But they should probably not be introduced before they are completely bug-free -- or at least more bug-free then the thing they will replace.
Mature software is almost always vastly less buggy than newer feature rich software. In any cycle of improvement the less buggy software is replaced with more feature rich software.
And considering that the *ix world is full of people who don't like change
I don't know that. I think there is a some change resistance in Linux now that didn't used to exist. The Unix world used to love change. I think it is a generational shift since the early 2000s. But the Android user base which is the vast majority of the *ix world seems pretty happy with the changes. As do iOS and OSX users. And frankly most Linux desktop users like systemd. And frankly most server people are using cloud solutions which either have or will shortly be switching to systemd easily to take advantage of those features.
There is a vocal minority what doesn't like this change.
doesn't solve the problem
Of course it does solve the problem which means you are just making stuff up.
That's well put and fairly characterizes the choices that Debian is going to be faced with going forward which is rare in the systemd debate. Agree 100% and well done.
That being said the bug he seems to be upset about is automatically switching old systems on upgrade.
I'm with you. I bought it for $99 and my daughter absolutely loves it, uses it all the time. I've used it more than enough to be worth the money, as has my wife though the split is like 90/5/5%
He most does support the 4 freedoms) He supports the existence of the kernel forks. So obviously the ability to modify and redistribute. His product is open source so the ability to change.There are no purpose restrictions nor have they ever been discussed.
He disagrees with RMS when it comes to the embedded space. He had strong disagreements with RMS on an architectural level about the directions of the GNU project which at this point even RMS has abandoned. In open source vs. free software and how best to relate to companies and commercial software entities i.e. political strategy they disagreed.
I think it is best to say the almost completely agreed on ends and had some strategic and tactical disputes on how to achieve those ends.
___
And no he does not think the Linux kernel should be the only kernel. He knows kernels too well. There are big advantages to some of the other kernels out there. What he wants is a free software ecosystem same as RMS.
The USA did not spend billions on any particular party. And in fact the Chinese do spend money working the economy of Canada. They do trade quite a bit more than $5b.
What Nuland said was, "Since the declaration of Ukrainian independence in 1991, the United States supported the Ukrainians in the development of democratic institutions and skills in promoting civil society and a good form of government - all that is necessary to achieve the objectives of Ukraine’s European. We have invested more than 5 billion dollars to help Ukraine to achieve these and other goals.” That's not sponsoring a junta that's working with a new quasi-European country over a quarter century.
After the coup, the U.S. immediately blessed the junta with legitimacy
That's not sponsorship that's friendship. I agree the USA is friendly towards Poroshenko. I'm disagreeing with sponsorship. The USA disagrees there was coup. They saw it as the population and the legislative branch replacing an executive. Turchynov became president. We have the same kind of mechanism in America if our president were to lose control. Arguably that's fairly close to how Gerald Ford (who had been minority leader) became president. Which incidentally was also a replacement that we immediately granted legitimacy to.
So was Ukraine, before the west decided that resource exploitation and crushing IMF austerity loans were too important to wait six months for the next set of elections, even with that $5 billion in pre-coup "aid".
What are you talking about? There was a disagreement about policy in Ukraine. There were mass protests. The president began slaughtering his own population. He was replaced by the legislature. That's what happened. We aren't the ones who talked Yanukovych into massacring his population and whipping up a popular revolt. That was the Ukrainians.
Saying that systemd is the right direction for mainstream Linux is a much weaker statement than saying it is the right solution for every use case of a computer on the planet. It might be that weather stations should use something else possibly not even Linux.
That being said...
a) Systemd exports plaintext logs. If you want that you can have it. Better though would be to have systemd just export to you regularly. Your system should be queuing the weatherstation directly.
b) If roadside weather stations are that complex to fix, having redundant equipment sounds like a good idea. There should be more than one system onsite including potentially a diagnostics box.
c) Certainly having a standard repair kit distributed to your service agency would be paid for by even one of those 600km trips. Give them a USB or DVD loaded with the appropriate software that boots their systems. Smarthand job is to get stuff connected. He shouldn't be doing anything in terms of looking IMHO.
Why? Because it's so under-utilized that it's practically free. People just don't use servers for consuming stuff.
