The remaining difference is probably how much the market size of a embedded system do influence the community that develop Linux distributions. My observation is that this relation is actually insignificant and I see no sign that this will change in a near future.
Well, Debian's glibc maintainer lost his position for negative attitude towards embedded. Some BIG changes in the kernel - like the whole Device Tree concept - were implemented strictly with Embedded in mind. Yes, it's not a primary focus by far, but no, "obligatory" changes that are deterimental to the embedded community are flatly rejected. Systemd being partially useful and for the rest optional, passed but it's not very loved, embedded or not.
Remember it's not just random developers - it's also big chip manufacturers that back embedded Linux. Cypress, AMD, Samsung, Altera, Texas Instruments, Freescale, - by making sure Linux runs smoothly on their chips they expand their customer base, and as such they do have active interest in its development direction. Again, Android, similarly to desktop systems and servers is "visible", but embedded is so ubiquitous that it's really significant - it's just overlooked frequently. But pretty much every single physical thing you have in your hands' reach was made with industrial machinery running embedded. Every paint, chemical and foodstuff was made with involvement of digital composition, QA and most frequently manufacturing too - all with embedded software. When you boot up your PC at least a dozen different OS installations start up in various controllers before your Windows or Linux starts booting.
These all are markets that either currently use Linux or soon will begin, and the manufacturers of their chips pay a close attention to make sure they are able to do so.
Personally, I believe Systemd - or a similar counterpart replacing SysVinit - will stay, but not in the current form. It's very cumbersome, the documentation is abysmally bad, it's monolithic and not modular in the least, and it has own quirks that annoy everyone who is still willing to cope with the other disadvantages. The original intent was noble, but the execution is so bad I'm quite sure it will change.
There's a difference between hiding your personal information (which is yours and nobody has any right to it) and hiding income (which is taxed; a shared resource with "tragedy of the commons" social trap triggered by violators.)
In Europe you can get most of 2) on debit cards as well - and in case of 1) you can set your own limits on the debit card; if you plan a big purchase or overspend in given month, just log into your internet banking account and increase your limits.
Good for you if you have a such simple system to manage, but I design systems far more complex than that, requiring constant monitoring of every applications, low power operation mode, automatic restarting in case of unexpected events, remote management, and dynamic adaptation on hardware events.
Yes, and in your case systemd makes sense. In mine it's very deterimental.
Please understand that if you work on really simple system where systemd features bring nothing to you, the vast majority of others developers works on more complex systems that require something like systemd.
Since you don't work on the systems as simple as mine (not that the devices are simple, there's just as much the OS can provide) - you don't notice how big a market this is. Just like PC software developers fail to realize how big the embedded market as a whole is.
Yes, it's a fairly narrow slice comparing to the whole, but considering how big "the whole" is, it's still very large.
Also, I wouldn't say "vast majority". Because there's a huge army of developers working on systems simpler than mine - "raw iron". My segment is roughly in half of the embedded market complexity-wise, and (if you exclude smartphone app developers) the employment split is roughly equally between the "above" and the "below" me. You just don't notice all the PIC, AVR, small ARM developers that are there and form a wide, solid foundation for your work. And they definitely don't need Systemd.
Yeah, it changes a number in the debt column... too bad it's a variable in a differential equation.
That means even if the value going directly into the debt column is the same, there's still interest to be paid, and the debt grows.
And sooner or later the debt+interest exceeds the credit capacity and your income goes poof. No, not through tax increase because it's too obvious, ugly and unpopular. It goes poof through inflation. You earn the same number of dollars but the dollars can buy you less. It's a process that's hard to notice; it even comes with something that looks and feels positive - as the value you earn dwindles your salary is raised and you earn a higher number of dollars. So what if you can still buy less with them than before...
The numbers stay the same, but their actual meaning is only as the percentage of the whole. And as "the whole" grows they mean a diminishing percentage - literally money is stolen from your wallet and your account - not the dollars as units, but their value.
They do. But regardless, several positive samples of data are more than no samples of data; this accounts for ability to pay debt but not for willingness to do so. A person with positive net score could just get a bunch of credit money and run to a remote country. A person in debt doesn't have the money to make that move.
OTOH the system can be gamed. If you do have a high positive net worth, you can just build up your credit score by taking increasingly bigger loans (and paying them back without ever spending the loaned money) until they build a credit rating good enough to make a big grab and run. But that requires time and effort.
This could be voluntary though. You don't want the credit - you don't share the details. If you desire to keep your net worth in positive numbers, you don't want a credit.
