Uh, you do realize that text-based logs suffer from the same problem, right? You just ignore it when you're reading it with your eyeballs, which is basically what you do when you look at the output of journalctl with your eyeballs.
I would have it quit corrupting log files. Either use a decent transaction mechanism or go back to appending text files where bits of corruption can generally be dealt with sanely.
How is journald NOT dealing with the corruption sanely? It just starts a new file. A text-based system would probably do the same thing, or insert a newline and keep going (basically just hiding the problem until your log parser wonders why there is malformatted line in the file).
No syslog implementation I'm aware of uses a transaction mechanism. That would basically require a journal of some kind and lots of fsyncing since Linux itself doesn't really support transactions on files.
If you're going to take the line that "udev is systemd" then sure, it takes more work to avoid it. Of course, Gentoo is about the only distro that even makes that an option on Linux.
Honestly, eudev isn't really all that different from udev. The build system is different, and it has some preferences codified as defaults like not having the new network names. I don't really get the excitement, but whatever floats your boat - Gentoo is used in a bazillion one-off configs which is basically its raison d'être...:)
It does feel like we get a lot of anti-systemd types on the lists, and perhaps that is because they don't have too many other places to go. The trolling really isn't welcome either way, but it is a pain to deal with.
I do think that singling out Gentoo is a bit unfair. Really, Gentoo is about choice and diversity, so we end up with users who don't want to use anything GPL, users who don't want to run glibc, users who run prefix on OSX, and so on. Basically you end up with a lot of non-mainstream users, because we're about the only show in town which makes being non-mainstream a practical option.
They're talking about big iron here. There are two ways to generally go about this stuff.
The new way is what you're describing - build your application such that it is distributed and nodes can go down at will without serious failure, and the hardware is commodity hardware.
The old way is big iron - the software is very simple in design and does not tolerate faults at all, and the hardware ensures that there never are any faults. That means super-redundant everything. IBM's z-series mainframes can actually run redundant CPUs with a watchdog monitoring their execution for discrepancies - if they disagree then the failing CPU is identified and dealt with. Think of ECC or RAID taken to the level of insanity. You could probably carefully smash virtually any single component in one of those things and have your software not miss a beat.
The big iron way means hot-pluggability. You don't want to shut down your software just because one of the 47 CPUs failed.
These things are practically rented with their support models. Often they sell you a big machine and you pay for the parts that you use - it might have 50 CPUs that allow for full redundancy, but you pay to unlock 2 of them without CPU redundancy, etc. If you pay for the works then your software will basically never go down. If something goes wrong the machine will phone home, somebody will show up and hot-swap the bad part, and not a single instruction will be mis-executed.
The reason big iron is going out of favor is that it is REALLY expensive. It makes more sense to just design the software to be fault-tolerant, and then you can often have geographic failovers as well.
Speaking of Slackware, it's situation is telling. No desire whatsoever to switch to systemd but with caveat that it may one day become impractical to continue resting. That sure sounds like there is a coercive element to all of this.
Or, maybe it just reflects that linux distros just repackage the stuff that everybody else is already writing, and that forking every other major project is too much work for a small band of volunteers?
If I write a simple FOSS kernel in my spare time that isn't POSIX-compatible, there is nothing wrong with that. If I write some software for it, or encourage a few friends to do so there is nothing wrong with that. On the other hand, if I then go on slashdot and moan and complain about how the fact that the folks at mozilla, apache, and mariadb haven't ported their massive codebases over to my new kernel, then my expectations are clearly off the rocker.
Systemd is just the same thing in reverse. Somebody writes that new non-POSIX kernel and then mozilla, apache, and mariadb announce that they've seen the light and will no longer support POSIX environments, and you're frustrated that everybody is abandoning linux as a result.
What is actually broken? I'm not aware of any syslog implementations that "fix" a log file if a line gets cut-off mid-write by a power outage.
Journald just rotates to a new file and starts writing a clean log. The old log is readable up until the last valid record I believe.
Sure, I guess you could have it delete the last record after-the-fact and then cleanly close the file, but that won't really "fix" anything. The resulting behavior would be the same. If anything by preserving the damaged file you have the opportunity to go poking inside to see if there is anything useful, while the log continues on in a new file after reboot.
He has foisted systemd on a bunch of people who don't want it. Now he expects those same people to send him flowers?
I know it is crazy. Last night I kissed good-night my Linux From Scratch (TM) box that is stripped of udev, bash, glibc, pulseaudio, and all that other crazy stuff. In truth even sysvinit and bash are too fluffy for me - I just build a minimal kernel with init=/bin/csh on the built-in command line and I'm still bummed that they stripped the bootloader code out of the kernel 15 years ago and I have to install bloatware like LILO or else it just tells me to bugger_off.
Then I went to bed and in the middle of the night a ninja hired by Lennart came in and wiped my hard drive and stole all my backups, and installed Arch running systemd and bloatware like agetty. I can't even figure out how to shut the thing down - killing my shell just logs me out and doesn't cause a panic like it is supposed to!