No that's not why. I've already explained to you why. You can continue to live in fantasy world or you can look at how everyone, including Amazon, is charged.
Employees in companies all the time have ideas. RedHat sponsored the project and put their institutional weight behind it. The fact that in 2010 Poetering thought of it as a personal project not withstanding.
When US is sponsoring junta in Ukraine (Russia's doorstep) and expect Russia to roll over and be cooperative, that's unreasonable
How is the USA sponsoring the new Ukrainian government? Friendly sure. Sponsoring? As for whether it is reasonable or unreasonable the question is what the criticism was in the USA media. The GP was misstating facts.
What would US do if Russia started 'democratic movements' in Mexico, or Texas with the goal to moving them away from US influence and towards 'independence'.
Mexico is independent. As for Texas it has a democratic movement. RT operates freely in the United States.
When has our government every said #1? I don't even know what you are talking about with respect to #2. And as for #3, that's Russian propaganda. I can't think of anyone talking about a Russian attack on America.
I think you should drop the dollar signs nonsense. And yes I've heard of the World Trade Organization the IMF and World Bank. They aren't anti-Russian Russia is a member of these organizations.
And membership in those organizations is voluntary. There are plenty of countries that are not members of any of them and plenty that have not taken loans or acted on their advice. The USA and Russia both being examples that have policies that disagree with the all 3.
What does you not likely the TPP (which I assume your government signed) have to do with the US media being anti-Russian. If you don't like your government's trade policies then work to change them. The TPP isn't something forced on anyone.
What hostility to Russia? The American media is mostly indifferent to Russia. When they aren't indifferent they either tend to be rather laudatory about the culture: ice skating, gymnastics, art... or hostile towards Russia being uncooperative with the USA.
Now the Russian media is genuinely hostile and paranoid.
WOW, no. Init was/is happy to get out of the way if you want to do process management.
It can't "get out of the way" part of process management is starting processes. It either does it or it doesn't do it. I get that process management can be done with init as a lower level. systemd takes a different approach where process initiation just becomes another aspect of process management.
So systemd is entirely inadequate? Because [runlevels ] is all it currently offers.
Not at all. The Requires and After keywords in the systemd system replace runlevels. There really is no need for runlevels with systemd at all. The whole concept is legacy. Distributions probably still make use of it mostly for compatibility.
One day, it might, but for today syslogd can log remotely and journald can't.
Hold on. journalctl --output=export sends remote log. It isn't a full featured remote logging system but most certainly it does do what syslogd does.
Moreover, you are contradicting yourself a bit. You basically arguing that system X is inadequate if there exists any feature what-so-ever it doesn't do. At the same time you are arguing that systems should have minimal feature sets. Systemd already offers way more features than init and the entire family. Systemd has different priorities and focuses. It is a paradigm shift. When it introduces full featured remote logging it is going to be tied to something much more sophisticated than syslogd, and that is future. Today though it is perfectly capable to spewing a stream on events over the network for a listener to pickup.
But in the real world, things systemd is "going to do one day" don't count today.
What is your entire charge about forcing people to do stuff other than forward looking statements? Again you are trying to have this both ways. If you want to make likely estimates about the future that's fine. If the only thing that counts is what currently has happened on November 3, 2014 then none of the forward looking statements about possible future problems apply.
You keep missing that I do advocate for a better system,
What does advocating for a better system mean in practice? Who is writing it? Where is this work being done? What is the project plan? When is it going to be completed? Who is paying? If 5 years from now a better system that systemd shows up then everyone can switch over to that. Systemd doesn't block the Unix community from moving to a better process management system or abandoning the idea of process management. If anyone wants to keep working on OpenRC no one is stopping them. But as of today, they are way behind and falling further behind. Unless you (or Ian) have some plausible plan then advocating doesn't mean anything.
I get that the anti-systemd people in some vague sense would tolerate better like your example of how xinitd allowed lots of servers without memory management. But mostly they are hostile to the paradigm shift. They object to better functionality on desktop. They object to better functionality for PaaS. That's the two uses cases they are objecting to.
and only claim that init does well enough today
Except that it doesn't do well enough today. I gave you a pretty simple chain and init couldn't come close to handling it. Init does nothing well. It is absolutely minimal process starting system. A whole range of hacks and problems exist because init sucks at its job. Going back decades, cron only exists because of deficiencies in init. The reason programs are switching to systemd is because they unhappy with the feature set of init. That's a problem today.
