which depends on C. B starts having problems and needs to restart:
i) What should the system do about A? How does it get notified? ii) How the the connection to C reestablished when B is restarted?
That sort of chain can't be handled. Daemontools can restart the simplest services but is structurally unable to handle the complexity of interdependencies. That's not a question of being clumsy or stupid. High volume vs, low volume has nothing to do with additional complexity in process management. And if you had extensive experience you would know that. I think you need to stop throwing around words like stupid, clumsy, bullshit and your claims to extensive experience and rather read about process management and why system administrators want it.
Ubuntu was the originator of Upstart. They announced they were considering the switch about 2 years ago, announced the switch Feb 2014. So I don't know where you are getting 2 days.
AIX System Resource Controller Solaris Service Management Facility Linux had a bunch of cludgy systems like daemon tools.
Etc... So you don't know what you are talking about those things most certainly exist.
As for faster boot times... your claim was what systemd does not one feature it had in 2010. Initial state during boot to running state with daemons states is just one of the states that systemd manages that most certainly is not the entirety or even the primary drivers of its functions.
Sovereign immunity applies to employees, congress hasn't waived it there. To go after them you would have to prove bad faith, illegal action, discrimination...
Systemd has been standard on distributions for about 2 years. There arrows are long past. Debian is a conservative distribution, they went when it was getting painful not to.
You are paranoid. The very first statement is wrong: There are no technical merits of systemd that are important or critical, just some convenience issues Nothing is critical. However in terms of important process management has been important for many years on many operating systems. It was part of most of the commercial Unixes and improved them considerably. So yes it is important.
As far as technical merits the technical merit is it does basic process management and thus allows for hooks to more advances process management.
Debian in so far as they picked either their issue was the license and that got resolves around 1999. KDE3 to 4 was never the cause of Debian tilting towards Gnome.
What do you mean by fair? From from the perspective of Jessie or fair from perspective of some broader Linux community? if you mean the formers: Debian can't control what upstream software producers do. If in 2016 an upstream package has no interest in fixing a security hole that only impacts sysv what are they going to do about it? They would then have to have a security update that could break init scripts.
As for how to handle upgrades on existing servers that have complex code I really don't have much of an opinion. Certainly systemd breaks stuff on existing servers as far as init scripts which is one of the reasons I think changing the default was important to do early before the forking between systemd-linux and sysv-linux got large enough that automated updating became unworkable. I assume basically that Debian wants updates to Jessie to be a lot like new installs and not have to support two systems in the field. So that end users understand version upgrades can break stuff, while minor updates don't. Knowing you are updating from 7 to 8 is the warning that breaking change could occur. And then they are moving them over to the new update.
As for 20 years disappearing overnight. Sort-of yes. The direction systemd is going towards is uniform process management like you see on mainframes and minis and away from the workstation process management that Unix uses. I think that's the right thing to do, but I'm not going to deny that in many ways the people who are supporting the systemd / IaaS / PaaS stacks are trying to put back into Unix the stuff that Ritchie, Thompson... were taking out.
No bigger fish are already doing that. The president of the United States has come out strongly for patent reform. The problem is:
a) Democrats in congress like legal fees b) There are Republicans who believe intellectual property laws should be strengthened.
However there is light. Innovation Act (H.R. 9) is really good and bipartisan. It doesn't fix the system but it does make very good minor change and starts the process of congress getting this system under control. Essentially it tightens up the rules for an infringement filing so that it goes from allowing filing of the form
Product X infringes patent Y to requiring Product X infringes patent Y by doing Z
You are just vastly decreasing the number of lawsuits. That might be a good reform. But essentially the burden on filing is so high and the costs so great that it would almost always (or possibly always) pay to resolve problems outside the justice system. Which means that structures that require lawsuits won't function. For example if debts aren't collectable the loan system collapses. The only way to get a loan would be a loan shark type investor who doesn't depend on the courts.
Your numbers need to be tweaked. 125% cap, liability for 100% of the amount sued for... too high. Something like a 300% cap and 10% liability and I'd agree with you.
I'd add to your list a requirement for due diligence: a party should be encouraged to investigate claims on their own before making them in court. rapid settlement: parties should be encouraged to make offers to one another in a timely matter punishment for stalling: encouragement towards truth telling: admitting flaws in your case rather than requiring the other party to prove them should always be to your advantage reduced cost: the system should work to adjudicate matters of fact cheaply. The expenses should be reasonable relative to the amount of money at stake.