That and polarized topics are such that someone can paint even a moderate view as extremist. I'm anti-abortion and pro-choice. They are orthogonal. Some people think that abortion is a valid contraception method, to be used casually (see the abortion rates and stats from the Soviet Union). Others would wish that nobody needed one, but so long as they are needed, they should be legal. They are pro-choice and anti-abortion. The "pro-life" group is anti-choice. Anti-abortion could fit a description of people from both sides.
I'll agree that pro-abortion and pro-choice are somewhat orthogonal, but the reason that people treat them as not being so is that their real assertion is that abortion is murder, and they just assume that you can't be pro-choice about murder. That is completely untrue, of course, pro-murder and pro-choice are also orthogonal in the same sense.
Let me rewrite your paragraph here and illustrate the challenge: That and polarized topics are such that someone can paint even a moderate view as extremist. I'm anti-murder and pro-choice. They are orthogonal. Some people think that murder is a valid dispute resolution method, to be used casually (see the murder rates and stats from the USA). Others would wish that nobody needed to kill anybody, but so long as it is necessary, it should be legal. They are pro-choice and anti-murder. The "pro-life" group is anti-choice. Anti-murder could fit a description of people from both sides.
Now, the reality is that almost nobody would self-identify with that statement, but it is a completely self-consistent position to have. I could very well think that murder is something to be avoided, but that it should be legal and up to every individual to decide whether to kill somebody over a disagreement.
I fully appreciate that almost everybody who is pro-choice does not consider abortion to be morally equivalent to murder, and so they will object to the analogy. However, I suspect that most who are pro-life DO consider abortion to be morally equivalent to murder, and so you shouldn't be shocked when they take objection to this simply being a matter of preference.
In my mind the more interesting moral question is whether abortion is murder. Once you make a decision on that one way or another, whether it should be a matter of choice would probably be fairly straightforward in most minds (but then again, considering that gay marriage is still illegal in 20 states and marijuana use in almost all of them, perhaps not).
Redhat and Poettering have some really, really big sticks: mostly a bank account to pay developers to have things and an existing code base/market. If you're debian, gentoo or any distro, forking half of Gnome3, etc, etc just to deal with this stuff is one hell of a tall ask considering how they are comprised and/or not funded.
I think most of the issue really boils down to this (that is, the code issues and not the social ones - which I'm not discounting). SystemD couldn't just "absorb" udev if there were an army of people already maintaining udev. The same is true of just about everything else.
An issue with FOSS is that people tend to work on the fun stuff. That is why you end up with two guys trying to maintain things like openssl or bash - the fun part ended 15 years ago. With corporate money you can make the boring stuff happen, and then the guy with the money can unilaterally make all the decisions.
The interesting thing is that the FOSS community often flocks to the new stuff all the same, because something new is almost always more fun than something that is in bugfix mode.
The other thing is that when you look at something like systemd, what is it even competing against? Look at the featureset of sysvinit plus a hodge-podge of bash scripts, and then look at the planned featureset of systemd. Which would you rather contribute to or work with?
I think the biggest issue with something like systemd is that it tends to get rolled out quickly, and there are growing pains as a result. However, distros don't have much choice - their devs want to work on what is new and flashy, and they actually prefer the buggy new system with more features than the stable old system. The bar on the customer side keeps raising too - people don't just want to boot a system and launch a few daemons - they want to be hosting containers, or running in containers, and they want to be able to plug in a USB network adapter and just have it do the right thing, etc.
Demanding other people to repeat easily searched reasons is your problem, not other people's.
Actually, it is everybody's problem. The issue is that technical arguments get made once, and insults get thrown around all day long.
The original poster made comments about excrement. Then when asked for an actual argument they said, "well, google it."
Why not just make a post that says "I have an unoriginal comment to add - type systemd into google and see what comes up." It would be about as helpful to the discussion.
If you'd like to know how I feel about you, just type your own name into Google and see what comes up.:)
SystemD is an abortion. It appeals to RedHat - who stacked the deck and manipulated the governing process to have it adopted by Debian. If they want an OS built like that?
Uh, you do realize that virtually every distro around is in the process of switching over to systemd, right? About the only one with no real movement towards doing so that I can see is Slackware. Did RedHat stack the deck and manipulate the governing process on all of them? Or, maybe do they just all want an OS built like that?
It's pretty simple. There's supposed to be a 2 second gap in good conditions. It's written into the handbook here in MA for driver's ed, although I believe it's been raised to 3-6-12 instead of 2-4-10.
I think the goal of the tailgater is to influence the driver ahead of them to change their behavior, by genuinely putting their life at risk (if they have to stop quickly, they'll be rammed and then they get to deal with that in addition to whatever they were trying to avoid ahead of them). Leaving a 2 second gap defeats that purpose, thus ensuring that the two guys driving 30 in a 50 zone side-by-side continue to do so.
I'm not saying that it is legal or ought to be legal. That is the rationale, however.
This is true, of course, but it should also be remembered (as Elenkov explains in detail) that Android's encryption features aren't new, they exist in hundreds of millions of devices already deployed.
I hope that Google plans to actually do better. Android does not support encryption unless you have a screen lock password turned on, and it does not support having the encryption key set to something other than the screen lock password.