Plus this would harm cheats the most. Honest citizens would remain unhurt.
And here I wonder - why not a debit card? From convenience point of view it's quite equivalent. But you don't pay with money you don't have, which you'd need to pay back at the end of the month. You pay with money you have - it's an electronic wallet - and they can't forbid you to pay as long as you do have the money.
So - what's the deal? Why are Americans so hooked on credit cards? Convenient electronic wallet and a looming wraith of debt with exorbitant "penalty" charges if you fail to fulfil the conditions. And for what - for having a month's worth of salary available a month ahead of time? That gain is only available once, the moment you get the card. Afterwards you're in the same place you were in the beginning - only with extra charges looming if you fail to satisfy the conditions.
Yes, it can, but there are still a few semi-sane people in the government that make sure that if everything crashes and burns you can still get a tenth of what you had back.
" If you don't need something, simply remove the unused package, it's very easy." - then something else fails and you remove it too. Then something else... and soon something essential fails too. The tree of dependencies is quite long and on a system that isn't neatly self-contained it's easy to break it. And in case of systemd, which touches so many functionalities of the OS (including ones you really don't want in your install) it's very easy.
In our systems we don't use systemd, sysv, ustart nor any startup manager. Startup is managed through a tiny, simple, easy to manage set of scripts, launched by init - that way we have a total control over what and when is started without traditional miles of checks and autoconfigs typical to the init systems. Getty on serial, inetd with 4 lines of inetd.conf, manually loading the 4 or so kernel modules required by given project, three lines of setting up the network, a simple condition of either launching the main app or dropping to a service console. Within 20 or so lines of startup script the system is up and doing exactly and only what we need. Oh, and no devfs - plain old/dev with exactly the entries we need.
Switching it all off through OS-provided services? First disabling some 40 services we don't need, then dealing with the leftover stubs of the init system that need to grind through all the options just to determine they are disabled. No, thanks.
Routers. Nice home/office environment. Eh, should have known.
Wanna pay a helicopter ride to Spitzbergen to fix one of our devices? -60C~+120C, running off a solar battery and a rechargeable battery (rated for the same conditions) that is to last through the polar night of a couple months, withstand 150km/h blizzard?
Your fancy box wouldn't last a day. But that one isn't running Linux. There won't be a Linux-capable hardware that can fulfil these requirements for decades yet.
What I'm working on, personally, is a board that has 16MB RAM, 32MB Flash, and is rated -40C~+120C, survive for 30 minutes on battery power and survive significant EMI interferences. It also uses a lot of features. And the board is manufactured in two variants, the other is 64MB RAM and 128MB Flash, and it's about $20 more. It's also manufactured in indoor-only variants (+4-+40C) which are significantly cheaper. But the device is the size of a refrigerator and not sold in bulk.
My co-worker next desk over though, works on a device that will be sold in bulk, size about half a box of cigarettes, running off two AAA accumulators and a solar battery about the size of an A5 paper sheet, maintenance-free, for years. It needs only Bluetooth and SPI. And again, it must survive outdoor conditions although it's not meant to operate below 0C, merely survive the winter inactive. Do you think he can afford to boot and run a dozen unnecessary daemons? Can you think of a chip and fully-featured distro to fulfill these requirements?
It's not just a trade-off between chip capacity and work difficulty. It's whether given device is possible at all.
And they can print washers of soft, radiation-proof rubber capable of surviving ~160K-300K temperature change without significant change in elasticity; forming an airtight seal despite dynamic stress of kilonewton order, resistant to exposure to hydrazine and its reaction products, precise to 0.1mm, and about 1m in diameter?
Because it's them getting most worn; hydrazine from jet engines and space radiation causing wear that makes them increasingly brittle. And they are installed in a way that doesn't really encourage replacing them on EVA. Never mind that replacing them requires temporarily splitting the station in half.
Back in the 1800's people were way more professional/courteous. Then it appeared that common workers want to be considered "people" too (look! How funny, he thinks he's people!) and the 1800's ended with some more or less bloody revolutions.
Your politeness is not a measure of your professionalism. Your politeness is not a measure of your humanity. Your politeness is not a measure of your value comparing to other people.
Your politeness only affects the superficial aesthetic impression you cause. And people, who value others by the superficial aesthetic expression are definitely not professional.
Oh, but what about consequences? If the other party felt offended by your assertions, they can file a complaint and regardless of merit of the argument, the side that (allegedly) offended the other is facing consequences.