A company in the industry doesn't do enough due diligence to check whether X did infringe or not. I don't have any problem with them getting their head handed to them. You have an obligation in your filings to have very good reason for believing what you are saying is true, and that should be more than just your gut instinct.
Instead, the fees are based on fairness- if you file a frivolous suit, you can plan on being ordered to pay the defendant's costs
The problem is that the courts have a minimalist view of what is frivolous and don't require due diligence or good faith to avoid a claim of frivolousness. If the courts were more aggressive about whether initial fillings were in good faith this would be fine. The loser pays is based on the belief that judges are not going to rulle out huge classes of disputes from being judged even though our society can't afford justice to be cheap.
Better IMHO would be tiers with very different procedures (and thus costs) at each tier: small claims: up to $20k reduced claims $10-200k standard: $100k - 5m enhanced expectations: $2.5m+
As for admitting it is a problem... part of the problem with the USA in terms of good faith is not admitting something is almost always to your advantage. That's something the courts would want to discourage. Admitting a wrong should reduce the claim while failure to admit it should increase it. That encourages people towards settlement.
good luck trying to win a case against Microsoft or IBM or similar patent trolls
A patent troll is a company that doesn't sell goods or services based upon a patent but enforces patent rights. Microsoft and IBM both make their money from making stuff. They are not by any means patent trolls. Patent troll should be reserved for the companies that just buy up patents otherwise you turn an argument about particularly noxious entities into a more fringe argument about intellectual property.
Small wronged person who is sure they have a solid case which they can prove can now sue big corporation and because it is rarer is more likely to be believed Small wrong person who has an iffy case can't sue.
Well you can go bankrupt more easily than they can.
Also uou could set it at a percentage. You sue Sony for $1m you have to put $100k in escrow that they can collect on for fees. You sue them for $1b you have to put $100m in escrow.
There were already emerging chains on Jessie last year where packages needed a recompile or work to run without systemd effectively That's not to say that every system had such chains but that they were starting to emerge and complicate work. The systemd advocates never said Jessie wouldn't work without systemd but rather that: a) Jessie was likely the last version that would work while being a broad based distribution without systemd b) They couldn't insure that upstream packages would continue to support initd based features for 2-3 years so not using it would introduce security issues c) Given the speed of systemd development and its ties with architecture the changeover was likely to be vastly more breaking 3 years later (i.e. 2015 is going to be much easier than 2018).
As for choice and flexibility... Debian is a compiled distribution Most packages don't introduce complex chains of dependencies but some do. And on those Debian has had to make choices about defaults. The way those choices are made is by looking at the direction of upstream. Debian's policy is pretty clear. If the initd people want initd then work with the upstream software to make sure their software is not introducing systemd dependencies and work with the package maintainers or easy option switches. Debian supports that. What they can't support is 2 different distribution. If Devuan ever comes to be then you'll have a long term systemd free Debian. If not Crux, Alpine... exist . No one is taking away choice from Linux.
The FAT32 example might be an anti-trust violation. Most of the other stuff probably has nothing to do with Windows desktop OS however. Remember Microsoft was a big player in phones for many years until Apple / Android.
l. I do, however, care about their OS, the stability and performance of which has been degrading steadily since the loss of Jobs.
That's just false. OSX stability and performance in 10,10 is far far better than say 10.4-6. Take for example the complexity of the video subsystems required to overlay 3 different screens for retina displays. The video subsystem handling of high performance video cards wasn't finished until 10.4 And wasn't stable or usable then. 10.7 is when what 10.7 does became possible. The memory handling for battery life requires a tremendously complex kernel. 10.10 is advanced over 10.9 over 10.8 and really before that you don't have anything remotely as complex.
So I'm going to throw it out this way. What subsystem is less stable or lower performance today and say 5 years ago? Let's hit your list:
Issues like the keyboard and trackpad freezing
That's a bug that gets fixed soon. Apple had bugs in 10.2, 10.3, 10.4...
Messages (which is now part of the OS) using over 2GB of RAM for its own process while making use of another kernel-level process that manages to eat 5GB (watching kernel_task go from over 6GB of RAM to 1.1GB just by closing Messages is freaking silly),
That is. You are loading something else. Run a diagnostic like etrecheck.
I experienced none of these issues in any version of OS X released while Jobs was active within the company.