Device encryption and screen locking are two different solutions to two different, but related, problems. It can make perfect sense to enable one without the other, and it makes almost no sense to use the same password for both.
The result is that most people don't bother enabling encryption at all, or if they do they use a key which is so weak it is useless.
Google needs to make both settings independently settable, with independent codes. By all means allow the user to sync them, but do not force this.
this is apparently the first brain function to go with the onset of Alzheimer's and may lead to greater understanding.
It's my understand that it's protein plaque forming between neurons, thus interfering with communications between the cells. As such, having information constantly being re-routed as the disease progresses makes sense that orientation would be disrupted. Would it not?
I think that is a definite maybe. I'd think it would depend on how this functionality actually works. If the brain re-wires its connections, then from one standpoint you'd think the new neurons could just take the place of the old ones. On the other hand, if the brain actually implemented 3D problem-solving by actually running the problems in a physical model of a 3D world that could break down. For example, if figuring out the fastest way from A to B involved sending signals out through a network and seeing what path leads to the shortest arrival time, then it is critical that your model have identical connectivity to the real world. If you have a bad neuron somewhere in the middle then direct transit across it will actually travel a longer physical distance around it than necessary. On the other hand, if actual latency/etc doesn't matter then you'd think that you could rewire neurons the way you move VMs around in a datacenter.
Countries that don't want to deal with this can just choose not to admit US warships into their ports.
I think by your description, that would be everyone.
I'd be shocked if it turned out that way. Why do these countries even want US warships in their ports in the first place? Those reasons aren't going to go away just because there is a risk a few local fishermen could get shot up. They don't exactly contribute heavily to political campaigns in democratic nations, and they aren't likely to be likely to related to the local despot elsewhere.
This is like suggesting that the result of all these NSA revelations is that countries are going to take steps to keep the NSA out. I'm sure 99% of these countries already knew what the NSA was up to already, and likely invited them in the first place. The outrage is just a show for the voters.
Do you have a missile system that can fire about a thousand or so missiles?
From my post: "You just need a few thousand of them manufactured for $500 apiece." The whole point is to fight swarms of cheap bots with swarms of cheap missiles. I don't see why this wouldn't be practical. What necessarily makes a missile expensive? It is a small warhead (cheap), a solid fuel rocket engine (cheap), a sensor of some kind (probably cheap, depending on design), and a computer (cheap). The cost is in the design/etc, but that gets cheaper the more you make.
And unless the state has specifically permitted this by legislation or given you an ironclad immunity deal (let's face it, not many countries are going to be happy to give foreign navies the right to point large guns at their civilians without recourse) such actions would be highly inadvisable. If you point a gun at another vessel and demand it cut its engines while in port you better be able to make a very strong case that it was the threat otherwise you may be spending a lot longer on shore leave than anticipated. In my own jurisdiction those actions would likely have you on the hook for assault and false imprisonment
The ship wouldn't be in port without the permission of the local government. The guy opening fire would only due so under the chain of command, and the local police wouldn't be able to do anything about it since everybody issuing orders would be in the middle of a US naval vessel, if they're even in the same country. The officers who made the call would report back to their superiors.
I'm sure the US navy makes arrangements with the governments where it sends its ships regarding how crisis situations get handled. Most likely the local government will express suitable outrage on TV and then reassure their diplomatic counterparts that everything is fine.
Another likely scenario is that the US spends billions on these boats but doesn't prevent anybody from approaching their vessels to avoid an incident, making the investment in these vessels worthless other than as pork.
Think about it. You're a peaceful merchant, or just a guy going sailing. A fast-attack boat sails up to you, points a 50 cal at you, and shouts on the bullhorn "cut off your engine immediately and put your hands in the air." What are you likely to do? You're going to stick your hands in the air, and thus you aren't likely to get shot. The guy controlling the drone isn't worried about never seeing his family again, so he isn't going to have a twitchy trigger finger. Then you sort the mess out and get an apology.
And this never goes wrong, right? You don't get a pilot who doesn't speak English or is deaf. Maybe the rudder broke and they can't cut the engine. Or maybe they're having a heart attack right now and are five minutes from death. Or maybe they're a terrorist faking one of the above and just trying to get their boat in close before they blow it up.
Sure it will go wrong, but never in a way that results in the US ship being sunk, which is the point. Even if the guy on the boat is deaf, after enough deaf people get shot up in their fishing boats the local deaf fishing community will figure out how to go fishing without getting shot.
The US doesn't build stuff like this for good community relations - they do it to protect their ships.
And taking out Beijing would only cause them to take out Los Angeles.
My point was that the Chinese don't have the ability to take out Los Angeles without the use of ICBMs. If they launch ICBMs at LA, then the US will launch land-based ICBMS at every Chinese military target before the Chinese ones are even at midcourse, and then everybody gets to find out what was in the Chinese warheads while the US nuclear warheads are halfway to China. If the Chinese didn't launch their nuclear arsenal in the first strike, they won't get a second one. The US would save the SLBMs for whatever comes next. Nobody plays games with ICBMs because the game theory basically boils down to "when one flies, they all fly." Again, fantasy hypotheses - the Chinese would never launch an ICBM against the US.
