What is needed, before "advancing" anything, is to advance acceptance of the Linux desktop, and IMHO this ain't helping.
That's only true if you care more about acceptance of the Linux desktop than about its functionality. If you work for a company like Canonical that is likely the case. If you get paid the same whether the whole world uses Linux or only the development community uses it, then it probably isn't the case.
People scratch their OWN itches when developing FOSS. That's why there will never be one true linux distro. Sure, the users of that distro would benefit from all the concentrated free labor, but people aren't donating their labor only to make a linux distro that YOU like.
It sounds like a really good use case for distributed version control. When this sort of thing happens, developers should be able to temporarily fork the repo and work on security issues in private, while everyone else is still able to access the main repo.
Sure, if you have infrastructure to run a hidden repo that only your devs can access. They likely don't have this, as is the case with most FOSS projects.
My post-paid plan had zero cost for the kids lines, actually, but it isn't a plan currently offered. That gets 500min/mo on each of the lines.
I do agree with you though. Plans are always changing, as are phones. When people tell me what phone they should get in six months I tell them to ask me again in six months. The same is true of plans. Last summer when I re-signed with T-Mo the value plans were the best option for my circumstances, but that could easily be different next time.
Still, $120/mo for unlimited voice and text on four lines, 2GB 3G data on two lines, 500MB 3G data on two lines, and unlimited 2G data on all lines is a VERY good deal. That would be hard to beat pre-paid assuming you needed all that data.
Unfortunately when it comes to physical infrastructure, redundancy is often just not possible.
That is simply a matter of priorities.
Moderately large bridges cost a few million dollars to build (we're talking about suspension bridges here - not the little ones that go over creeks). We probably could have replaced every suspension bridge in the US and built 5x as many more for much less than the cost of one of our recent wars, and this would have likely had a HUGE impact on the economy.
Sure, when you build a bigger highway more people drive on it, but there is a REASON they drive on it, and a benefit from them doing so. More transportation means less redundant industry, lower costs of living, more job portability, and so on.
Building new fighter jets involves high levels of project risk, huge amounts of money, and little economic benefit (unless you're actually at risk of being invaded). Sure, you will always need a military, but does it have to be 10X larger than the next 5 combined?
Building a new bridge involves almost no project risk. It costs money, but it is a solved problem. The only reason replacing bridges causes so much disruption is because we're running our infrastructure way over what it really is able to manage. If a Google query took 30 seconds to respond you can bet they'd be concerned about the costs of taking down hardware to upgrade it. However, Google is a sensible business and realizes that fast queries make them more money, so when they replace hardware the impact of outages is measured in microseconds.
The only problem with this approach is that it is usually more expensive (no linkage to family plan, etc). Agree that it is a good teaching approach, though - maybe that makes it worth it in the end.
However, I think there is still a need for a legislative solution. We don't need to regulate plan offerings or anything like that. Instead, just let customers put a limit on their bill, which can be as low as whatever number was advertised on the window (which will need to include all taxes/fees). If the provider delivers any service in excess of that amount, the service is free (this isn't a payment plan - they can't charge for the service ever). Providers can of course cut off service or not deliver optional services that aren't going to be paid for until the next billing cycle.
The basis for providing a service should be a conscious agreement between buyer and seller. Anything else is little more than a scam. If the buyer doesn't realize they're buying something, then the sale is null and void.
I've done a fair bit of comparison for home mailing purposes. For small packages USPS tends to be the best option. Once you start getting moderately large/heavy (nothing you can't pick up) UPS gets a lot cheaper. Also, if you care about full tracking UPS or Fedex can be cheaper - the cheapest USPS services don't get you full tracking, signature, etc. Oh, don't forget declared value either. For UPS it is free up to some value (maybe $100?), and for USPS it always costs money and I think you have to be at a higher class of service.
I usually check all three services anytime I ship - I can't say I consistently find any to always be cheaper. USPS tends to be best when you don't care about tracking and the box is small. For some intermediate cases Fedex does well, especially for light things that you want to track. UPS tends to do well for medium-sized packages in general.