And even what it claims to be good at, it sucks at. On most systems falling back to lower runlevels doesn't actually work. I've tried taking a system
You DO realizew that gtk is a set of gimp libraries, yes? You know Gimp is a photoshop like app, yes? Would you care to have photoshop load every time yuou use notepad?
No of course I wouldn't want the app to load. But I would want gtk to have features from Gimp so that other applications can use chunks of Gimp code to accomplish different tasks. This is the way Cocoa works. For example webkit is used by a 1/2 dozen applications on OSX and iOS to accomplish web features even if you aren't using Safari.
Upload (out) is what's expensive. Download (in) is what's nearly free.
You have upload and download reversed it is to and from them not to and from your local systems. But ignoring the terminology reversal, they don't charge you for the bandwidth when you are putting stuff on their server because they are going to collect storage fees.
; they charge those particular prices because of supply and demand.
No they charge those particular prices because in one direction they make lots of money on storage and in the other they either make nothing or are going to lose their storage fees.
The fixed costs and cost of the pipe is irrelevant; the price they charge is chosen because that is what that market can bear.
The poster is arguing the AWS reflects on the cost of the pipe.
Consider, what will you do if systemd makes a bigger hash of it on your preferred hardware? Go back to the older stuff that worked, of course. But for some reason you can't get the GUI desktop to come up any more. I guess it depended on systemd which fails miserably on your laptop.
I don't see why systemd would make a hash of my hardware rather than be better. What I would do is use a distribution that didn't depend on systemd. Linux is a diverse ecosystem of software. Or if that didn't work, pick between Linux and the hardware. Like I said I currently have hardware Linux doesn't work on. I'm not blaming initd for that. But the power management problems I and everyone has had before well yeah that was initd's fault. Initd didn't offer enough services to do power management well easily. Windows did a great job because of the how deeply embedded powerstate was in Windows. Integration helped. OSX BTW lagged for similar reasons though with they have in the last few years pulled way ahead of Windows.
And it sounds like the older, simpler distros worlked better than the ones with dbus and gnome and such with consolekit and policykit.
No that's not the issue. I suspect if I were still using the older hardware the more complex distributions would boot. It is the more complex hardware (EFI, dual GPU, complex resolution and sizing, lots of inputs on trackpad...) that is throwing the the newer distributions. The Linux community used to work hard on getting all x86 hardware to work. There is a push away from desktop Linux and one of the falloffs is that Linux just isn't working as hard on hassle free installs.
So why were you so happy to have gtk depend on gimp? It is both silly and counterproductive.
Because that's not the form it would take. The form it would take would be is what I described gtk depending on gimp libraries. Which would mean that all gtk applications would get a rich collection of routines from gimp that could be easily integrated into gtk applications. Which is very similar to what does exist with Cocoa and the Quartz graphical libraries. Having the widget set have additional features is a good thing about Cocoa not a bad thing.
Amazon is allowing you to upload free because when you upload you are generally going to be buying storage. That doesn't reflect Amazon's underlying network costs. They don't have the same cost structures as carriers because AWS offers far more services. So, no it does not cost Amazon 0 for you to upload data. They offset the cost. When they aren't offsetting the cost they charge between $30-120 / tb to move data out. Which is being sold at a reasonable profit but is more reflective of what the actual cost of bandwidth is.
Your ISP can similarly offset home costs by making money on selling for example television.
In any event, there's this thing called supply and demand, which says (among other things) that prices of products are set completely independently of their cost.
I think you should go back read about supply and demand. The supply curve is a curve which indicates how supply increases as cost increases. The whole point of the intersection point being on that curve is that price is not independent of cost of providing the service. In particular, carriers are not going to give you services for far less than they cost to deliver.
. Why do you claim that power management on sysvinit sucks but suddenly you think it's great when systemd does the same thing no better?
I didn't claim that. What I claimed was that power management properly belongs with process management because of the advantages it could offer. You then claimed that systemd didn't implement power management and I was showing that it most certainly did. Now on top of all that, it is already better in terms of success it already works better than it did with init.
So change runlevels?