US attacks on the Chinese mainland would be launched by subs - a capability the Chinese basically don't have right now, and which is much harder to counter than sinking an aircraft carrier since subs rely on stealth to a much greater degree.
Also, carriers aren't quite as vulnerable as people think after a war starts. You can't target a carrier with a missile unless you have a sense of where it is. We take that for granted because satellite surveillance is ubiquitous. However, once a real war starts, the first thing everybody will do is shoot down everybody else's satellites and as a result the human race won't be able to put anything in orbit for a few centuries due to debris. Finding a naval task force at sea without the use of satellites is a much harder problem, and one which the US has a major advantage at due to its large navy.
This is basically a password that cannot be changed. Just like fingerprints and retina scans. And all these things can be faked so easily. You don't bother with a false fingerprint or "a practiced mouse grip pattern". No, you replace the device (mouse or fingerprint reader) and fake the digital output from the authentication device. This is much easier.
This is a fundamental weakness of biometrics, but that doesn't make biometrics useless. If the computer is in a supervised area swapping out a mouse isn't going to be trivial, especially if the correct mouse authenticates itself in some way.
When I see stuff like this I think "government installation designed to thwart ninja sneaking in the air ducts and changing into a uniform."
The Persian gulf. The issue there is that lots of tiny boats could swarm US ships and destroy them. This is an Iranian plan by the way. To use the proximity of the US to the shore and just hundreds of little boats. The drone swarm idea appears to be a counter strategy.
I think that ships like this would be more useful for investigating approaching traffic and warning it off so that fishing boats and such don't get accidentally destroyed.
If you wanted to deal with swarms of attack craft I think this is only perhaps useful as a first step. The problem is that these things are just armed with machine guns, which is plenty to sink a small craft at close range, but if you send 200 speed boats against an aircraft carrier and one is carrying an antiship missile that is lethal at 20 miles, then I don't think you can count on stopping that attack with a bunch of drones with short-range weapons.
I think the better option is a medium-range missile system that is inexpensive to deploy in quantity. You can use the fancy Harpoons and such for the fancy warships that have armor and air defenses and all that (and which are so expensive that the enemy only owns a handful). When the enemy sends in 200 zodiacs then you basically just need some cheap slow rockets that fire off in a ballistic trajectory, glide in to a designated target (radar or maybe laser, or maybe even optical), and set off a warhead not much bigger than a hand grenade. You just need a few thousand of them manufactured for $500 apiece. They're not going to do more than scratch the paint on a real warship and maybe a well-trained soldier with a shotgun could shoot them out of the sky, but the enemy won't be able to field hundreds of attack boats capable of defending against a system like this.
Now, you could very well use boats like this as a launch platform for these cheap anti-ship rockets, to further reduce their range/cost. It wouldn't be too hard to station a cheap defense drone every mile along a perimeter outside of the enemy's effective missile range.
The part where you have autonomous boats running around in a busy harbor blocking and possibly destroying other boats.
Think about it. You're a peaceful merchant, or just a guy going sailing. A fast-attack boat sails up to you, points a 50 cal at you, and shouts on the bullhorn "cut off your engine immediately and put your hands in the air." What are you likely to do? You're going to stick your hands in the air, and thus you aren't likely to get shot. The guy controlling the drone isn't worried about never seeing his family again, so he isn't going to have a twitchy trigger finger. Then you sort the mess out and get an apology.
The whole point is to enforce a perimeter far enough away that the folks working the guns can take their time and work things out.
Countries that don't want to deal with this can just choose not to admit US warships into their ports.
How many times have the US Ships put into port / refueled without getting hit? Just seems that it would be a way to spend a ton of money for something that overly complicates normal procedures, and only wards off that 1% of attacks. (Note, percentage pulled out of my rear).
It would ward off the 100% of successful attacks. How many attacks have US naval vessels actually come under in the last decade? These boats are probably going to save more lives than the billion-dollar antiaircraft defense systems they're carrying.
This is about pushing the perimeter away from the boat. They can challenge approaching vessels from a distance, and it gives them the option of getting in close without putting the ship at risk. I'm sure these little boats could be evaded/destroyed/jammed/etc, but at that point the intruder has demonstrated hostile intent and they'll be sunk before they can get anywhere near the real target.
Also.... any autonomous craft would surely need a remote control system. You can't stop the signal (Mal). It wouldn't be impossible for another country / faction to take control of said boats, and use them to accomplish their goal.
Hijacking a military drone is not a trivial matter. All communications will almost certainly be encrypted and authenticated, and tested by folks like the NSA. Jamming is a more likely possibility, but even that isn't necessarily easy, and that just alerts the ship being defended that an attack is underway.
This is about defense in depth. You keep everybody at arms length, and then anybody who tries to get close can be treated as unfriendly. It is like putting up a chain link fence on the prison property line, with a big wall further back. Anybody walking around at the base of the wall has already broken into the property and can be shot on sight.