There are a lot of other things to consider. If you're dealing with a party you don't trust signatures are a must, as is a declared value. Oh, and consider packaging as well. If you ship via a certified partner and pay them some token fee to do the packaging and declare the value and require a signature then you're basically zero-risk in terms of anything that could go wrong. The last time I sold a cell phone I did that - cost me all of maybe $10 more, but I didn't have to worry about complaints about broken/lost/etc, reversals, etc. When you're shipping to some individual and you aren't willing to eat the cost of fraud it is nice to basically have everything insured.
Uh, if you're shipping lead ingots all the way across the country the flat-rate boxes are probably the best deal. For just about anything else, not so much. Their main advantage is being flat-rate, and free standard-sized boxes. If you're paying the guy who is doing all the postage work it makes more sense as a result.
Then again, if you're running a business you can just get a postage mailing system with integrated scale/printer/etc and flat-rate has basically no advantage at all.
Telecom in the US has been full with the equivalent of city window washers for decades. It all comes down to a flawed trust system.
If I want to change services, I should call up the phone company and switch my services. However, that isn't as vulnerable to impulse buying, cold calling, and general scamming, so that approach simply won't do in the US.
The result is that if I run some kind of registered phone service I can basically send a bill to any telephone company in the country for any phone number in the country and get paid money. Then the poor sucker I scammed has to beg and plead with the phone company to try to reverse the charges while I retire to someplace in the Caribbean. There are phone numbers that charge you if you call them, international area codes that don't look like they're international, and so on.
SMS scams are just another example of this. Your kid texts some joke-of-the-day service and the parent gets a $40 bill. Sure, you can yell at the kid, but in what world are we supposed to allow children to sign contracts that are binding on their parents?
Yup, the nice thing about T-Mobile is that you can't get hit with per-MB charges, ever. The worst that can happen is you get throttled. That means you don't have to buy an extra-large plan "just in case" or check your data every other day.
Only Sprint offers a similar deal. Verizon and ATT just love the add-on charges. Back when I had Verizon I found myself having to over-spend on my plan just to avoid getting the surprise $600 phone bill. That is especially a big concern if you have kids. Sure, you can yell at them and it doesn't happen often, but it is cheaper to pay the extra $20/mo as "insurance." I resented this and am saving money and don't have to worry about the insurance at all...
Keep in mind you compared an unlimited data plan to a 3GB data plan. The 2GB data plan for T-Mobile is only $10/month.
Oh, I've found that T-Mobile has much better family plans as well. For $120/month you can have four phones with unlimited calling and 2GB data on two of them. For ATT that would be around $200/month. Oh, and it still is apples and oranges as T-Mo is now throwing in 500MB on all lines and NO overage fees (it just drops to 2G). ATT would charge $240/mo to bump the other two lines up to only 200MB and if you get one too many emails, say hello to per-MB pricing.
In my experience T-Mobile is way cheaper in just about every way these days. Their only issue is coverage, but even on vacation I haven't found it to be bad. I live in a suburb.
I don't think they support WiFi calling on phones they don't sell. It doesn't always work on modded phones either.
While I love T-Mobile I'm not sure I'd ever buy another phone from them. Their phones tend to be expensive (well, all the other major carriers are too). You can save quite a bit by buying from Google, or from some Asian supplier.
Bullet striations are about as reliable as a lie detector test.
You may as well consult an astrologist.
Maybe, but they'll convict you on it just the same. It isn't any different from any of the other forensic techniques dating back to the 80s and earlier, including many forms of DNA testing. The prosecutor doesn't care they work, they just care that the necessary precedents are in place so that they can send the defendant to jail.
The purpose of a prosecutor is to punish somebody for a crime as severely as possible. The purpose of a court is to ensure the prosecutor jumps through the legally-required hoops before doing so. Or, at least, that is what all the incentives promote.
Something others didn't mention is that this is SATELLITE data, not data from aerial photos.
When you look at Google Maps "satellite" view you are likely looking at a photo taken by a plane. Obviously it is much easier to get a high-resolution photo of a house from a plane a mile or two up than from a satellite 350 miles up.
Satellite photos have the advantage of being easier to acquire more regularly. The satellite flies over the country every day whether you need a photo or not. It will never be able to compete with a photo taken from a plane, let alone one taken from the ground. These are technologies that solve different problems.
There is definitely a use for regular civilian satellite images of the entire Earth's surface.
FOSS projects smaller than KDE manage to back up their data. All it requires is that somebody cares.