No. Changing runlevels is shutting process down and breaking potential dependencies. That isn't suspending or hibernating them. Telling a process to shutdown is different than telling a process to wrap up immediate problems possibly save data and sleep. Or being able to know the state of the process to evaluate when it is safe to swap it out.
Too bad people typically need to diagnose failure when the system has failed.
No they don't. If a node has failed to the point it won't boot the node gets pulled and possibly diagnosed offline. Once a system fails, diagnosing is about whether the OS needs to be fixed or hardware needs to fixed. In terms of the PaaS it doesn't really matter much as to why.
Now ask yourself, they clearly know how to speak syslogd's language, and clearly journald can't actually replace syslogd in the scenarios they claim excellence in, so why not speak syslogd all around and make journald a separate and optional component?
I don't agree it can't replace syslogd. Nor do I believe that's where they claim excellence. Where they claim excellence is where a process has failed not where a system has failed. They don't make journald a separate component is they don't want their journal held back by using a 30 year old text based logger. syslogd is a bad way to do logs. It is not remotely sophisticated enough for process management. If there were a good reason to send journal to a remote system they would (and to some extent do) do that.
Part of systemd is they want you to move off the old paradigm for critical servers where jobs are tied to individual nodes. It is a paradigm change. You like to talk about force all the time. Initd was forcing people to process management the single box (Unix / NT) way. Lots of people (in particular developers) think that way sucks and instead are advocating for the way mainframes have done it. So they are introducing the capability to do it mainframe style. They are introducing an option that's all.
You keep asking "if all they are trying to do is replace initd why are they replacing all this other stuff" and the answer is always the same. The goal is not just to replace initd but rather to introduce process management, initd being just one part of that.
The commercial Unixes failed because the sellers demanded too much money for them and failed to innovate.
Too much money is too much money relative to Linux and Windows NT. They were cheap relative to minis and mainframes. That's Linux (and NT) forcing the kind of disruption you are complaining about with systemd.
As for failing to innovate you can't see the irony here in your defense of initd?
Linux didn't come up with a dirty deal to make gcc depend on the Linux kernel and make X depend on gcc in order to make it too much work to maintain solaris without the linux kernel.
I said CC. Linux pushed tremendous resources into GCC which then became good enough that it displaced CC even on Solaris.
As for Gnome1, it didn't do any such thing. Gnome 1 was reasonably interoperable. Gnome apps worked fine in other desktop environments.
I didn't say it didn't work on other desktop environments. What I said was it tried to create an e
Now for small business at endpoints they offer cheap packages but all the carriers do that. The small business products are mostly offering bandwidth down plus phone (i.e. they look like residential from a network standpoint). Maybe they won't freak about small business servers but they certainly would if the business were doing anything remotely like 1gbs regularly. They charge a lot more for that.
They aren't doing anything different than any other residential carrier. It is not $80 to distribute 1gb/1gb throughout the USA or just about every service in country would be running from their network with rates that cheap.
In major data centers, download is used very little, to the point of being free.
Where are you getting that from? More importantly how is that relevant? Major carrier based data centers are about every 100km along all the major fiber routes. Even if traffic were one direction, which it isn't, from the standpoint of the national map it would still be very expensive to distribute between them.
Upload to customers is what is costly, and since different channels are used for upload and download,
Different channels are not used for upload and download in any dedicated sort of way.
But what even if your idea were true which it isn't at all, that still wouldn't change the fact that you running a server at an endpoint would be "upload" from their standpoint.
What you're proposing is eliminating these cost signals that help allocate capital, and keep usage of it efficient and allocated to the most urgently demanded uses.
No I'm not. I'm proposing that people deal with the reality of how little of the cost What I'm proposing is that you stop pretending that your $50 / mo is covering the cost of national access rather than realizing you are offsetting local costs and that's it. The reason you get national internet is because someone else is paying for you. to consume their fully paid for bandwidth. Which is precisely what the current pricing policies do reflect.
An Apple product meant for OSX doesn't work perfectly with linux?
Linux used to work well on OSX laptops. The one previous to this and the one before that ran just about any distribution fine. Before that there was Yellow Dog which was specifically designed for Apple products, RedHat's PPC distribution, Gentoo PPC (surprisingly good)... Things have gotten worse. The point was you were claiming that power management wasn't a problem. That everything was fine I was saying that I happen to be using one of the most common x86 products possible and its very far from power management working.