Slashdot Mirror
Uh, you do realize that text-based logs suffer from the same problem, right? You just ignore it when you're reading it with your eyeballs, which is basically what you do when you look at the output of journalctl with your eyeballs.
I would have it quit corrupting log files. Either use a decent transaction mechanism or go back to appending text files where bits of corruption can generally be dealt with sanely.
How is journald NOT dealing with the corruption sanely? It just starts a new file. A text-based system would probably do the same thing, or insert a newline and keep going (basically just hiding the problem until your log parser wonders why there is malformatted line in the file).
No syslog implementation I'm aware of uses a transaction mechanism. That would basically require a journal of some kind and lots of fsyncing since Linux itself doesn't really support transactions on files.
If you're going to take the line that "udev is systemd" then sure, it takes more work to avoid it. Of course, Gentoo is about the only distro that even makes that an option on Linux.
Honestly, eudev isn't really all that different from udev. The build system is different, and it has some preferences codified as defaults like not having the new network names. I don't really get the excitement, but whatever floats your boat - Gentoo is used in a bazillion one-off configs which is basically its raison d'être... :)
++
It does feel like we get a lot of anti-systemd types on the lists, and perhaps that is because they don't have too many other places to go. The trolling really isn't welcome either way, but it is a pain to deal with.
I do think that singling out Gentoo is a bit unfair. Really, Gentoo is about choice and diversity, so we end up with users who don't want to use anything GPL, users who don't want to run glibc, users who run prefix on OSX, and so on. Basically you end up with a lot of non-mainstream users, because we're about the only show in town which makes being non-mainstream a practical option.
They're talking about big iron here. There are two ways to generally go about this stuff.
The new way is what you're describing - build your application such that it is distributed and nodes can go down at will without serious failure, and the hardware is commodity hardware.
The old way is big iron - the software is very simple in design and does not tolerate faults at all, and the hardware ensures that there never are any faults. That means super-redundant everything. IBM's z-series mainframes can actually run redundant CPUs with a watchdog monitoring their execution for discrepancies - if they disagree then the failing CPU is identified and dealt with. Think of ECC or RAID taken to the level of insanity. You could probably carefully smash virtually any single component in one of those things and have your software not miss a beat.
The big iron way means hot-pluggability. You don't want to shut down your software just because one of the 47 CPUs failed.
These things are practically rented with their support models. Often they sell you a big machine and you pay for the parts that you use - it might have 50 CPUs that allow for full redundancy, but you pay to unlock 2 of them without CPU redundancy, etc. If you pay for the works then your software will basically never go down. If something goes wrong the machine will phone home, somebody will show up and hot-swap the bad part, and not a single instruction will be mis-executed.
The reason big iron is going out of favor is that it is REALLY expensive. It makes more sense to just design the software to be fault-tolerant, and then you can often have geographic failovers as well.
Speaking of Slackware, it's situation is telling. No desire whatsoever to switch to systemd but with caveat that it may one day become impractical to continue resting. That sure sounds like there is a coercive element to all of this.
Or, maybe it just reflects that linux distros just repackage the stuff that everybody else is already writing, and that forking every other major project is too much work for a small band of volunteers?
If I write a simple FOSS kernel in my spare time that isn't POSIX-compatible, there is nothing wrong with that. If I write some software for it, or encourage a few friends to do so there is nothing wrong with that. On the other hand, if I then go on slashdot and moan and complain about how the fact that the folks at mozilla, apache, and mariadb haven't ported their massive codebases over to my new kernel, then my expectations are clearly off the rocker.
Systemd is just the same thing in reverse. Somebody writes that new non-POSIX kernel and then mozilla, apache, and mariadb announce that they've seen the light and will no longer support POSIX environments, and you're frustrated that everybody is abandoning linux as a result.
What is actually broken? I'm not aware of any syslog implementations that "fix" a log file if a line gets cut-off mid-write by a power outage.
Journald just rotates to a new file and starts writing a clean log. The old log is readable up until the last valid record I believe.
Sure, I guess you could have it delete the last record after-the-fact and then cleanly close the file, but that won't really "fix" anything. The resulting behavior would be the same. If anything by preserving the damaged file you have the opportunity to go poking inside to see if there is anything useful, while the log continues on in a new file after reboot.
He has foisted systemd on a bunch of people who don't want it. Now he expects those same people to send him flowers?
I know it is crazy. Last night I kissed good-night my Linux From Scratch (TM) box that is stripped of udev, bash, glibc, pulseaudio, and all that other crazy stuff. In truth even sysvinit and bash are too fluffy for me - I just build a minimal kernel with init=/bin/csh on the built-in command line and I'm still bummed that they stripped the bootloader code out of the kernel 15 years ago and I have to install bloatware like LILO or else it just tells me to bugger_off.
Then I went to bed and in the middle of the night a ninja hired by Lennart came in and wiped my hard drive and stole all my backups, and installed Arch running systemd and bloatware like agetty. I can't even figure out how to shut the thing down - killing my shell just logs me out and doesn't cause a panic like it is supposed to!