Just try to publish performance comparison at specific technical jobs between male and female employees. No matter how much scientific scrutiny you have applied to accuracy of the comparison, your career has just ended, you sexist swine.
Some people are willing to cope with co-workers using strong language if it means they don't need to watch their own language every turn and feel safe to joke about forking someone's repository or using dongles, without fear of being fired, their technical merit notwithstanding.
The "friendly, gentle and safe" environment envisioned by the feminists and the likes turns out to be quite oppressive to a number of professionals, with need to think twice before you choose to speak your mind, lest you offend some sensitive ears and face disciplinary action for violation of code of conduct.
- A developer of a complex project can, and does use abusive language in case of bad screw-up. If you fucked up, you may expect some bad tongue-lashing. It's clear who screwed up. If they don't improve nevertheless, they are expunged.
- Nobody on the team is allowed to use abusive language no matter what the situation and how badly the other party fails, and keeps failing; only mild and sensitive criticism is allowed and can be ignored without consequences. Violation of the good behaviour rules regardless of the situation results in disciplinary consequences including expulsion. Blame for project failure is shared equally between members of the team; nobody gets singled out and called out on specific reason for failure; everyone suffers the same consequences, usually related to finances and promotions.
Are you claiming the former situation will drive more developers away than the latter?
If you call about $20 savings per board, for a device that is produced in 5mln copies "no gain", then yes, there is no gain. And if some *unnecessary* subsystem of a fully-featured Linux fails 6 months after release (say, because it was writing logs to the same piece of flash over and over), the cost of troubleshooting and fixing 5mln devices already out in customers' hands overshadows any savings on development time....especially, that learning Yocto and integrating your solution with it is easier than learning Systemd and integrating your solution with THAT.
New microcontrollers are designed all the time, but not only faster and with bigger capacity, but in versions that are better at saving energy, smaller or ones that are simply cheaper than old counterparts. They are not vanishing. Plus the situation isn't nearly as happy as you want to picture it with heavily miniaturized parts - if you need to put a GSM module with antenna and SIM, an SD card reader, an USB passive electronics, and fit with this a chip to run Linux plus RAM it needs - and fit it all in a thumb drive package - good luck locating a quad-core >1.2GHz embedded CPU with integrated flash and necessary peripherals that will fit with them. *Maybe* they will appear in a couple years.
You seem to live in some tiny segment of the embedded world, far away from large-volume production and "small embedded", and you have completely lost the big image. You don't see the long line of devices that will not grow into "big" Linux for now but would benefit from actual OS, that are just waiting for chips cheaper, smaller and less power-hungry, but not faster or with more RAM.
And yeah, maybe someday systemd will be the standard for the "small ones" too. A subsystem similar to Busybox conceptually, but handling demons and services, would be quite welcome. But first it must become more modular, less demanding and easier to integrate with custom-purpose services.
The significant difference is that Linus isn't a pointless jerk - he just reacts aggressively to stupidity or arrogance. But you won't find him parking his car on a disabled people's place, or cheating a co-worker/partner out of a fair share of profits.
It will need to be scrapped. There are parts that are about impossible to replace, that decay over time and once they fail the astronauts would die. There are problems - like growing molds/bacteria - that are about impossible to get rid of now, with the station not built with removal of them in mind. There are scientific experiments that ran their course and no longer needed, and ones that can't be done on the ISS. The station is a zoo of docking port standards, with lots of adapters to connect incompatible modules. It's a sink of resources, instead of being a supply station for departing missions. It is undermanned at all times, as there's only one Soyuz escape capsule so it can't operate with the full crew. These are all significant shortcomings that can be addressed in a next space station.
It doesn't mean it needs to be disposed of whole. There's a lot of systems that will be good for another fifty years. Massive outer structures that cost a fortune to be brought to orbit and that simply don't have anything in them that could fail. They can be reused in a new space station.
Last mission to Salyut 7 was to pick up everything that would be of use on brand new then Mir. Simply, useful equipment was reused in a newer, bigger space station. So don't think about crashing ISS as "a great loss". Think of it as "a thorough upgrade".
By the way, still not all modules are in place... The Nauka module - a Russian laboratory, and NEM-1 - a support/habitat/power/storage module to supplement Nauka - are still missing.
People who shun central planning, giving the Soviet Union economy as example of over-regulated market being a failure, and claiming that shortcomings of other markets come from government regulation, tend to forget that there IS one country where the trade is completely free - not regulated by the government in the least. It shows exactly what you can, and should expect from a totally free market in the real world.