As for how is systemd likely to fix this. It isn't. But if the other problems were to be fixed it is likely to fix power management.
If you think your laptop is challenged now, I can only imagine how fast it would crap out if GTK required gimp (note, GTK = Gimp Tool Kit, Gnome came later) I can just see it now, you wait for half an hour while a paintbrush renders the desktop by painting it with a brush.
What is interesting is that the FTC actually finally seems to be addressing the predominant industry FRAUD centering around terms like "unlimited data/bandwidth".
I agree. The ISPs are the ones who IMHO created this confusion by mislabeling the services they are providing.
The ISPs should not get to shape the goings on of the internet like that.
Again. Consumers should mostly be consuming bandwidth from people paying full fare. Business accounts can generate bits. Other than that basic asymmetry they really aren't aiming to shape too much.
What you ought to be able to have is the same $70/mo connection your neighbor has, and the ability to use your share of network resources, however you damn well please ('legally' bla bla). Maybe the server you run will spend 99% of its traffic only with neighbors in your apartment complex, thereby not impacting at all anybody elses service. Or maybe your quake3 server will use less bandwidth than that rich asshole neighbor with their 4k HDTV jacked into GoogleHangouts.
Well certainly if somehow you were making sure you aren't generating much traffic that is non local with your server then it doesn't really matte in terms of price. In general though most people are using LAN not WAN technology to achieve local services. There just isn't that much demand for local WAN. If there were something like $.03 / gb would be a fair charge for pushing data purely locally. Your friend using HDTV on GoogleHangouts is having his bandwidth paid for by Google so it isn't really relevant.
Slashdot Mirror
Mature software is almost always vastly less buggy than newer feature rich software. In any cycle of improvement the less buggy software is replaced with more feature rich software.
I don't know that. I think there is a some change resistance in Linux now that didn't used to exist. The Unix world used to love change. I think it is a generational shift since the early 2000s. But the Android user base which is the vast majority of the *ix world seems pretty happy with the changes. As do iOS and OSX users. And frankly most Linux desktop users like systemd. And frankly most server people are using cloud solutions which either have or will shortly be switching to systemd easily to take advantage of those features.
There is a vocal minority what doesn't like this change.
Of course it does solve the problem which means you are just making stuff up.
Can you explain what is he quitting over? (Briefly if you want) I'm having trouble separating the issues here and getting the context.
That's well put and fairly characterizes the choices that Debian is going to be faced with going forward which is rare in the systemd debate. Agree 100% and well done.
That being said the bug he seems to be upset about is automatically switching old systems on upgrade.
I'm with you. I bought it for $99 and my daughter absolutely loves it, uses it all the time. I've used it more than enough to be worth the money, as has my wife though the split is like 90/5/5%
He most does support the 4 freedoms) He supports the existence of the kernel forks. So obviously the ability to modify and redistribute. His product is open source so the ability to change.There are no purpose restrictions nor have they ever been discussed.
He disagrees with RMS when it comes to the embedded space. He had strong disagreements with RMS on an architectural level about the directions of the GNU project which at this point even RMS has abandoned. In open source vs. free software and how best to relate to companies and commercial software entities i.e. political strategy they disagreed.
I think it is best to say the almost completely agreed on ends and had some strategic and tactical disputes on how to achieve those ends.
___
And no he does not think the Linux kernel should be the only kernel. He knows kernels too well. There are big advantages to some of the other kernels out there. What he wants is a free software ecosystem same as RMS.
The USA did not spend billions on any particular party. And in fact the Chinese do spend money working the economy of Canada. They do trade quite a bit more than $5b.
As for the rest i'm going with mainstream history: http://en.wikipedia.org/wiki/2...
What Nuland said was, "Since the declaration of Ukrainian independence in 1991, the United States supported the Ukrainians in the development of democratic institutions and skills in promoting civil society and a good form of government - all that is necessary to achieve the objectives of Ukraine’s European. We have invested more than 5 billion dollars to help Ukraine to achieve these and other goals.” That's not sponsoring a junta that's working with a new quasi-European country over a quarter century.