They have websites, they have mirrors, they have servers. They can afford backups. They likely have servers donated by companies that could back up those servers for them if asked. With git anybody can do a clone a day and rotate those clones, and those would be backups.
I'm involved in a much smaller FOSS project with a much smaller budget, and yet we spend thousands of dollars a year on hardware/hosting/etc and we do backups.
I believe a properly administered git mirror will do everything the GP wanted, except for testing the backups (which isn't git's problem) and the "2-sided crypto-hash compare" because, in my amateurishness, I don't know what that is.
The git mirror will overwrite the last backup with the next - it does not rotate media. Now, you can do a git clone to a new location every time and THAT might accomplish all the backup best-practices (assume you test).
You need to rotate your backups because you can't assume that you'll detect a problem with your data before your previous backup gets overwritten. What happens if you find a glitch that goes back a month?
The further back you want to go the more expensive things are, but in general you want to be able to go back at least a few days. For something as small as a git repository (even a big one) there is no reason not to rotate back at least a month or two.
More accurately the problem is that the hardware resources available to KDE are very limited and the KDE repository is one of the largest git repositories in the world.
I doubt it is more than a few GB per backup compressed. If they rotated 20 backups using daily/weekly/monthly rotations they'd have a week of dailies and a year of less frequent backups. That would cost them all of maybe a few bucks a month even with retail-cost solutions like Amazon S3 - WAY less than they must be paying for their servers. They'd only need backups going back a few days to save themselves from a rapidly-detected disaster.
Oh, and anybody could do a backup of KDE like this, assuming they allow anybody to clone their full repositories. A clone is a point-in-time snapshot, and then if you don't overwrite each snapshot with the next you have backups. It isn't like it takes special hardware to pull this off.
If you can backup a live database you can certainly backup a live git repository.
Stick it on zfs or btrfs, create snapshots, and then offline backup the snapshots. That's really easy to do, and atomic.
Or just do a git clone or create a bundle and back that up offline. That is definitely atomic. Oh sure, you won't get every last commit, but this is a BACKUP, not a mirror, and you're not going to rotate them every five minutes. If you have to resort to backups you'll lose some commits, but at least you won't lose your last decade of history or whatever.
Or you can offline the repository for a few minutes while you umount it, LVM snapshot it, and then remount it. That gives you a more defined recovery point for what it's worth, but at the cost of some downtime.
Lots of ways to backup a live repository. Any of them are better than not having backups. They can still have mirrors - they're useful as well, but they solve a different set of problems.
Yup, it all starts with defining the failure modes you want to protect against, and appreciating what techniques protect against what modes.
However, if you don't really appreciate all that stuff or can't be bothered to, then a safe default is rotational backup to offline media stored at a separate location from the device being backed up. There are plenty of services who make this sort of thing easy to do if you can't be bothered to get it right yourself - many are cloud-based and designed for clueless PC owners, and commercial solutions exist for Linux as well. There is no excuse for the maintainers of an FOSS project to not have an adequate backup solution for their source repositories.
Mirrors and RAID don't involve rotation and that makes them effective against only a small number of failure modes. Those who think they're "good enough" for something like this don't have good imaginations. I live with RAID-only only for things that I can reproduce (such as re-installing, re-ripping, etc), or things I don't mind losing (like hundreds of hours of MythTV recordings). For anything with actual value I do rotational backups uploaded to an offsite location, with periodic testing (in particular anytime my backup software is updated). During testing I carefully peruse my backed up data so I have a pretty good appreciation for what I will and won't get back in the event of a disaster. These days decent backup is pretty cheap - there is no excuse for not having it where it matters.
Re:A thousand times. (Unless online mirrors roll b
on
Too Perfect a Mirror
·
· Score: 1
Now, improving uptime and reducing downtimes is important, but it is not what a backup does.
Well, a backup does contribute to reducing downtimes, albeit not so much as RAID/etc. Compared to doing a full reinstall/reconfiguration restoring from backup is likely to be much faster. That is why backup can be useful even on systems that do not contain unreproducible data. There are other strategies that have other advantages (like automatic builds/etc) which are also effective if there is no data involved.
I do agree that the primary purpose of a backup is to prevent the loss of data in as many failure modes as possible/practical. Mirrors are definitely not backups (or at least, not very good ones - there is a continuum when it comes to backups, just as there is a continuum when it comes to disasters).
