It look like we are from the same generation, but contrary to you I absolutely like the systemd move. This situation is not new: Git faced the same kind of critics for example, as udev has also over the static/dev.
The basic of systemd is learned in a single day of work, really. For all the others issue, you will find the answer in a minute on the internet. Maybe you have to learn hard to understand system V init and his reluctant stack of script at his time, and that experience make you feel like you are an expert about it. But today way of learning something like systemd is accessible to a very big amount of peoples.
Or maybe you are too old to understand that system V init have absolutely no way to provides so much features that systemd bring on the table.
I use Debian since Bo, and I delivered this week to a customer my first embedded system based on Jessie with systemd. From this background, I say that Jessie is probably the best Debian release since his creation. For me, Jessie make SB2, WRT, Buildroot, OpenEmbedded, Yocto and all the others totally obsolete. Never was so happy to work building embedded systems than with Debian Jessie. Multiarch, systemd and MATE are exactly what I need.
I updated from system V init to systemd on the embedded system during the development and this was surprisingly smooth. Backward compatibility with system V init worked out of the box and the few minors issues was usual part of a first deployment or specific to my target.
I really don't care of any others operating system since at least 20 years. Fine for me. There still have system V init, and system V init is still running on Linux. What the problem ? Linus didn't ask others OS before making system calls that are unique to Linux. Why systemd (or any applications for that matter) should be prevented to use them ?
Remember me how the Node.js team reacted to my proposition to use Linux timerfd to improve the repeatable timer precision: there removed the repeatable timer from the API because Windows would have nothing to match the Linux timerfd precision.
I first used systemd on a complex custom embedded project that dynamically switch between multiple loads and it have saved me days of work compared to previous designs. Dependencies are managed the right way, and status+analysis tools are very good.
PulseAudio is unstable I agree (I juste deleted a ~1.5GB.xsessionerror filled with insanity from it, and it crash at least one a day), but not so systemd, even if journald need some more work to fix some minor issues.
And by posting here with your infected machine you spread even more your disaster... Now my shinny new 4K screen degraded to early Archean in a stack of mineral.
systemd service files are far less complex that the stack of init scripts need on top of system V init. The status and reporting is incomparably better than anything before it.
I actually found systemd reporting, status and analysis tools pretty verbose compared to the corresponding system V init scripts. What kind of bug did you try to catch ?
Gnome 2, MATE, XCFE, and maybe other desktops allow vertical toolbars too if you like it (not my case). Alt-TAB window switching work equally well on all of them too. I personally found both Gnome 3 and Unity catastrophic in term of productivity. I actually work with close to 40 virtual desktops and over 100 windows. I configured to switch around them using right CTRL+arrow keys, without any animation or effect. Blazing fast and simple.
AFAIK in a electric car, the motor and his surrounding infrastructure is cheap compared to an thermal combustion motor. The 'only' problem is the storage where the cost, mass, capacity and longevity is hard to compare with a simple tank filled by fuel. I am confident that at some point an innovative solution will open the path of way to overcome this limitation at an acceptable level for a bigger chunk of the car mass market.
I found your point interesting. I we have to replace all the fossil petroleum by synthetic fuel, how costly will be to do so compared of the same amount of electric energy ?
I wondering how much a energy a given surface can generate in one year if it was used to grow organic stuff (taking water and insecticide in the process) and then distilled into usable fuel, vs if the same given surface was used to produce electricity using solar panel and wind power plant.
Not certain that you get the good picture, probably because that my explanation and my English are not good enough. The parliament can vote to appoint any citizen (almost everybody) as a member of the head of state (The Federal Council: the highest political level), even a citizen that have do nothing to be voted for. Maybe this URL is worth reading http://en.wikipedia.org/wiki/S...
But the main feature is that there vote 7 members, not just one, so this grant that all leading parties are represented up to the highest level, lowering greatly the frustration of a large part of the population and forcing every parties to continuously and constructively negotiate between them. The whole process is under direct control of all the citizens that can revert any decision and even impose new decision to the government. The Federal Council, the direct democracy by referendum and initiative, are something different from most other country in Europe. Some countries have referendum and initiative in there constitution but in a form that make then more theoretical than a real powerful tool practiced many time per years by all citizens. For example, French citizens referendum are excessively rare and the result of the last most important one was inverted by the government than disliked the popular decision. No better way to spread frustration at large scale and break confidence on the political system...
The Swiss constitution was mainly setup after the civil war of 1847. At that time enough politics worked seriously together to compare various existing foreign constitutions. The French Revolution and the USA constitution played a major role in the redaction of the first text. The major Swiss ingredient was to introduce a proportional representation of the parties up to the head of state. A new version of the constitution adjusted many parameters and introduced the referendum, and the version just after introduced the popular initiative.