That's not sponsorship that's friendship. I agree the USA is friendly towards Poroshenko. I'm disagreeing with sponsorship. The USA disagrees there was coup. They saw it as the population and the legislative branch replacing an executive. Turchynov became president. We have the same kind of mechanism in America if our president were to lose control. Arguably that's fairly close to how Gerald Ford (who had been minority leader) became president. Which incidentally was also a replacement that we immediately granted legitimacy to.
What are you talking about? There was a disagreement about policy in Ukraine. There were mass protests. The president began slaughtering his own population. He was replaced by the legislature. That's what happened. We aren't the ones who talked Yanukovych into massacring his population and whipping up a popular revolt. That was the Ukrainians.
Saying that systemd is the right direction for mainstream Linux is a much weaker statement than saying it is the right solution for every use case of a computer on the planet. It might be that weather stations should use something else possibly not even Linux.
That being said...
a) Systemd exports plaintext logs. If you want that you can have it. Better though would be to have systemd just export to you regularly. Your system should be queuing the weatherstation directly.
b) If roadside weather stations are that complex to fix, having redundant equipment sounds like a good idea. There should be more than one system onsite including potentially a diagnostics box.
c) Certainly having a standard repair kit distributed to your service agency would be paid for by even one of those 600km trips. Give them a USB or DVD loaded with the appropriate software that boots their systems. Smarthand job is to get stuff connected. He shouldn't be doing anything in terms of looking IMHO.
No that's not why. I've already explained to you why. You can continue to live in fantasy world or you can look at how everyone, including Amazon, is charged.
Employees in companies all the time have ideas. RedHat sponsored the project and put their institutional weight behind it. The fact that in 2010 Poetering thought of it as a personal project not withstanding.
How is the USA sponsoring the new Ukrainian government? Friendly sure. Sponsoring? As for whether it is reasonable or unreasonable the question is what the criticism was in the USA media. The GP was misstating facts.
Mexico is independent. As for Texas it has a democratic movement. RT operates freely in the United States.
When has our government every said #1? I don't even know what you are talking about with respect to #2. And as for #3, that's Russian propaganda. I can't think of anyone talking about a Russian attack on America.
You are just fabricating stuff.
I think you should drop the dollar signs nonsense. And yes I've heard of the World Trade Organization the IMF and World Bank. They aren't anti-Russian Russia is a member of these organizations.
And membership in those organizations is voluntary. There are plenty of countries that are not members of any of them and plenty that have not taken loans or acted on their advice. The USA and Russia both being examples that have policies that disagree with the all 3.
What does you not likely the TPP (which I assume your government signed) have to do with the US media being anti-Russian. If you don't like your government's trade policies then work to change them. The TPP isn't something forced on anyone.
What hostility to Russia? The American media is mostly indifferent to Russia. When they aren't indifferent they either tend to be rather laudatory about the culture: ice skating, gymnastics, art... or hostile towards Russia being uncooperative with the USA.
Now the Russian media is genuinely hostile and paranoid.
It can't "get out of the way" part of process management is starting processes. It either does it or it doesn't do it. I get that process management can be done with init as a lower level. systemd takes a different approach where process initiation just becomes another aspect of process management.
Not at all. The Requires and After keywords in the systemd system replace runlevels. There really is no need for runlevels with systemd at all. The whole concept is legacy. Distributions probably still make use of it mostly for compatibility.
Hold on. journalctl --output=export sends remote log. It isn't a full featured remote logging system but most certainly it does do what syslogd does.
Moreover, you are contradicting yourself a bit. You basically arguing that system X is inadequate if there exists any feature what-so-ever it doesn't do. At the same time you are arguing that systems should have minimal feature sets. Systemd already offers way more features than init and the entire family. Systemd has different priorities and focuses. It is a paradigm shift. When it introduces full featured remote logging it is going to be tied to something much more sophisticated than syslogd, and that is future. Today though it is perfectly capable to spewing a stream on events over the network for a listener to pickup.
What is your entire charge about forcing people to do stuff other than forward looking statements? Again you are trying to have this both ways. If you want to make likely estimates about the future that's fine. If the only thing that counts is what currently has happened on November 3, 2014 then none of the forward looking statements about possible future problems apply.