There were many more bugs in Job’s day. You sound like you have a worm or something, that isn’t OSX.
Care to give any examples of what was un-balanced about Apple's machines under Jobs
Sure. The G4 had terrible throughput for memory and hard drives relative to CPU speed. The result was that the machine pulled a lot of no-ops. It was a bad CPU in a period when Intel CPUs were cheap and much more powerful. The G5 was excellent but then Jobs wouldn’t commit to a laptop version so just as his CPU problems were fixed he migrated away.
Another area where Jobs made sacrifices was on his memory sourcing. Apple customers often had to pay 5x or more street price for memory.
2nd or 3rd in every category isn't beating Android. The players are iPhone, Android, Windows Phone, and Blackberry
By 2nd or 3rd I meant compared to individual phones. i.e. HTC One M9, Samsung Galaxy S6, HTC Desire Eye, Motorola Moto X, Lumia 1520
and major apps that exist on both platforms (like Adobe's suite) are routinely found to perform better on Windows.
While the opposite is true on Android vs. iOS. If this were about Tim Cook that shouldn’t be happening.
In the end we disagree that there has been slippage in the software to any great degree. I don’t disagree with your point philosophically: were OSX’s all around experience worse than Windows the hardware wouldn’t make up for that. What I disagree with you on is a matter of fact, that OSX’s experience is worse.
A trusted application is trusted to authorize applications. That's what it means to trust. If you want applications that are only semi-trusted: capability computing, sandboxing, virtual machines... permissions systems are not the way to go.
Slashdot Mirror
Actually there is. Systemd does lots of things that run at very low levels and thus performance matters. Processes that run multiple times per second aren't scripted in Unix. Unix has always accepted that you use a high level scripting language to tie together low level binary function, not that you script the whole system. What you are asking for is a design like what you see in LISP machines where the scripting languages are designed to degrade into very fast code so that it is safe to script almost the entire OS. Readying your criticism of systemd it would apply equally well to the kernel or to many of the standard libraries on Linux.
The kind of operating system you really want doesn't exist anymore on hardware What you really want is a LISP machine. Quite literally the essential binary part of the operating system was a few hundred assembly language functions and everything else scripted. I think where you do see this is in cloud computing. Many of the web based systems and IaaS manipulations are scriptable easily changeable and modular.
Systemd itself allows for that, but not system admins. It is modular for developers. So what you would need is say 6-15 guys who change systemd over into a meta process management system with a scriptable configuration system. The question is whether you would use such a thing? And I don't think most of the system admins who object to systemd would. Where I do see potential users is in the embedded space as systemd makes it way into complex embedded systems. Once it is developed for embedded Linux though, it could be used by other distributions. I'd guess you are talking the years 2025-35 is when that becomes available in a mainstream way.
Or even better to have desktop / workstation distributions that are tweaked that for that use. So for example mainly support server applications in developer modes rather than being tweaked for large numbers of users, or have much the ability to shift security levels up and down easily to see if what security feature is introducing a bug in interprocess communications. Caldera, Mandrake, Corel Linux / Xandros...
In complex projects there is a lag. Software isn't done by magic. Once software people decide to start moving towards a standard the actual use of standard in release product lag. Ubuntu announced it couldn't continue with upstart and needed to switch away two years ago. That doesn't cause the repositories to instantly change.
You had just claimed they didn't exist, that it was "utter bullshit" that "process management was part of most of the commercial Unixes and improved them considerably". That they didn't exist.
As for things like daemontools working well... service A depends on B which depends on C. B starts having problems and needs to restart:
i) What should the system do about A? How does it get notified?
ii) How the the connection to C reestablished when B is restarted?
That sort of chain can't be handled. Daemontools can restart the simplest services but is structurally unable to handle the complexity of interdependencies. That's not a question of being clumsy or stupid. High volume vs, low volume has nothing to do with additional complexity in process management. And if you had extensive experience you would know that. I think you need to stop throwing around words like stupid, clumsy, bullshit and your claims to extensive experience and rather read about process management and why system administrators want it.
Ubuntu was the originator of Upstart. They announced they were considering the switch about 2 years ago, announced the switch Feb 2014. So I don't know where you are getting 2 days.
AIX System Resource Controller
Solaris Service Management Facility
Linux had a bunch of cludgy systems like daemon tools.