Tend to agree on this one, and it is the sort of thing that has been running companies into the ground lately (usually due to too much MBA-think - they forget they're supposed to manage people and not just spreadsheets).
Employees who just check out is something that is hard to measure, and yet it can cost a company a fortune. Then you get tons of training programs encouraging employees to speak up about problems and try to make the workplace happier, and they typically neglect the underlying issues that are causing those problems in the first place so they have no effect.
If an airport receives a significant volume of traffic then it should have a tower (and that need not be more than an office with a reasonable view of the field and a radio - it doesn't have to be some huge complex with supervisors, etc).
If an airport does not receive a significant volume of traffic it should be uncontrolled, and aircraft procedures already enable pilots to handle this situation. You can even fly at such an airport when visibility is poor - the local air traffic control simply creates a bubble around the airport for a specified period of time so that there isn't a conflict (though this means waiting for anybody who wants to takeoff/land since only one aircraft can hold such a clearance at a time).
The interstate commerce clause has been stretched so far it will cover anything.
The prosecutor will simply argue that because you made your own howitzer in your basement in Kansas you didn't end up buying one from a licensed howitzer manufacturer in Vermont, and therefore you had an effect on interstate commerce. Never mind that you couldn't have legally bought that howitzer from Vermont anyway.
These kinds of arguments have been used to apply Federal regulations to virtually everything. After all, you could apply that sort of thing to anything. If I watch a movie across the street I satisfy entertainment urges that might have otherwise have required me to drive across a state line to the nearest beach, so that movie theater is affecting interstate commerce, and so on...
Not only that, it was done illegally by bypassing the cabinet.
How can a President "illegally" bypass the cabinet? The president isn't constitutionally even required to have a cabinet in the first place, as far as I can see. I don't buy into the whole "unitary executive means the president can just ignore the law" bit, but the President is, well, the President. Ultimately he's the one responsible for upholding the law and conducting the affairs of the executive branch, and everybody else is just there to help out.
If I missed some clause in the constitution by all means let me know.
I do tend to think that wages make a difference. I also think one problem lies in keeping them in the profession as opposed to going into a more lucrative field. I haven't seen much data to back that up though.
I think half the problem is how the pay is structured. In US public schools teacher pay is based on seniority, full stop.
The art teacher who has been working for 15 years gets paid the same as the math/science teacher who has been working for 15 years. The teacher who sticks a page of notes on a projector screen and tells their students to copy it gets paid the same as the teacher who uses the Socratic method and whose students perform 3x better.
As a result, if you have an extra $10M to spend on wages the only way to spend it is to take the money, divide it by the number of teachers, apply some curve so that more goes to those who have been there the longest, and then give everybody an extra $100-1000 dollars. No private company facing a crisis would ever spend money that way - they'd target how the money could best be used and they'd spend the same amount of money in a much smaller area and get a dramatic result.
If teachers were paid market wages per subject then you could actually hire great new math/science teachers without having to pay the music teacher with 25 years seniority $500k/yr. You can't expect somebody with an MS in a physical science to start out at $20k/yr and pay their dues as they work their way up. The only people you hire that way are those who couldn't get a job elsewhere (mixed in with a few salt-of-the-earth types who would find a way to teach even if you didn't pay them at all).
Pay is just the start though - there is quite a bit that needs to be reformed. The other big problem is that you can't really force kids to learn - that is a culture thing that starts at home/etc. You can have Feyman teaching your physics class and it won't do any good for the kid who is just there to hang out with friends and do his time. In many ways I think the pay problem is just a symptom of larger societal problems. Why work hard to get A's in physics if you can get a much higher salary by being in the right fraternity and landing a job in some bank?
The US military is really good at blowing things up - nobody does it better. As long as that's how you use them things go just fine. The problem is that when your tool is blowing things up and your goal is to establish a government with liberty and justice for all, things don't always work out well.
The locals need to want true democracy before you can try to establish it. If this were about being the French and blockading the British so that the US revolutionaries could finish setting up a democracy that would be one thing. However, what this is more likely to turn into is providing for one particular religious faction so that they can wipe out all the other ones, then hold an election where a new dictator for life is selected. Maybe it will go slightly better, but there is a huge culture difference between the Middle East and the US/Europe. I guess they just haven't had enough time to get tired of all the Salem Witch Trials and Inquisitions and they still want to see if they work.