What does advocating for a better system mean in practice? Who is writing it? Where is this work being done? What is the project plan? When is it going to be completed? Who is paying? If 5 years from now a better system that systemd shows up then everyone can switch over to that. Systemd doesn't block the Unix community from moving to a better process management system or abandoning the idea of process management. If anyone wants to keep working on OpenRC no one is stopping them. But as of today, they are way behind and falling further behind. Unless you (or Ian) have some plausible plan then advocating doesn't mean anything.
I get that the anti-systemd people in some vague sense would tolerate better like your example of how xinitd allowed lots of servers without memory management. But mostly they are hostile to the paradigm shift. They object to better functionality on desktop. They object to better functionality for PaaS. That's the two uses cases they are objecting to.
Except that it doesn't do well enough today. I gave you a pretty simple chain and init couldn't come close to handling it. Init does nothing well. It is absolutely minimal process starting system. A whole range of hacks and problems exist because init sucks at its job. Going back decades, cron only exists because of deficiencies in init. The reason programs are switching to systemd is because they unhappy with the feature set of init. That's a problem today.
And even what it claims to be good at, it sucks at. On most systems falling back to lower runlevels doesn't actually work. I've tried taking a system
No of course I wouldn't want the app to load. But I would want gtk to have features from Gimp so that other applications can use chunks of Gimp code to accomplish different tasks. This is the way Cocoa works. For example webkit is used by a 1/2 dozen applications on OSX and iOS to accomplish web features even if you aren't using Safari.
You have upload and download reversed it is to and from them not to and from your local systems. But ignoring the terminology reversal, they don't charge you for the bandwidth when you are putting stuff on their server because they are going to collect storage fees.
No they charge those particular prices because in one direction they make lots of money on storage and in the other they either make nothing or are going to lose their storage fees.
The poster is arguing the AWS reflects on the cost of the pipe.
I don't see why systemd would make a hash of my hardware rather than be better. What I would do is use a distribution that didn't depend on systemd. Linux is a diverse ecosystem of software. Or if that didn't work, pick between Linux and the hardware. Like I said I currently have hardware Linux doesn't work on. I'm not blaming initd for that. But the power management problems I and everyone has had before well yeah that was initd's fault. Initd didn't offer enough services to do power management well easily. Windows did a great job because of the how deeply embedded powerstate was in Windows. Integration helped. OSX BTW lagged for similar reasons though with they have in the last few years pulled way ahead of Windows.
No that's not the issue. I suspect if I were still using the older hardware the more complex distributions would boot. It is the more complex hardware (EFI, dual GPU, complex resolution and sizing, lots of inputs on trackpad...) that is throwing the the newer distributions. The Linux community used to work hard on getting all x86 hardware to work. There is a push away from desktop Linux and one of the falloffs is that Linux just isn't working as hard on hassle free installs.
Because that's not the form it would take. The form it would take would be is what I described gtk depending on gimp libraries. Which would mean that all gtk applications would get a rich collection of routines from gimp that could be easily integrated into gtk applications. Which is very similar to what does exist with Cocoa and the Quartz graphical libraries. Having the widget set have additional features is a good thing about Cocoa not a bad thing.
Amazon is allowing you to upload free because when you upload you are generally going to be buying storage. That doesn't reflect Amazon's underlying network costs. They don't have the same cost structures as carriers because AWS offers far more services. So, no it does not cost Amazon 0 for you to upload data. They offset the cost. When they aren't offsetting the cost they charge between $30-120 / tb to move data out. Which is being sold at a reasonable profit but is more reflective of what the actual cost of bandwidth is.
Your ISP can similarly offset home costs by making money on selling for example television.
I think you should go back read about supply and demand. The supply curve is a curve which indicates how supply increases as cost increases. The whole point of the intersection point being on that curve is that price is not independent of cost of providing the service. In particular, carriers are not going to give you services for far less than they cost to deliver.
I didn't claim that. What I claimed was that power management properly belongs with process management because of the advantages it could offer. You then claimed that systemd didn't implement power management and I was showing that it most certainly did. Now on top of all that, it is already better in terms of success it already works better than it did with init.
No. Changing runlevels is shutting process down and breaking potential dependencies. That isn't suspending or hibernating them. Telling a process to shutdown is different than telling a process to wrap up immediate problems possibly save data and sleep. Or being able to know the state of the process to evaluate when it is safe to swap it out.