Etc... So you don't know what you are talking about those things most certainly exist.
As for faster boot times ... your claim was what systemd does not one feature it had in 2010. Initial state during boot to running state with daemons states is just one of the states that systemd manages that most certainly is not the entirety or even the primary drivers of its functions.
Sovereign immunity applies to employees, congress hasn't waived it there. To go after them you would have to prove bad faith, illegal action, discrimination...
Systemd has been standard on distributions for about 2 years. There arrows are long past. Debian is a conservative distribution, they went when it was getting painful not to.
You are paranoid. The very first statement is wrong: There are no technical merits of systemd that are important or critical, just some convenience issues Nothing is critical. However in terms of important process management has been important for many years on many operating systems. It was part of most of the commercial Unixes and improved them considerably. So yes it is important.
As far as technical merits the technical merit is it does basic process management and thus allows for hooks to more advances process management.
Debian in so far as they picked either their issue was the license and that got resolves around 1999. KDE3 to 4 was never the cause of Debian tilting towards Gnome.
What do you mean by fair? From from the perspective of Jessie or fair from perspective of some broader Linux community? if you mean the formers: Debian can't control what upstream software producers do. If in 2016 an upstream package has no interest in fixing a security hole that only impacts sysv what are they going to do about it? They would then have to have a security update that could break init scripts.
As for how to handle upgrades on existing servers that have complex code I really don't have much of an opinion. Certainly systemd breaks stuff on existing servers as far as init scripts which is one of the reasons I think changing the default was important to do early before the forking between systemd-linux and sysv-linux got large enough that automated updating became unworkable. I assume basically that Debian wants updates to Jessie to be a lot like new installs and not have to support two systems in the field. So that end users understand version upgrades can break stuff, while minor updates don't. Knowing you are updating from 7 to 8 is the warning that breaking change could occur. And then they are moving them over to the new update.
As for 20 years disappearing overnight. Sort-of yes. The direction systemd is going towards is uniform process management like you see on mainframes and minis and away from the workstation process management that Unix uses. I think that's the right thing to do, but I'm not going to deny that in many ways the people who are supporting the systemd / IaaS / PaaS stacks are trying to put back into Unix the stuff that Ritchie, Thompson... were taking out.
The federal government has sovereign immunity. Congress has a to waive immunity otherwise they can't be sued.
No bigger fish are already doing that. The president of the United States has come out strongly for patent reform. The problem is:
a) Democrats in congress like legal fees
b) There are Republicans who believe intellectual property laws should be strengthened.
However there is light. Innovation Act (H.R. 9) is really good and bipartisan. It doesn't fix the system but it does make very good minor change and starts the process of congress getting this system under control. Essentially it tightens up the rules for an infringement filing so that it goes from allowing filing of the form
Product X infringes patent Y to requiring
Product X infringes patent Y by doing Z
You are just vastly decreasing the number of lawsuits. That might be a good reform. But essentially the burden on filing is so high and the costs so great that it would almost always (or possibly always) pay to resolve problems outside the justice system. Which means that structures that require lawsuits won't function. For example if debts aren't collectable the loan system collapses. The only way to get a loan would be a loan shark type investor who doesn't depend on the courts.
Your numbers need to be tweaked. 125% cap, liability for 100% of the amount sued for... too high. Something like a 300% cap and 10% liability and I'd agree with you.
Very good list of traits for a legal system.
I'd add to your list a requirement for
due diligence: a party should be encouraged to investigate claims on their own before making them in court.
rapid settlement: parties should be encouraged to make offers to one another in a timely matter
punishment for stalling:
encouragement towards truth telling: admitting flaws in your case rather than requiring the other party to prove them should always be to your advantage
reduced cost: the system should work to adjudicate matters of fact cheaply. The expenses should be reasonable relative to the amount of money at stake.
A company in the industry doesn't do enough due diligence to check whether X did infringe or not. I don't have any problem with them getting their head handed to them. You have an obligation in your filings to have very good reason for believing what you are saying is true, and that should be more than just your gut instinct.
The problem is that the courts have a minimalist view of what is frivolous and don't require due diligence or good faith to avoid a claim of frivolousness. If the courts were more aggressive about whether initial fillings were in good faith this would be fine. The loser pays is based on the belief that judges are not going to rulle out huge classes of disputes from being judged even though our society can't afford justice to be cheap.