It was a constructive work to fix the problems found in the early versions. It mainly work because of the first decision of making a proportional head of state. Without this feature, too many effort is lost by the parties to place there members on the most powerful place, making them willing to promise the impossible to take more vote: there are basically required to lie and to play on the emotional level. And when you play with lie and emotion, you get something unmanageable that split the citizens mostly between two main parties that constantly fight each at the other. Whenever the party that win, the system mostly grant to have as close as possible a rate of half of citizen frustrated by the result of the vote. By contrast the Swiss political system limit the possibility that a significant opposition exists because it will be integrated early in the government.
The argument that this work only because this is a small country have no merit. The Switzerland with 26 strong sates with each there own constitution, and 4 official languages, is used at a bigger scale than in many bigger countries where the political system is far too much centralized. The current state of the multiple parties coalition in Germany is interesting and is obviously a good step forward the proportional representation, even if it's informal. Juts hope that the will not miss the opportunity to grant that in an updated constitution.
The citizens vote to elect the two parliament chambers members, representing proportionally the states and the peoples. Then the parliament (all members of the two chambers) vote to appoint the 7 Federal Council members. There can appoint any citizen of the country even it's not a candidate. As strange as it look like, yes this has happened: a citizen without high view on the politic can be boosted up to the highest level in a few hours. The main goal is to form a stable Federal Council with a proportional diversity of the leading parties. Then 7 Federal Council members decide by them self there minister assignation between them.
The same happens in all states. Prime ministers to presidents all do that.
Not in Switzerland for example.
Here the parliament (directly elected by the citizens) elect the Federal Council where each of this 7 members act as minister. There is no upper layer like prime minister or formal president (there is an annually elected president into the Federal Council, but this is strictly honorific without any added power). So each leading parties are fully represented in every politic layer up to the head of state, making a lot of conflict useless. The Federal Council is required to act as a single body after there have voted on a decision. Every decision can be changed back by citizens using referendum and every change to the constitution require a mandatory referendum, so the parliament and the Federal Council cannot act against the citizen wishes. In addition citizens can force a change in the constitution using the popular initiative. Here every citizens vote on a lot of important subjects many times per years making everyone more concerned about the politic of the country.
Exactly the contrary of the French citizens where there live in a modern kind of monarchy from a small self proclaimed elites that fight to place there favorite king at the most powerful place to mainly serve there own interest. I really hope that there will quickly evolve to something better.
It's only a democratic country because there can vote for the king, but the king choose the ministers and the ministers can do anything. The parliament is a joke to keep some credibility that some discussion exists. The citizen are left with there problems and never take seriously.
Re:Remote link failure imply stand alone operation
on
Planes Without Pilots
·
· Score: 1
Sorry for my English, it's not my native language.
Yes we fully agree that automation on board is the best solution. This is critically important for safety operation, far more than a remote co-pilot.
From my point of view radar and radio are the same category of system as civil aircraft operator mostly use secondary radar that rely on a radio link to get back information from the aircraft. Primary ground radar will be useless for the aircraft until it send information to them, but this again depend on a radio link.
Re:Remote link failure imply stand alone operation
on
Planes Without Pilots
·
· Score: 1
Radar failure rate show how fragile is a radio link. Now you have to remember that this link is required to have a kind of "ground based co-pilot". So the availability of the remote co-pilot will be far lower than a on board co-pilot, human or automate. The net result of a remote co-pilot is a decease of the safety level. On board automated co-pilot can potentially increase the safety level because it's possible to have multiple instance of them to increase there availability. It's a fairly simple logic. An other hint is that complex systems that need to maintain safety under degraded operational conditions use distributed autonomous subsystems not central radio link.
If you are still not convinced, just look at accident reports where the stress on the flight deck required the full attention of the two pilots very quickly to save the situation. A remote co-pilot will be completely unable to do anything useful, especially if it already monitor multiples flights at the same time. An on board automated co-pilot can react immediately, and probably even faster than an human.
Re:Remote link failure imply stand alone operation
on
Planes Without Pilots
·
· Score: 1
You forget that the link is only a link: the ground facility could be unable to operate for hours for any reason.
Re:Remote link failure imply stand alone operation
on
Planes Without Pilots
·
· Score: 1
all planes start circling at different altitudes until the system is working again..