No they don't. If a node has failed to the point it won't boot the node gets pulled and possibly diagnosed offline. Once a system fails, diagnosing is about whether the OS needs to be fixed or hardware needs to fixed. In terms of the PaaS it doesn't really matter much as to why.
I don't agree it can't replace syslogd. Nor do I believe that's where they claim excellence. Where they claim excellence is where a process has failed not where a system has failed. They don't make journald a separate component is they don't want their journal held back by using a 30 year old text based logger. syslogd is a bad way to do logs. It is not remotely sophisticated enough for process management. If there were a good reason to send journal to a remote system they would (and to some extent do) do that.
Part of systemd is they want you to move off the old paradigm for critical servers where jobs are tied to individual nodes. It is a paradigm change. You like to talk about force all the time. Initd was forcing people to process management the single box (Unix / NT) way. Lots of people (in particular developers) think that way sucks and instead are advocating for the way mainframes have done it. So they are introducing the capability to do it mainframe style. They are introducing an option that's all.
You keep asking "if all they are trying to do is replace initd why are they replacing all this other stuff" and the answer is always the same. The goal is not just to replace initd but rather to introduce process management, initd being just one part of that.
Too much money is too much money relative to Linux and Windows NT. They were cheap relative to minis and mainframes. That's Linux (and NT) forcing the kind of disruption you are complaining about with systemd.
As for failing to innovate you can't see the irony here in your defense of initd?
I said CC. Linux pushed tremendous resources into GCC which then became good enough that it displaced CC even on Solaris.
I didn't say it didn't work on other desktop environments. What I said was it tried to create an e
Bengie take a look at what they charge for real business lines: http://www.sonic.net/solutions...
They also want to distribute via. colo not at endpoints: http://www.sonic.net/colocatio...
Now for small business at endpoints they offer cheap packages but all the carriers do that. The small business products are mostly offering bandwidth down plus phone (i.e. they look like residential from a network standpoint). Maybe they won't freak about small business servers but they certainly would if the business were doing anything remotely like 1gbs regularly. They charge a lot more for that.
They aren't doing anything different than any other residential carrier. It is not $80 to distribute 1gb/1gb throughout the USA or just about every service in country would be running from their network with rates that cheap.
Where are you getting that from? More importantly how is that relevant? Major carrier based data centers are about every 100km along all the major fiber routes. Even if traffic were one direction, which it isn't, from the standpoint of the national map it would still be very expensive to distribute between them.
Different channels are not used for upload and download in any dedicated sort of way.
But what even if your idea were true which it isn't at all, that still wouldn't change the fact that you running a server at an endpoint would be "upload" from their standpoint.
No I'm not. I'm proposing that people deal with the reality of how little of the cost What I'm proposing is that you stop pretending that your $50 / mo is covering the cost of national access rather than realizing you are offsetting local costs and that's it. The reason you get national internet is because someone else is paying for you. to consume their fully paid for bandwidth. Which is precisely what the current pricing policies do reflect.
Linux used to work well on OSX laptops. The one previous to this and the one before that ran just about any distribution fine. Before that there was Yellow Dog which was specifically designed for Apple products, RedHat's PPC distribution, Gentoo PPC (surprisingly good) ... Things have gotten worse. The point was you were claiming that power management wasn't a problem. That everything was fine I was saying that I happen to be using one of the most common x86 products possible and its very far from power management working.
As for how is systemd likely to fix this. It isn't. But if the other problems were to be fixed it is likely to fix power management.
That's just silly.
I agree. The ISPs are the ones who IMHO created this confusion by mislabeling the services they are providing.
Again. Consumers should mostly be consuming bandwidth from people paying full fare. Business accounts can generate bits. Other than that basic asymmetry they really aren't aiming to shape too much.
Well certainly if somehow you were making sure you aren't generating much traffic that is non local with your server then it doesn't really matte in terms of price. In general though most people are using LAN not WAN technology to achieve local services. There just isn't that much demand for local WAN. If there were something like $.03 / gb would be a fair charge for pushing data purely locally. Your friend using HDTV on GoogleHangouts is having his bandwidth paid for by Google so it isn't really relevant.