Slashdot Mirror
Well, Debian's glibc maintainer lost his position for negative attitude towards embedded. Some BIG changes in the kernel - like the whole Device Tree concept - were implemented strictly with Embedded in mind. Yes, it's not a primary focus by far, but no, "obligatory" changes that are deterimental to the embedded community are flatly rejected. Systemd being partially useful and for the rest optional, passed but it's not very loved, embedded or not.
Remember it's not just random developers - it's also big chip manufacturers that back embedded Linux. Cypress, AMD, Samsung, Altera, Texas Instruments, Freescale, - by making sure Linux runs smoothly on their chips they expand their customer base, and as such they do have active interest in its development direction. Again, Android, similarly to desktop systems and servers is "visible", but embedded is so ubiquitous that it's really significant - it's just overlooked frequently. But pretty much every single physical thing you have in your hands' reach was made with industrial machinery running embedded. Every paint, chemical and foodstuff was made with involvement of digital composition, QA and most frequently manufacturing too - all with embedded software. When you boot up your PC at least a dozen different OS installations start up in various controllers before your Windows or Linux starts booting.
These all are markets that either currently use Linux or soon will begin, and the manufacturers of their chips pay a close attention to make sure they are able to do so.
Personally, I believe Systemd - or a similar counterpart replacing SysVinit - will stay, but not in the current form. It's very cumbersome, the documentation is abysmally bad, it's monolithic and not modular in the least, and it has own quirks that annoy everyone who is still willing to cope with the other disadvantages. The original intent was noble, but the execution is so bad I'm quite sure it will change.
There's a difference between hiding your personal information (which is yours and nobody has any right to it) and hiding income (which is taxed; a shared resource with "tragedy of the commons" social trap triggered by violators.)
In Europe you can get most of 2) on debit cards as well - and in case of 1) you can set your own limits on the debit card; if you plan a big purchase or overspend in given month, just log into your internet banking account and increase your limits.
Yes, and in your case systemd makes sense. In mine it's very deterimental.
Since you don't work on the systems as simple as mine (not that the devices are simple, there's just as much the OS can provide) - you don't notice how big a market this is. Just like PC software developers fail to realize how big the embedded market as a whole is.
Yes, it's a fairly narrow slice comparing to the whole, but considering how big "the whole" is, it's still very large.
Also, I wouldn't say "vast majority". Because there's a huge army of developers working on systems simpler than mine - "raw iron". My segment is roughly in half of the embedded market complexity-wise, and (if you exclude smartphone app developers) the employment split is roughly equally between the "above" and the "below" me. You just don't notice all the PIC, AVR, small ARM developers that are there and form a wide, solid foundation for your work. And they definitely don't need Systemd.
Yeah, it changes a number in the debt column... too bad it's a variable in a differential equation.
That means even if the value going directly into the debt column is the same, there's still interest to be paid, and the debt grows.
And sooner or later the debt+interest exceeds the credit capacity and your income goes poof. No, not through tax increase because it's too obvious, ugly and unpopular. It goes poof through inflation. You earn the same number of dollars but the dollars can buy you less. It's a process that's hard to notice; it even comes with something that looks and feels positive - as the value you earn dwindles your salary is raised and you earn a higher number of dollars. So what if you can still buy less with them than before...
The numbers stay the same, but their actual meaning is only as the percentage of the whole. And as "the whole" grows they mean a diminishing percentage - literally money is stolen from your wallet and your account - not the dollars as units, but their value.
They do. But regardless, several positive samples of data are more than no samples of data; this accounts for ability to pay debt but not for willingness to do so. A person with positive net score could just get a bunch of credit money and run to a remote country. A person in debt doesn't have the money to make that move.
OTOH the system can be gamed. If you do have a high positive net worth, you can just build up your credit score by taking increasingly bigger loans (and paying them back without ever spending the loaned money) until they build a credit rating good enough to make a big grab and run. But that requires time and effort.
This could be voluntary though. You don't want the credit - you don't share the details. If you desire to keep your net worth in positive numbers, you don't want a credit.
Plus this would harm cheats the most. Honest citizens would remain unhurt.
And here I wonder - why not a debit card? From convenience point of view it's quite equivalent. But you don't pay with money you don't have, which you'd need to pay back at the end of the month. You pay with money you have - it's an electronic wallet - and they can't forbid you to pay as long as you do have the money.