Slashdot Mirror
What is needed, before "advancing" anything, is to advance acceptance of the Linux desktop, and IMHO this ain't helping.
That's only true if you care more about acceptance of the Linux desktop than about its functionality. If you work for a company like Canonical that is likely the case. If you get paid the same whether the whole world uses Linux or only the development community uses it, then it probably isn't the case.
People scratch their OWN itches when developing FOSS. That's why there will never be one true linux distro. Sure, the users of that distro would benefit from all the concentrated free labor, but people aren't donating their labor only to make a linux distro that YOU like.
It sounds like a really good use case for distributed version control. When this sort of thing happens, developers should be able to temporarily fork the repo and work on security issues in private, while everyone else is still able to access the main repo.
Sure, if you have infrastructure to run a hidden repo that only your devs can access. They likely don't have this, as is the case with most FOSS projects.
My post-paid plan had zero cost for the kids lines, actually, but it isn't a plan currently offered. That gets 500min/mo on each of the lines.
I do agree with you though. Plans are always changing, as are phones. When people tell me what phone they should get in six months I tell them to ask me again in six months. The same is true of plans. Last summer when I re-signed with T-Mo the value plans were the best option for my circumstances, but that could easily be different next time.
Still, $120/mo for unlimited voice and text on four lines, 2GB 3G data on two lines, 500MB 3G data on two lines, and unlimited 2G data on all lines is a VERY good deal. That would be hard to beat pre-paid assuming you needed all that data.
Unfortunately when it comes to physical infrastructure, redundancy is often just not possible.
That is simply a matter of priorities.
Moderately large bridges cost a few million dollars to build (we're talking about suspension bridges here - not the little ones that go over creeks). We probably could have replaced every suspension bridge in the US and built 5x as many more for much less than the cost of one of our recent wars, and this would have likely had a HUGE impact on the economy.
Sure, when you build a bigger highway more people drive on it, but there is a REASON they drive on it, and a benefit from them doing so. More transportation means less redundant industry, lower costs of living, more job portability, and so on.
Building new fighter jets involves high levels of project risk, huge amounts of money, and little economic benefit (unless you're actually at risk of being invaded). Sure, you will always need a military, but does it have to be 10X larger than the next 5 combined?
Building a new bridge involves almost no project risk. It costs money, but it is a solved problem. The only reason replacing bridges causes so much disruption is because we're running our infrastructure way over what it really is able to manage. If a Google query took 30 seconds to respond you can bet they'd be concerned about the costs of taking down hardware to upgrade it. However, Google is a sensible business and realizes that fast queries make them more money, so when they replace hardware the impact of outages is measured in microseconds.
The only problem with this approach is that it is usually more expensive (no linkage to family plan, etc). Agree that it is a good teaching approach, though - maybe that makes it worth it in the end.
However, I think there is still a need for a legislative solution. We don't need to regulate plan offerings or anything like that. Instead, just let customers put a limit on their bill, which can be as low as whatever number was advertised on the window (which will need to include all taxes/fees). If the provider delivers any service in excess of that amount, the service is free (this isn't a payment plan - they can't charge for the service ever). Providers can of course cut off service or not deliver optional services that aren't going to be paid for until the next billing cycle.
The basis for providing a service should be a conscious agreement between buyer and seller. Anything else is little more than a scam. If the buyer doesn't realize they're buying something, then the sale is null and void.
I've done a fair bit of comparison for home mailing purposes. For small packages USPS tends to be the best option. Once you start getting moderately large/heavy (nothing you can't pick up) UPS gets a lot cheaper. Also, if you care about full tracking UPS or Fedex can be cheaper - the cheapest USPS services don't get you full tracking, signature, etc. Oh, don't forget declared value either. For UPS it is free up to some value (maybe $100?), and for USPS it always costs money and I think you have to be at a higher class of service.
I usually check all three services anytime I ship - I can't say I consistently find any to always be cheaper. USPS tends to be best when you don't care about tracking and the box is small. For some intermediate cases Fedex does well, especially for light things that you want to track. UPS tends to do well for medium-sized packages in general.