That and polarized topics are such that someone can paint even a moderate view as extremist. I'm anti-abortion and pro-choice. They are orthogonal. Some people think that abortion is a valid contraception method, to be used casually (see the abortion rates and stats from the Soviet Union). Others would wish that nobody needed one, but so long as they are needed, they should be legal. They are pro-choice and anti-abortion. The "pro-life" group is anti-choice. Anti-abortion could fit a description of people from both sides.
I'll agree that pro-abortion and pro-choice are somewhat orthogonal, but the reason that people treat them as not being so is that their real assertion is that abortion is murder, and they just assume that you can't be pro-choice about murder. That is completely untrue, of course, pro-murder and pro-choice are also orthogonal in the same sense.
Let me rewrite your paragraph here and illustrate the challenge:
That and polarized topics are such that someone can paint even a moderate view as extremist. I'm anti-murder and pro-choice. They are orthogonal. Some people think that murder is a valid dispute resolution method, to be used casually (see the murder rates and stats from the USA). Others would wish that nobody needed to kill anybody, but so long as it is necessary, it should be legal. They are pro-choice and anti-murder. The "pro-life" group is anti-choice. Anti-murder could fit a description of people from both sides.
Now, the reality is that almost nobody would self-identify with that statement, but it is a completely self-consistent position to have. I could very well think that murder is something to be avoided, but that it should be legal and up to every individual to decide whether to kill somebody over a disagreement.
I fully appreciate that almost everybody who is pro-choice does not consider abortion to be morally equivalent to murder, and so they will object to the analogy. However, I suspect that most who are pro-life DO consider abortion to be morally equivalent to murder, and so you shouldn't be shocked when they take objection to this simply being a matter of preference.
In my mind the more interesting moral question is whether abortion is murder. Once you make a decision on that one way or another, whether it should be a matter of choice would probably be fairly straightforward in most minds (but then again, considering that gay marriage is still illegal in 20 states and marijuana use in almost all of them, perhaps not).
Redhat and Poettering have some really, really big sticks: mostly a bank account to pay developers to have things and an existing code base/market. If you're debian, gentoo or any distro, forking half of Gnome3, etc, etc just to deal with this stuff is one hell of a tall ask considering how they are comprised and/or not funded.
I think most of the issue really boils down to this (that is, the code issues and not the social ones - which I'm not discounting). SystemD couldn't just "absorb" udev if there were an army of people already maintaining udev. The same is true of just about everything else.
An issue with FOSS is that people tend to work on the fun stuff. That is why you end up with two guys trying to maintain things like openssl or bash - the fun part ended 15 years ago. With corporate money you can make the boring stuff happen, and then the guy with the money can unilaterally make all the decisions.
The interesting thing is that the FOSS community often flocks to the new stuff all the same, because something new is almost always more fun than something that is in bugfix mode.
The other thing is that when you look at something like systemd, what is it even competing against? Look at the featureset of sysvinit plus a hodge-podge of bash scripts, and then look at the planned featureset of systemd. Which would you rather contribute to or work with?
I think the biggest issue with something like systemd is that it tends to get rolled out quickly, and there are growing pains as a result. However, distros don't have much choice - their devs want to work on what is new and flashy, and they actually prefer the buggy new system with more features than the stable old system. The bar on the customer side keeps raising too - people don't just want to boot a system and launch a few daemons - they want to be hosting containers, or running in containers, and they want to be able to plug in a USB network adapter and just have it do the right thing, etc.
Demanding other people to repeat easily searched reasons is your problem, not other people's.
Actually, it is everybody's problem. The issue is that technical arguments get made once, and insults get thrown around all day long.
The original poster made comments about excrement. Then when asked for an actual argument they said, "well, google it."
Why not just make a post that says "I have an unoriginal comment to add - type systemd into google and see what comes up." It would be about as helpful to the discussion.
If you'd like to know how I feel about you, just type your own name into Google and see what comes up. :)
SystemD is an abortion. It appeals to RedHat - who stacked the deck and manipulated the governing process to have it adopted by Debian. If they want an OS built like that?
Uh, you do realize that virtually every distro around is in the process of switching over to systemd, right? About the only one with no real movement towards doing so that I can see is Slackware. Did RedHat stack the deck and manipulate the governing process on all of them? Or, maybe do they just all want an OS built like that?
It's pretty simple. There's supposed to be a 2 second gap in good conditions. It's written into the handbook here in MA for driver's ed, although I believe it's been raised to 3-6-12 instead of 2-4-10.
I think the goal of the tailgater is to influence the driver ahead of them to change their behavior, by genuinely putting their life at risk (if they have to stop quickly, they'll be rammed and then they get to deal with that in addition to whatever they were trying to avoid ahead of them). Leaving a 2 second gap defeats that purpose, thus ensuring that the two guys driving 30 in a 50 zone side-by-side continue to do so.
I'm not saying that it is legal or ought to be legal. That is the rationale, however.
This is true, of course, but it should also be remembered (as Elenkov explains in detail) that Android's encryption features aren't new, they exist in hundreds of millions of devices already deployed.
I hope that Google plans to actually do better. Android does not support encryption unless you have a screen lock password turned on, and it does not support having the encryption key set to something other than the screen lock password.