So - what's the deal? Why are Americans so hooked on credit cards? Convenient electronic wallet and a looming wraith of debt with exorbitant "penalty" charges if you fail to fulfil the conditions. And for what - for having a month's worth of salary available a month ahead of time? That gain is only available once, the moment you get the card. Afterwards you're in the same place you were in the beginning - only with extra charges looming if you fail to satisfy the conditions.
Yes, it can, but there are still a few semi-sane people in the government that make sure that if everything crashes and burns you can still get a tenth of what you had back.
" If you don't need something, simply remove the unused package, it's very easy." - then something else fails and you remove it too. Then something else... and soon something essential fails too. The tree of dependencies is quite long and on a system that isn't neatly self-contained it's easy to break it. And in case of systemd, which touches so many functionalities of the OS (including ones you really don't want in your install) it's very easy.
In our systems we don't use systemd, sysv, ustart nor any startup manager. Startup is managed through a tiny, simple, easy to manage set of scripts, launched by init - that way we have a total control over what and when is started without traditional miles of checks and autoconfigs typical to the init systems. Getty on serial, inetd with 4 lines of inetd.conf, manually loading the 4 or so kernel modules required by given project, three lines of setting up the network, a simple condition of either launching the main app or dropping to a service console. Within 20 or so lines of startup script the system is up and doing exactly and only what we need. Oh, and no devfs - plain old /dev with exactly the entries we need.
Switching it all off through OS-provided services? First disabling some 40 services we don't need, then dealing with the leftover stubs of the init system that need to grind through all the options just to determine they are disabled. No, thanks.
Routers. Nice home/office environment. Eh, should have known.
Wanna pay a helicopter ride to Spitzbergen to fix one of our devices? -60C~+120C, running off a solar battery and a rechargeable battery (rated for the same conditions) that is to last through the polar night of a couple months, withstand 150km/h blizzard?
Your fancy box wouldn't last a day. But that one isn't running Linux. There won't be a Linux-capable hardware that can fulfil these requirements for decades yet.
What I'm working on, personally, is a board that has 16MB RAM, 32MB Flash, and is rated -40C~+120C, survive for 30 minutes on battery power and survive significant EMI interferences. It also uses a lot of features. And the board is manufactured in two variants, the other is 64MB RAM and 128MB Flash, and it's about $20 more. It's also manufactured in indoor-only variants (+4-+40C) which are significantly cheaper. But the device is the size of a refrigerator and not sold in bulk.
My co-worker next desk over though, works on a device that will be sold in bulk, size about half a box of cigarettes, running off two AAA accumulators and a solar battery about the size of an A5 paper sheet, maintenance-free, for years. It needs only Bluetooth and SPI. And again, it must survive outdoor conditions although it's not meant to operate below 0C, merely survive the winter inactive. Do you think he can afford to boot and run a dozen unnecessary daemons? Can you think of a chip and fully-featured distro to fulfill these requirements?
It's not just a trade-off between chip capacity and work difficulty. It's whether given device is possible at all.
And they can print washers of soft, radiation-proof rubber capable of surviving ~160K-300K temperature change without significant change in elasticity; forming an airtight seal despite dynamic stress of kilonewton order, resistant to exposure to hydrazine and its reaction products, precise to 0.1mm, and about 1m in diameter?
Because it's them getting most worn; hydrazine from jet engines and space radiation causing wear that makes them increasingly brittle. And they are installed in a way that doesn't really encourage replacing them on EVA. Never mind that replacing them requires temporarily splitting the station in half.
Back in the 1800's people were way more professional/courteous. Then it appeared that common workers want to be considered "people" too (look! How funny, he thinks he's people!) and the 1800's ended with some more or less bloody revolutions.
Your politeness is not a measure of your professionalism.
Your politeness is not a measure of your humanity.
Your politeness is not a measure of your value comparing to other people.
Your politeness only affects the superficial aesthetic impression you cause. And people, who value others by the superficial aesthetic expression are definitely not professional.
Oh, but what about consequences? If the other party felt offended by your assertions, they can file a complaint and regardless of merit of the argument, the side that (allegedly) offended the other is facing consequences.
Just try to publish performance comparison at specific technical jobs between male and female employees. No matter how much scientific scrutiny you have applied to accuracy of the comparison, your career has just ended, you sexist swine.
Some people are willing to cope with co-workers using strong language if it means they don't need to watch their own language every turn and feel safe to joke about forking someone's repository or using dongles, without fear of being fired, their technical merit notwithstanding.