There are a lot of other things to consider. If you're dealing with a party you don't trust signatures are a must, as is a declared value. Oh, and consider packaging as well. If you ship via a certified partner and pay them some token fee to do the packaging and declare the value and require a signature then you're basically zero-risk in terms of anything that could go wrong. The last time I sold a cell phone I did that - cost me all of maybe $10 more, but I didn't have to worry about complaints about broken/lost/etc, reversals, etc. When you're shipping to some individual and you aren't willing to eat the cost of fraud it is nice to basically have everything insured.
Uh, if you're shipping lead ingots all the way across the country the flat-rate boxes are probably the best deal. For just about anything else, not so much. Their main advantage is being flat-rate, and free standard-sized boxes. If you're paying the guy who is doing all the postage work it makes more sense as a result.
Then again, if you're running a business you can just get a postage mailing system with integrated scale/printer/etc and flat-rate has basically no advantage at all.
Telecom in the US has been full with the equivalent of city window washers for decades. It all comes down to a flawed trust system.
If I want to change services, I should call up the phone company and switch my services. However, that isn't as vulnerable to impulse buying, cold calling, and general scamming, so that approach simply won't do in the US.
The result is that if I run some kind of registered phone service I can basically send a bill to any telephone company in the country for any phone number in the country and get paid money. Then the poor sucker I scammed has to beg and plead with the phone company to try to reverse the charges while I retire to someplace in the Caribbean. There are phone numbers that charge you if you call them, international area codes that don't look like they're international, and so on.
SMS scams are just another example of this. Your kid texts some joke-of-the-day service and the parent gets a $40 bill. Sure, you can yell at the kid, but in what world are we supposed to allow children to sign contracts that are binding on their parents?
Yup, the nice thing about T-Mobile is that you can't get hit with per-MB charges, ever. The worst that can happen is you get throttled. That means you don't have to buy an extra-large plan "just in case" or check your data every other day.
Only Sprint offers a similar deal. Verizon and ATT just love the add-on charges. Back when I had Verizon I found myself having to over-spend on my plan just to avoid getting the surprise $600 phone bill. That is especially a big concern if you have kids. Sure, you can yell at them and it doesn't happen often, but it is cheaper to pay the extra $20/mo as "insurance." I resented this and am saving money and don't have to worry about the insurance at all...
Keep in mind you compared an unlimited data plan to a 3GB data plan. The 2GB data plan for T-Mobile is only $10/month.
Oh, I've found that T-Mobile has much better family plans as well. For $120/month you can have four phones with unlimited calling and 2GB data on two of them. For ATT that would be around $200/month. Oh, and it still is apples and oranges as T-Mo is now throwing in 500MB on all lines and NO overage fees (it just drops to 2G). ATT would charge $240/mo to bump the other two lines up to only 200MB and if you get one too many emails, say hello to per-MB pricing.
In my experience T-Mobile is way cheaper in just about every way these days. Their only issue is coverage, but even on vacation I haven't found it to be bad. I live in a suburb.
I don't think they support WiFi calling on phones they don't sell. It doesn't always work on modded phones either.
While I love T-Mobile I'm not sure I'd ever buy another phone from them. Their phones tend to be expensive (well, all the other major carriers are too). You can save quite a bit by buying from Google, or from some Asian supplier.
Bullet striations are about as reliable as a lie detector test.
You may as well consult an astrologist.
Maybe, but they'll convict you on it just the same. It isn't any different from any of the other forensic techniques dating back to the 80s and earlier, including many forms of DNA testing. The prosecutor doesn't care they work, they just care that the necessary precedents are in place so that they can send the defendant to jail.
The purpose of a prosecutor is to punish somebody for a crime as severely as possible. The purpose of a court is to ensure the prosecutor jumps through the legally-required hoops before doing so. Or, at least, that is what all the incentives promote.
Something others didn't mention is that this is SATELLITE data, not data from aerial photos.
When you look at Google Maps "satellite" view you are likely looking at a photo taken by a plane. Obviously it is much easier to get a high-resolution photo of a house from a plane a mile or two up than from a satellite 350 miles up.
Satellite photos have the advantage of being easier to acquire more regularly. The satellite flies over the country every day whether you need a photo or not. It will never be able to compete with a photo taken from a plane, let alone one taken from the ground. These are technologies that solve different problems.
There is definitely a use for regular civilian satellite images of the entire Earth's surface.
FOSS projects smaller than KDE manage to back up their data. All it requires is that somebody cares.