Certainly not an option for a failure that last longer than a single hour as aircraft will risk to run out of fuel. There is a chunk of documented incidents at ground level that affected a lot of flight and that lasted far more than a single hour. Aircraft need to be able to land automatically by itself without remote operation and there need to use it every day to ensure that the system is at the required safety level. There is nothing worst than a safety feature that is only used in case of a incident: most of the issues will be hidden until the incidents.
Remote link failure imply stand alone operation
on
Planes Without Pilots
·
· Score: 2
Remote operation imply data transfer, usually by radio, and this is the weakest part of the system. To ensure that the aircraft stay in safe operation without remote link, it fist must be able to sustain stand alone operation, including landing before running out of fuel.
Concentrating control to a single point will increase the risk, concentrating control to a remote point will add an another layer of risk. The only solution is to allow distributed control, and this imply that each aircraft is able to operate safely by itself without remote control.
If you are not convinced, just imagine that a remote control point is unable to operate for some reason: you have now dozen of flight without co-pilot and there need to all land as soon as possible, raising a another wave of problems.
If I understand you correctly, this confirm the possibility that Microsoft have the possibility to manage 2 classes of keys: the first keys class is the current one where Microsoft is willing to sign binaries not from them; the second keys class could be for 'lock in' machines where Microsoft keep full control on.
To be fair, I think that the 'lock in' keys class it's a logical step for Microsoft branded machines. But this could go very wrong if OEMs start to do the same by using the argument 'designed for Microsoft OS' because this will add 'and nothing else could run on it' to the argument. I suspect that the goal is to reserve top machines specifications to Microsoft and to only allow degraded specifications machines to run other OS. The market already have products with this kind of bias.
And yes, you are right. This evil plan was draw decades ago with the deep knowledge that it will only work at the time when the security feature will be so standard that no chip will be manufactured without it anymore.
The fact that Windows 10 is announced to be virtually free for almost everyone having a previous copy of Windows somewhere is a clear singe that the time have changed. The OS have no value anymore. The number of new software that only run on a single OS will drastically shrink, exacerbating the OS value problem. So the 'lock in' machines with exclusive specifications will be the only market where Microsoft could make money from the OS.
From my analysis, the Microsoft message is dual: 1) you don't need anything other that Windows 10 as it's virtually free for everyone; 2) You need Windows 10 to run top specifications machines. OEM market will almost certainly split the product range accordingly if no reaction prevent this.
Yes. Microsoft could release a special version of his OS for 'lock in' machines with a new key that only boot on secure BIOS from this machines. This will make this machines unable to run any previous OS, including Linux. Or did I miss something ?
It look like we are from the same generation, but contrary to you I absolutely like the systemd move. This situation is not new: Git faced the same kind of critics for example, as udev has also over the static /dev.
The basic of systemd is learned in a single day of work, really. For all the others issue, you will find the answer in a minute on the internet. Maybe you have to learn hard to understand system V init and his reluctant stack of script at his time, and that experience make you feel like you are an expert about it. But today way of learning something like systemd is accessible to a very big amount of peoples.
Or maybe you are too old to understand that system V init have absolutely no way to provides so much features that systemd bring on the table.
I use Debian since Bo, and I delivered this week to a customer my first embedded system based on Jessie with systemd. From this background, I say that Jessie is probably the best Debian release since his creation. For me, Jessie make SB2, WRT, Buildroot, OpenEmbedded, Yocto and all the others totally obsolete. Never was so happy to work building embedded systems than with Debian Jessie. Multiarch, systemd and MATE are exactly what I need.
I updated from system V init to systemd on the embedded system during the development and this was surprisingly smooth. Backward compatibility with system V init worked out of the box and the few minors issues was usual part of a first deployment or specific to my target.
I really don't care of any others operating system since at least 20 years. Fine for me. There still have system V init, and system V init is still running on Linux. What the problem ? Linus didn't ask others OS before making system calls that are unique to Linux. Why systemd (or any applications for that matter) should be prevented to use them ?
Remember me how the Node.js team reacted to my proposition to use Linux timerfd to improve the repeatable timer precision: there removed the repeatable timer from the API because Windows would have nothing to match the Linux timerfd precision.
I first used systemd on a complex custom embedded project that dynamically switch between multiple loads and it have saved me days of work compared to previous designs. Dependencies are managed the right way, and status+analysis tools are very good.
PulseAudio is unstable I agree (I juste deleted a ~1.5GB .xsessionerror filled with insanity from it, and it crash at least one a day), but not so systemd, even if journald need some more work to fix some minor issues.
And by posting here with your infected machine you spread even more your disaster... Now my shinny new 4K screen degraded to early Archean in a stack of mineral.