The "friendly, gentle and safe" environment envisioned by the feminists and the likes turns out to be quite oppressive to a number of professionals, with need to think twice before you choose to speak your mind, lest you offend some sensitive ears and face disciplinary action for violation of code of conduct.
So we have two typical team development models:
- A developer of a complex project can, and does use abusive language in case of bad screw-up. If you fucked up, you may expect some bad tongue-lashing. It's clear who screwed up. If they don't improve nevertheless, they are expunged.
- Nobody on the team is allowed to use abusive language no matter what the situation and how badly the other party fails, and keeps failing; only mild and sensitive criticism is allowed and can be ignored without consequences. Violation of the good behaviour rules regardless of the situation results in disciplinary consequences including expulsion. Blame for project failure is shared equally between members of the team; nobody gets singled out and called out on specific reason for failure; everyone suffers the same consequences, usually related to finances and promotions.
Are you claiming the former situation will drive more developers away than the latter?
If you call about $20 savings per board, for a device that is produced in 5mln copies "no gain", then yes, there is no gain. And if some *unnecessary* subsystem of a fully-featured Linux fails 6 months after release (say, because it was writing logs to the same piece of flash over and over), the cost of troubleshooting and fixing 5mln devices already out in customers' hands overshadows any savings on development time. ...especially, that learning Yocto and integrating your solution with it is easier than learning Systemd and integrating your solution with THAT.
New microcontrollers are designed all the time, but not only faster and with bigger capacity, but in versions that are better at saving energy, smaller or ones that are simply cheaper than old counterparts. They are not vanishing. Plus the situation isn't nearly as happy as you want to picture it with heavily miniaturized parts - if you need to put a GSM module with antenna and SIM, an SD card reader, an USB passive electronics, and fit with this a chip to run Linux plus RAM it needs - and fit it all in a thumb drive package - good luck locating a quad-core >1.2GHz embedded CPU with integrated flash and necessary peripherals that will fit with them. *Maybe* they will appear in a couple years.
You seem to live in some tiny segment of the embedded world, far away from large-volume production and "small embedded", and you have completely lost the big image. You don't see the long line of devices that will not grow into "big" Linux for now but would benefit from actual OS, that are just waiting for chips cheaper, smaller and less power-hungry, but not faster or with more RAM.
And yeah, maybe someday systemd will be the standard for the "small ones" too. A subsystem similar to Busybox conceptually, but handling demons and services, would be quite welcome. But first it must become more modular, less demanding and easier to integrate with custom-purpose services.
The significant difference is that Linus isn't a pointless jerk - he just reacts aggressively to stupidity or arrogance. But you won't find him parking his car on a disabled people's place, or cheating a co-worker/partner out of a fair share of profits.
It will need to be scrapped. There are parts that are about impossible to replace, that decay over time and once they fail the astronauts would die. There are problems - like growing molds/bacteria - that are about impossible to get rid of now, with the station not built with removal of them in mind. There are scientific experiments that ran their course and no longer needed, and ones that can't be done on the ISS. The station is a zoo of docking port standards, with lots of adapters to connect incompatible modules. It's a sink of resources, instead of being a supply station for departing missions. It is undermanned at all times, as there's only one Soyuz escape capsule so it can't operate with the full crew. These are all significant shortcomings that can be addressed in a next space station.
It doesn't mean it needs to be disposed of whole. There's a lot of systems that will be good for another fifty years. Massive outer structures that cost a fortune to be brought to orbit and that simply don't have anything in them that could fail. They can be reused in a new space station.
Last mission to Salyut 7 was to pick up everything that would be of use on brand new then Mir. Simply, useful equipment was reused in a newer, bigger space station. So don't think about crashing ISS as "a great loss". Think of it as "a thorough upgrade".
By the way, still not all modules are in place... The Nauka module - a Russian laboratory, and NEM-1 - a support/habitat/power/storage module to supplement Nauka - are still missing.
And it's badly in need of room service to tidy up the place. That place is a mess.
99.65% over a single pass or totally? Because everything that passes the filter returns to the system... and then returns to the filter eventually.
Putin will kill all these pathogens with his bare fists.
People who shun central planning, giving the Soviet Union economy as example of over-regulated market being a failure, and claiming that shortcomings of other markets come from government regulation, tend to forget that there IS one country where the trade is completely free - not regulated by the government in the least. It shows exactly what you can, and should expect from a totally free market in the real world.
It's Somalia.
Why? The photons from the laser reach some CCD that passes the results to some display device. Switch the CCD off.
Sounds interesting, can you provide some links on Sony suing themselves?