They have websites, they have mirrors, they have servers. They can afford backups. They likely have servers donated by companies that could back up those servers for them if asked. With git anybody can do a clone a day and rotate those clones, and those would be backups.
I'm involved in a much smaller FOSS project with a much smaller budget, and yet we spend thousands of dollars a year on hardware/hosting/etc and we do backups.
I believe a properly administered git mirror will do everything the GP wanted, except for testing the backups (which isn't git's problem) and the "2-sided crypto-hash compare" because, in my amateurishness, I don't know what that is.
The git mirror will overwrite the last backup with the next - it does not rotate media. Now, you can do a git clone to a new location every time and THAT might accomplish all the backup best-practices (assume you test).
You need to rotate your backups because you can't assume that you'll detect a problem with your data before your previous backup gets overwritten. What happens if you find a glitch that goes back a month?
The further back you want to go the more expensive things are, but in general you want to be able to go back at least a few days. For something as small as a git repository (even a big one) there is no reason not to rotate back at least a month or two.
More accurately the problem is that the hardware resources available to KDE are very limited and the KDE repository is one of the largest git repositories in the world.
I doubt it is more than a few GB per backup compressed. If they rotated 20 backups using daily/weekly/monthly rotations they'd have a week of dailies and a year of less frequent backups. That would cost them all of maybe a few bucks a month even with retail-cost solutions like Amazon S3 - WAY less than they must be paying for their servers. They'd only need backups going back a few days to save themselves from a rapidly-detected disaster.
Oh, and anybody could do a backup of KDE like this, assuming they allow anybody to clone their full repositories. A clone is a point-in-time snapshot, and then if you don't overwrite each snapshot with the next you have backups. It isn't like it takes special hardware to pull this off.
If you can backup a live database you can certainly backup a live git repository.
Stick it on zfs or btrfs, create snapshots, and then offline backup the snapshots. That's really easy to do, and atomic.
Or just do a git clone or create a bundle and back that up offline. That is definitely atomic. Oh sure, you won't get every last commit, but this is a BACKUP, not a mirror, and you're not going to rotate them every five minutes. If you have to resort to backups you'll lose some commits, but at least you won't lose your last decade of history or whatever.
Or you can offline the repository for a few minutes while you umount it, LVM snapshot it, and then remount it. That gives you a more defined recovery point for what it's worth, but at the cost of some downtime.
Lots of ways to backup a live repository. Any of them are better than not having backups. They can still have mirrors - they're useful as well, but they solve a different set of problems.
Yup, it all starts with defining the failure modes you want to protect against, and appreciating what techniques protect against what modes.
However, if you don't really appreciate all that stuff or can't be bothered to, then a safe default is rotational backup to offline media stored at a separate location from the device being backed up. There are plenty of services who make this sort of thing easy to do if you can't be bothered to get it right yourself - many are cloud-based and designed for clueless PC owners, and commercial solutions exist for Linux as well. There is no excuse for the maintainers of an FOSS project to not have an adequate backup solution for their source repositories.
Mirrors and RAID don't involve rotation and that makes them effective against only a small number of failure modes. Those who think they're "good enough" for something like this don't have good imaginations. I live with RAID-only only for things that I can reproduce (such as re-installing, re-ripping, etc), or things I don't mind losing (like hundreds of hours of MythTV recordings). For anything with actual value I do rotational backups uploaded to an offsite location, with periodic testing (in particular anytime my backup software is updated). During testing I carefully peruse my backed up data so I have a pretty good appreciation for what I will and won't get back in the event of a disaster. These days decent backup is pretty cheap - there is no excuse for not having it where it matters.
Now, improving uptime and reducing downtimes is important, but it is not what a backup does.
Well, a backup does contribute to reducing downtimes, albeit not so much as RAID/etc. Compared to doing a full reinstall/reconfiguration restoring from backup is likely to be much faster. That is why backup can be useful even on systems that do not contain unreproducible data. There are other strategies that have other advantages (like automatic builds/etc) which are also effective if there is no data involved.
I do agree that the primary purpose of a backup is to prevent the loss of data in as many failure modes as possible/practical. Mirrors are definitely not backups (or at least, not very good ones - there is a continuum when it comes to backups, just as there is a continuum when it comes to disasters).
Tend to agree on this one, and it is the sort of thing that has been running companies into the ground lately (usually due to too much MBA-think - they forget they're supposed to manage people and not just spreadsheets).