Device encryption and screen locking are two different solutions to two different, but related, problems. It can make perfect sense to enable one without the other, and it makes almost no sense to use the same password for both.
The result is that most people don't bother enabling encryption at all, or if they do they use a key which is so weak it is useless.
Google needs to make both settings independently settable, with independent codes. By all means allow the user to sync them, but do not force this.
this is apparently the first brain function to go with the onset of Alzheimer's and may lead to greater understanding.
It's my understand that it's protein plaque forming between neurons, thus interfering with communications between the cells. As such, having information constantly being re-routed as the disease progresses makes sense that orientation would be disrupted. Would it not?
I think that is a definite maybe. I'd think it would depend on how this functionality actually works. If the brain re-wires its connections, then from one standpoint you'd think the new neurons could just take the place of the old ones. On the other hand, if the brain actually implemented 3D problem-solving by actually running the problems in a physical model of a 3D world that could break down. For example, if figuring out the fastest way from A to B involved sending signals out through a network and seeing what path leads to the shortest arrival time, then it is critical that your model have identical connectivity to the real world. If you have a bad neuron somewhere in the middle then direct transit across it will actually travel a longer physical distance around it than necessary. On the other hand, if actual latency/etc doesn't matter then you'd think that you could rewire neurons the way you move VMs around in a datacenter.
Countries that don't want to deal with this can just choose not to admit US warships into their ports.
I think by your description, that would be everyone.
I'd be shocked if it turned out that way. Why do these countries even want US warships in their ports in the first place? Those reasons aren't going to go away just because there is a risk a few local fishermen could get shot up. They don't exactly contribute heavily to political campaigns in democratic nations, and they aren't likely to be likely to related to the local despot elsewhere.
This is like suggesting that the result of all these NSA revelations is that countries are going to take steps to keep the NSA out. I'm sure 99% of these countries already knew what the NSA was up to already, and likely invited them in the first place. The outrage is just a show for the voters.
Do you have a missile system that can fire about a thousand or so missiles?
From my post: "You just need a few thousand of them manufactured for $500 apiece." The whole point is to fight swarms of cheap bots with swarms of cheap missiles. I don't see why this wouldn't be practical. What necessarily makes a missile expensive? It is a small warhead (cheap), a solid fuel rocket engine (cheap), a sensor of some kind (probably cheap, depending on design), and a computer (cheap). The cost is in the design/etc, but that gets cheaper the more you make.
LOL at the notion of $500 guided missiles. Anti-tank missiles with a range of a mile or two cost almost six figures.
These missiles don't have to destroy a tank.
And unless the state has specifically permitted this by legislation or given you an ironclad immunity deal (let's face it, not many countries are going to be happy to give foreign navies the right to point large guns at their civilians without recourse) such actions would be highly inadvisable. If you point a gun at another vessel and demand it cut its engines while in port you better be able to make a very strong case that it was the threat otherwise you may be spending a lot longer on shore leave than anticipated. In my own jurisdiction those actions would likely have you on the hook for assault and false imprisonment
The ship wouldn't be in port without the permission of the local government. The guy opening fire would only due so under the chain of command, and the local police wouldn't be able to do anything about it since everybody issuing orders would be in the middle of a US naval vessel, if they're even in the same country. The officers who made the call would report back to their superiors.
I'm sure the US navy makes arrangements with the governments where it sends its ships regarding how crisis situations get handled. Most likely the local government will express suitable outrage on TV and then reassure their diplomatic counterparts that everything is fine.
Another likely scenario is that the US spends billions on these boats but doesn't prevent anybody from approaching their vessels to avoid an incident, making the investment in these vessels worthless other than as pork.
Think about it. You're a peaceful merchant, or just a guy going sailing. A fast-attack boat sails up to you, points a 50 cal at you, and shouts on the bullhorn "cut off your engine immediately and put your hands in the air." What are you likely to do? You're going to stick your hands in the air, and thus you aren't likely to get shot. The guy controlling the drone isn't worried about never seeing his family again, so he isn't going to have a twitchy trigger finger. Then you sort the mess out and get an apology.
And this never goes wrong, right? You don't get a pilot who doesn't speak English or is deaf. Maybe the rudder broke and they can't cut the engine. Or maybe they're having a heart attack right now and are five minutes from death. Or maybe they're a terrorist faking one of the above and just trying to get their boat in close before they blow it up.
Sure it will go wrong, but never in a way that results in the US ship being sunk, which is the point. Even if the guy on the boat is deaf, after enough deaf people get shot up in their fishing boats the local deaf fishing community will figure out how to go fishing without getting shot.
The US doesn't build stuff like this for good community relations - they do it to protect their ships.
And taking out Beijing would only cause them to take out Los Angeles.
My point was that the Chinese don't have the ability to take out Los Angeles without the use of ICBMs. If they launch ICBMs at LA, then the US will launch land-based ICBMS at every Chinese military target before the Chinese ones are even at midcourse, and then everybody gets to find out what was in the Chinese warheads while the US nuclear warheads are halfway to China. If the Chinese didn't launch their nuclear arsenal in the first strike, they won't get a second one. The US would save the SLBMs for whatever comes next. Nobody plays games with ICBMs because the game theory basically boils down to "when one flies, they all fly." Again, fantasy hypotheses - the Chinese would never launch an ICBM against the US.