Are you certain that Debian fit your definition ?
systemd service files are far less complex that the stack of init scripts need on top of system V init. The status and reporting is incomparably better than anything before it.
I actually found systemd reporting, status and analysis tools pretty verbose compared to the corresponding system V init scripts.
What kind of bug did you try to catch ?
Gnome 2, MATE, XCFE, and maybe other desktops allow vertical toolbars too if you like it (not my case).
Alt-TAB window switching work equally well on all of them too.
I personally found both Gnome 3 and Unity catastrophic in term of productivity. I actually work with close to 40 virtual desktops and over 100 windows. I configured to switch around them using right CTRL+arrow keys, without any animation or effect. Blazing fast and simple.
Debian Jessie with MATE is very good too. Should be released tomorrow AFAIK.
AFAIK in a electric car, the motor and his surrounding infrastructure is cheap compared to an thermal combustion motor. The 'only' problem is the storage where the cost, mass, capacity and longevity is hard to compare with a simple tank filled by fuel. I am confident that at some point an innovative solution will open the path of way to overcome this limitation at an acceptable level for a bigger chunk of the car mass market.
I found your point interesting. I we have to replace all the fossil petroleum by synthetic fuel, how costly will be to do so compared of the same amount of electric energy ?
I wondering how much a energy a given surface can generate in one year if it was used to grow organic stuff (taking water and insecticide in the process) and then distilled into usable fuel, vs if the same given surface was used to produce electricity using solar panel and wind power plant.
Not certain that you get the good picture, probably because that my explanation and my English are not good enough. The parliament can vote to appoint any citizen (almost everybody) as a member of the head of state (The Federal Council: the highest political level), even a citizen that have do nothing to be voted for. Maybe this URL is worth reading http://en.wikipedia.org/wiki/S...
But the main feature is that there vote 7 members, not just one, so this grant that all leading parties are represented up to the highest level, lowering greatly the frustration of a large part of the population and forcing every parties to continuously and constructively negotiate between them. The whole process is under direct control of all the citizens that can revert any decision and even impose new decision to the government. The Federal Council, the direct democracy by referendum and initiative, are something different from most other country in Europe. Some countries have referendum and initiative in there constitution but in a form that make then more theoretical than a real powerful tool practiced many time per years by all citizens. For example, French citizens referendum are excessively rare and the result of the last most important one was inverted by the government than disliked the popular decision. No better way to spread frustration at large scale and break confidence on the political system...
The Swiss constitution was mainly setup after the civil war of 1847. At that time enough politics worked seriously together to compare various existing foreign constitutions. The French Revolution and the USA constitution played a major role in the redaction of the first text. The major Swiss ingredient was to introduce a proportional representation of the parties up to the head of state. A new version of the constitution adjusted many parameters and introduced the referendum, and the version just after introduced the popular initiative.
It was a constructive work to fix the problems found in the early versions. It mainly work because of the first decision of making a proportional head of state. Without this feature, too many effort is lost by the parties to place there members on the most powerful place, making them willing to promise the impossible to take more vote: there are basically required to lie and to play on the emotional level. And when you play with lie and emotion, you get something unmanageable that split the citizens mostly between two main parties that constantly fight each at the other. Whenever the party that win, the system mostly grant to have as close as possible a rate of half of citizen frustrated by the result of the vote. By contrast the Swiss political system limit the possibility that a significant opposition exists because it will be integrated early in the government.
The argument that this work only because this is a small country have no merit. The Switzerland with 26 strong sates with each there own constitution, and 4 official languages, is used at a bigger scale than in many bigger countries where the political system is far too much centralized. The current state of the multiple parties coalition in Germany is interesting and is obviously a good step forward the proportional representation, even if it's informal. Juts hope that the will not miss the opportunity to grant that in an updated constitution.
The citizens vote to elect the two parliament chambers members, representing proportionally the states and the peoples. Then the parliament (all members of the two chambers) vote to appoint the 7 Federal Council members. There can appoint any citizen of the country even it's not a candidate. As strange as it look like, yes this has happened: a citizen without high view on the politic can be boosted up to the highest level in a few hours. The main goal is to form a stable Federal Council with a proportional diversity of the leading parties. Then 7 Federal Council members decide by them self there minister assignation between them.
The same happens in all states. Prime ministers to presidents all do that.
Not in Switzerland for example.