Employees who just check out is something that is hard to measure, and yet it can cost a company a fortune. Then you get tons of training programs encouraging employees to speak up about problems and try to make the workplace happier, and they typically neglect the underlying issues that are causing those problems in the first place so they have no effect.
Tend to agree, and I'm an airplane buff.
If an airport receives a significant volume of traffic then it should have a tower (and that need not be more than an office with a reasonable view of the field and a radio - it doesn't have to be some huge complex with supervisors, etc).
If an airport does not receive a significant volume of traffic it should be uncontrolled, and aircraft procedures already enable pilots to handle this situation. You can even fly at such an airport when visibility is poor - the local air traffic control simply creates a bubble around the airport for a specified period of time so that there isn't a conflict (though this means waiting for anybody who wants to takeoff/land since only one aircraft can hold such a clearance at a time).
The interstate commerce clause has been stretched so far it will cover anything.
The prosecutor will simply argue that because you made your own howitzer in your basement in Kansas you didn't end up buying one from a licensed howitzer manufacturer in Vermont, and therefore you had an effect on interstate commerce. Never mind that you couldn't have legally bought that howitzer from Vermont anyway.
These kinds of arguments have been used to apply Federal regulations to virtually everything. After all, you could apply that sort of thing to anything. If I watch a movie across the street I satisfy entertainment urges that might have otherwise have required me to drive across a state line to the nearest beach, so that movie theater is affecting interstate commerce, and so on...
Not only that, it was done illegally by bypassing the cabinet.
How can a President "illegally" bypass the cabinet? The president isn't constitutionally even required to have a cabinet in the first place, as far as I can see. I don't buy into the whole "unitary executive means the president can just ignore the law" bit, but the President is, well, the President. Ultimately he's the one responsible for upholding the law and conducting the affairs of the executive branch, and everybody else is just there to help out.
If I missed some clause in the constitution by all means let me know.
I do tend to think that wages make a difference. I also think one problem lies in keeping them in the profession as opposed to going into a more lucrative field. I haven't seen much data to back that up though.
I think half the problem is how the pay is structured. In US public schools teacher pay is based on seniority, full stop.
The art teacher who has been working for 15 years gets paid the same as the math/science teacher who has been working for 15 years. The teacher who sticks a page of notes on a projector screen and tells their students to copy it gets paid the same as the teacher who uses the Socratic method and whose students perform 3x better.
As a result, if you have an extra $10M to spend on wages the only way to spend it is to take the money, divide it by the number of teachers, apply some curve so that more goes to those who have been there the longest, and then give everybody an extra $100-1000 dollars. No private company facing a crisis would ever spend money that way - they'd target how the money could best be used and they'd spend the same amount of money in a much smaller area and get a dramatic result.
If teachers were paid market wages per subject then you could actually hire great new math/science teachers without having to pay the music teacher with 25 years seniority $500k/yr. You can't expect somebody with an MS in a physical science to start out at $20k/yr and pay their dues as they work their way up. The only people you hire that way are those who couldn't get a job elsewhere (mixed in with a few salt-of-the-earth types who would find a way to teach even if you didn't pay them at all).
Pay is just the start though - there is quite a bit that needs to be reformed. The other big problem is that you can't really force kids to learn - that is a culture thing that starts at home/etc. You can have Feyman teaching your physics class and it won't do any good for the kid who is just there to hang out with friends and do his time. In many ways I think the pay problem is just a symptom of larger societal problems. Why work hard to get A's in physics if you can get a much higher salary by being in the right fraternity and landing a job in some bank?
He was obviously being sarcastic.
The US military is really good at blowing things up - nobody does it better. As long as that's how you use them things go just fine. The problem is that when your tool is blowing things up and your goal is to establish a government with liberty and justice for all, things don't always work out well.
The locals need to want true democracy before you can try to establish it. If this were about being the French and blockading the British so that the US revolutionaries could finish setting up a democracy that would be one thing. However, what this is more likely to turn into is providing for one particular religious faction so that they can wipe out all the other ones, then hold an election where a new dictator for life is selected. Maybe it will go slightly better, but there is a huge culture difference between the Middle East and the US/Europe. I guess they just haven't had enough time to get tired of all the Salem Witch Trials and Inquisitions and they still want to see if they work.