Better IMHO would be tiers with very different procedures (and thus costs) at each tier:
small claims: up to $20k
reduced claims $10-200k
standard: $100k - 5m
enhanced expectations: $2.5m+
As for admitting it is a problem... part of the problem with the USA in terms of good faith is not admitting something is almost always to your advantage. That's something the courts would want to discourage. Admitting a wrong should reduce the claim while failure to admit it should increase it. That encourages people towards settlement.
A patent troll is a company that doesn't sell goods or services based upon a patent but enforces patent rights. Microsoft and IBM both make their money from making stuff. They are not by any means patent trolls. Patent troll should be reserved for the companies that just buy up patents otherwise you turn an argument about particularly noxious entities into a more fringe argument about intellectual property.
Small wronged person who is sure they have a solid case which they can prove can now sue big corporation and because it is rarer is more likely to be believed
Small wrong person who has an iffy case can't sue.
Well you can go bankrupt more easily than they can.
Also uou could set it at a percentage. You sue Sony for $1m you have to put $100k in escrow that they can collect on for fees. You sue them for $1b you have to put $100m in escrow.
There were already emerging chains on Jessie last year where packages needed a recompile or work to run without systemd effectively That's not to say that every system had such chains but that they were starting to emerge and complicate work. The systemd advocates never said Jessie wouldn't work without systemd but rather that:
a) Jessie was likely the last version that would work while being a broad based distribution without systemd
b) They couldn't insure that upstream packages would continue to support initd based features for 2-3 years so not using it would introduce security issues
c) Given the speed of systemd development and its ties with architecture the changeover was likely to be vastly more breaking 3 years later (i.e. 2015 is going to be much easier than 2018).
As for choice and flexibility... Debian is a compiled distribution Most packages don't introduce complex chains of dependencies but some do. And on those Debian has had to make choices about defaults. The way those choices are made is by looking at the direction of upstream. Debian's policy is pretty clear. If the initd people want initd then work with the upstream software to make sure their software is not introducing systemd dependencies and work with the package maintainers or easy option switches. Debian supports that. What they can't support is 2 different distribution. If Devuan ever comes to be then you'll have a long term systemd free Debian. If not Crux, Alpine... exist . No one is taking away choice from Linux.
The FAT32 example might be an anti-trust violation. Most of the other stuff probably has nothing to do with Windows desktop OS however. Remember Microsoft was a big player in phones for many years until Apple / Android.
That's just false. OSX stability and performance in 10,10 is far far better than say 10.4-6. Take for example the complexity of the video subsystems required to overlay 3 different screens for retina displays. The video subsystem handling of high performance video cards wasn't finished until 10.4 And wasn't stable or usable then. 10.7 is when what 10.7 does became possible. The memory handling for battery life requires a tremendously complex kernel. 10.10 is advanced over 10.9 over 10.8 and really before that you don't have anything remotely as complex.
So I'm going to throw it out this way. What subsystem is less stable or lower performance today and say 5 years ago? Let's hit your list:
That's a bug that gets fixed soon. Apple had bugs in 10.2, 10.3, 10.4...
There were many more bugs in Job’s day. You sound like you have a worm or something, that isn’t OSX.
Sure.
The G4 had terrible throughput for memory and hard drives relative to CPU speed. The result was that the machine pulled a lot of no-ops. It was a bad CPU in a period when Intel CPUs were cheap and much more powerful. The G5 was excellent but then Jobs wouldn’t commit to a laptop version so just as his CPU problems were fixed he migrated away.
Another area where Jobs made sacrifices was on his memory sourcing. Apple customers often had to pay 5x or more street price for memory.
By 2nd or 3rd I meant compared to individual phones. i.e. HTC One M9, Samsung Galaxy S6, HTC Desire Eye, Motorola Moto X, Lumia 1520
While the opposite is true on Android vs. iOS. If this were about Tim Cook that shouldn’t be happening.
In the end we disagree that there has been slippage in the software to any great degree. I don’t disagree with your point philosophically: were OSX’s all around experience worse than Windows the hardware wouldn’t make up for that. What I disagree with you on is a matter of fact, that OSX’s experience is worse.
A trusted application is trusted to authorize applications. That's what it means to trust. If you want applications that are only semi-trusted: capability computing, sandboxing, virtual machines... permissions systems are not the way to go.
Well clearly it is not the best OS in 2015. Yet they continued using it for years. Ergo...