US attacks on the Chinese mainland would be launched by subs - a capability the Chinese basically don't have right now, and which is much harder to counter than sinking an aircraft carrier since subs rely on stealth to a much greater degree.
Also, carriers aren't quite as vulnerable as people think after a war starts. You can't target a carrier with a missile unless you have a sense of where it is. We take that for granted because satellite surveillance is ubiquitous. However, once a real war starts, the first thing everybody will do is shoot down everybody else's satellites and as a result the human race won't be able to put anything in orbit for a few centuries due to debris. Finding a naval task force at sea without the use of satellites is a much harder problem, and one which the US has a major advantage at due to its large navy.
This is basically a password that cannot be changed. Just like fingerprints and retina scans. And all these things can be faked so easily. You don't bother with a false fingerprint or "a practiced mouse grip pattern". No, you replace the device (mouse or fingerprint reader) and fake the digital output from the authentication device. This is much easier.
This is a fundamental weakness of biometrics, but that doesn't make biometrics useless. If the computer is in a supervised area swapping out a mouse isn't going to be trivial, especially if the correct mouse authenticates itself in some way.
When I see stuff like this I think "government installation designed to thwart ninja sneaking in the air ducts and changing into a uniform."
The Persian gulf. The issue there is that lots of tiny boats could swarm US ships and destroy them. This is an Iranian plan by the way. To use the proximity of the US to the shore and just hundreds of little boats. The drone swarm idea appears to be a counter strategy.
I think that ships like this would be more useful for investigating approaching traffic and warning it off so that fishing boats and such don't get accidentally destroyed.
If you wanted to deal with swarms of attack craft I think this is only perhaps useful as a first step. The problem is that these things are just armed with machine guns, which is plenty to sink a small craft at close range, but if you send 200 speed boats against an aircraft carrier and one is carrying an antiship missile that is lethal at 20 miles, then I don't think you can count on stopping that attack with a bunch of drones with short-range weapons.
I think the better option is a medium-range missile system that is inexpensive to deploy in quantity. You can use the fancy Harpoons and such for the fancy warships that have armor and air defenses and all that (and which are so expensive that the enemy only owns a handful). When the enemy sends in 200 zodiacs then you basically just need some cheap slow rockets that fire off in a ballistic trajectory, glide in to a designated target (radar or maybe laser, or maybe even optical), and set off a warhead not much bigger than a hand grenade. You just need a few thousand of them manufactured for $500 apiece. They're not going to do more than scratch the paint on a real warship and maybe a well-trained soldier with a shotgun could shoot them out of the sky, but the enemy won't be able to field hundreds of attack boats capable of defending against a system like this.
Now, you could very well use boats like this as a launch platform for these cheap anti-ship rockets, to further reduce their range/cost. It wouldn't be too hard to station a cheap defense drone every mile along a perimeter outside of the enemy's effective missile range.
The part where you have autonomous boats running around in a busy harbor blocking and possibly destroying other boats.
Think about it. You're a peaceful merchant, or just a guy going sailing. A fast-attack boat sails up to you, points a 50 cal at you, and shouts on the bullhorn "cut off your engine immediately and put your hands in the air." What are you likely to do? You're going to stick your hands in the air, and thus you aren't likely to get shot. The guy controlling the drone isn't worried about never seeing his family again, so he isn't going to have a twitchy trigger finger. Then you sort the mess out and get an apology.
The whole point is to enforce a perimeter far enough away that the folks working the guns can take their time and work things out.
Countries that don't want to deal with this can just choose not to admit US warships into their ports.
How many times have the US Ships put into port / refueled without getting hit? Just seems that it would be a way to spend a ton of money for something that overly complicates normal procedures, and only wards off that 1% of attacks. (Note, percentage pulled out of my rear).
It would ward off the 100% of successful attacks. How many attacks have US naval vessels actually come under in the last decade? These boats are probably going to save more lives than the billion-dollar antiaircraft defense systems they're carrying.
This is about pushing the perimeter away from the boat. They can challenge approaching vessels from a distance, and it gives them the option of getting in close without putting the ship at risk. I'm sure these little boats could be evaded/destroyed/jammed/etc, but at that point the intruder has demonstrated hostile intent and they'll be sunk before they can get anywhere near the real target.
Also.... any autonomous craft would surely need a remote control system. You can't stop the signal (Mal). It wouldn't be impossible for another country / faction to take control of said boats, and use them to accomplish their goal.
Hijacking a military drone is not a trivial matter. All communications will almost certainly be encrypted and authenticated, and tested by folks like the NSA. Jamming is a more likely possibility, but even that isn't necessarily easy, and that just alerts the ship being defended that an attack is underway.
This is about defense in depth. You keep everybody at arms length, and then anybody who tries to get close can be treated as unfriendly. It is like putting up a chain link fence on the prison property line, with a big wall further back. Anybody walking around at the base of the wall has already broken into the property and can be shot on sight.