Here the parliament (directly elected by the citizens) elect the Federal Council where each of this 7 members act as minister. There is no upper layer like prime minister or formal president (there is an annually elected president into the Federal Council, but this is strictly honorific without any added power). So each leading parties are fully represented in every politic layer up to the head of state, making a lot of conflict useless. The Federal Council is required to act as a single body after there have voted on a decision. Every decision can be changed back by citizens using referendum and every change to the constitution require a mandatory referendum, so the parliament and the Federal Council cannot act against the citizen wishes. In addition citizens can force a change in the constitution using the popular initiative. Here every citizens vote on a lot of important subjects many times per years making everyone more concerned about the politic of the country.
Exactly the contrary of the French citizens where there live in a modern kind of monarchy from a small self proclaimed elites that fight to place there favorite king at the most powerful place to mainly serve there own interest. I really hope that there will quickly evolve to something better.
It's only a democratic country because there can vote for the king, but the king choose the ministers and the ministers can do anything. The parliament is a joke to keep some credibility that some discussion exists. The citizen are left with there problems and never take seriously.
Sorry for my English, it's not my native language.
Yes we fully agree that automation on board is the best solution. This is critically important for safety operation, far more than a remote co-pilot.
From my point of view radar and radio are the same category of system as civil aircraft operator mostly use secondary radar that rely on a radio link to get back information from the aircraft. Primary ground radar will be useless for the aircraft until it send information to them, but this again depend on a radio link.
Radar failure rate show how fragile is a radio link. Now you have to remember that this link is required to have a kind of "ground based co-pilot". So the availability of the remote co-pilot will be far lower than a on board co-pilot, human or automate. The net result of a remote co-pilot is a decease of the safety level. On board automated co-pilot can potentially increase the safety level because it's possible to have multiple instance of them to increase there availability. It's a fairly simple logic. An other hint is that complex systems that need to maintain safety under degraded operational conditions use distributed autonomous subsystems not central radio link.
If you are still not convinced, just look at accident reports where the stress on the flight deck required the full attention of the two pilots very quickly to save the situation. A remote co-pilot will be completely unable to do anything useful, especially if it already monitor multiples flights at the same time. An on board automated co-pilot can react immediately, and probably even faster than an human.
You forget that the link is only a link: the ground facility could be unable to operate for hours for any reason.
all planes start circling at different altitudes until the system is working again..
Certainly not an option for a failure that last longer than a single hour as aircraft will risk to run out of fuel. There is a chunk of documented incidents at ground level that affected a lot of flight and that lasted far more than a single hour. Aircraft need to be able to land automatically by itself without remote operation and there need to use it every day to ensure that the system is at the required safety level. There is nothing worst than a safety feature that is only used in case of a incident: most of the issues will be hidden until the incidents.
Remote operation imply data transfer, usually by radio, and this is the weakest part of the system. To ensure that the aircraft stay in safe operation without remote link, it fist must be able to sustain stand alone operation, including landing before running out of fuel.
Concentrating control to a single point will increase the risk, concentrating control to a remote point will add an another layer of risk. The only solution is to allow distributed control, and this imply that each aircraft is able to operate safely by itself without remote control.
If you are not convinced, just imagine that a remote control point is unable to operate for some reason: you have now dozen of flight without co-pilot and there need to all land as soon as possible, raising a another wave of problems.
http://hcc.web.cern.ch/hcc/
If I understand you correctly, this confirm the possibility that Microsoft have the possibility to manage 2 classes of keys: the first keys class is the current one where Microsoft is willing to sign binaries not from them; the second keys class could be for 'lock in' machines where Microsoft keep full control on.
To be fair, I think that the 'lock in' keys class it's a logical step for Microsoft branded machines. But this could go very wrong if OEMs start to do the same by using the argument 'designed for Microsoft OS' because this will add 'and nothing else could run on it' to the argument. I suspect that the goal is to reserve top machines specifications to Microsoft and to only allow degraded specifications machines to run other OS. The market already have products with this kind of bias.
And yes, you are right. This evil plan was draw decades ago with the deep knowledge that it will only work at the time when the security feature will be so standard that no chip will be manufactured without it anymore.
The fact that Windows 10 is announced to be virtually free for almost everyone having a previous copy of Windows somewhere is a clear singe that the time have changed. The OS have no value anymore. The number of new software that only run on a single OS will drastically shrink, exacerbating the OS value problem. So the 'lock in' machines with exclusive specifications will be the only market where Microsoft could make money from the OS.
From my analysis, the Microsoft message is dual: 1) you don't need anything other that Windows 10 as it's virtually free for everyone; 2) You need Windows 10 to run top specifications machines. OEM market will almost certainly split the product range accordingly if no reaction prevent this.
Yes. Microsoft could release a special version of his OS for 'lock in' machines with a new key that only boot on secure BIOS from this machines. This will make this machines unable to run any previous OS, including Linux. Or did I miss something ?