NTFS supports a weird "copy-before-write" mode. It's not the same as copy-on-write as it copies the data to the shadow area before writing the new data to the file. This allows it to keep the shadow area and/or snapshots on separate media to the live copy.
What about all of us who live in older apartment buildings or terrace houses that don't have off-street parking? We have council parking permits allowing us to park on the street, we don't have a place to plug in and charge overnight.
I mostly remember the one in the film Dark City. Seems like a strange idea to be honest - the Japanese style of buying a ticket for the meal from a vending machine and then giving it to the person behind the counter makes more sense.
I don't think the GPL stops them doing this. They aren't stopping you from redistributing GPL software, they're just saying that if you redistribute the software, they won't give you future updates. GPL doesn't require supplying future updates, it just says that you must provide an offer of source with binaries, and can't restrict redistribution of source/binaries. It looks like they've found another way to follow the letter of the GPL without following the spirit of it.
So someone who buys some version of grsecurity can redistributes it, and the people they redistribute it from can also redistribute it. The vendor is free to refuse to do business with all these people. But it only takes one customer who no longer cares about receiving future updates to release all the versions they've received, or potentially one rogue employee who doesn't want their employer to receive future grsecurity updates.
The emergency call establishment doesn't have a number in in. You dial 112, 911 or whatever and the phone just says "get me emergency". It can't specify any more than that. The network decides how to route the emergency call. So yes, you can make a custom SIM that allows you to dial an arbitrary number for free, as long as you're happy for that number to just connect to emergency services rather than being routed as it usually would be.
Emergency numbers are actually recognised by the phone itself, and it initiates a different call establishment sequence for emergency calls. This allows the network to give emergency calls priority, and allow emergency calls without a SIM card, etc. All GSM-compliant phones (anything with WCDMA 3G or LTE) will recognise 112 as well as any additional emergency numbers programmed in the firmware or the SIM card. Phones sold in the US will recognise at least 112 and 911, phones sold in Australia recognise 112, 000 and usually 911 as well, phones sold in Hong Kong and UK recognise 112 and 999 and possibly other emergency numbers.
So while the story is wrong, you're wrong too. Unless you're using a Sprint-style CDMA phone, 112 will always work as an emergency number, and the other numbers that work depend on your phone and SIM card. If it's a US phone and/or SIM card, 911 will definitely work, but if you're using a phone and SIM card from overseas, that may not be the case. It has nothing to do with a courtesy to European visitors in cities.
The story probably got started early in US GSM rollout before the carriers thought to program 911 as an alternate emergency number in phones and SIM cards. If the phone/SIM card don't recognise 911 as an emergency number, it will be established as a normal call, and routed by the network. It will still go to the same destination US but it won't get priority the way an emergency call establishment sequence will. The same was true in Australia before 2000 - most phones/SIM cards didn't recognise 000 as an emergency number.
Yeah, it's an "electric multiple unit trainset" if you want to be specific, but that can be shortened to "EMU train" or just "EMU", and it's still a "trainset" or "train" if you don't want to specify the type.
What? He made money with investment banking and arguably won the lottery with his investments, but he didn't come from a privileged background. He grew up with a single dad below the poverty line.
That was Symantec issuing google.com certificates for their test environment, not WoSign. The thing everyone jumped on WoSign for was doing a customer a favour. Some significant Australian customer wasn't ready for SHA1 certificates being phased out and asked if WoSign could help them out. WoSign issued back-dated SHA1 certificates for the customer.
IMO, Symantec was worse yet they haven't been punished anywhere near as hard, and the timing seemed rather "convenient" in that it coincided with the launch of Let's Encrypt. Free StartCom certificates won't be trusted any more. Who else are you going to get free certificates from now, buddy?
I'm not so sure about that. I think it's becoming possible to write the modern dialect of C++ without ever learning most of the old cruft. You probably do need someone around who knows the dusty corners, but largely you can ignore them.
It's at the point where there's a lot to learn before you'll know what the right tool for the job is. You could be looking at a perfect use case for std::enable_if, or perhaps you could flatten that thing at compile time with a recursive variadic template. Did you know && is used for both universal references in templates and rvalue references? When should you use std::move vs std::forward? Can you effectively use std::promise? A lot of the new library features won't hurt you if you ignore them, you just might not be getting the most out of the tools you have available. But you do have to be very careful with rvalue references or you can end up inadvertently hurting performance. It's always the way though - if you give someone a gun, it's up to them not to shoot themselves in the foot.
I don't think you're quite right about C++ - I think we've finally got to the point where it's worth using C++. For years, C++ compilers were absolutely atrocious, and standard library implementations were even worse. C++ compilers finally became usable with GCC 4 and MSVC 2005. With these compilers, most of the useful C++ language features finally worked properly.
Then you had the problem with the standards committee arguing for years, but they've finally got a process that mostly works, and we're getting our updated standards on a three-year cycle. Implementations have mostly caught up with C++14, although support for things like variable templates is still weak. C++11 added a whole pile of stuff that was sorely needed. In particular, variadic templates, constexpr, lambdas, and the concurrency primitives have reduced the need for writing a lot of glue for every project.
Qt isn't the only extended C++ library. There's also Microsoft's MFC/ATL, although that's stagnated, and of course Boost. No sane person would say Boost is perfect, but it solves a lot of common problems and it's well-understood, so people understand its issues and know how to work around it.
The biggest problem with C++ now is that the size of it makes it a steep learning curve. There's a lot to get your head around if you want to use it effectively, and it gives you lots of ways to shoot yourself in the foot. Java has similar issues with the size of its standard library, but C++ allows you to do all sorts of things where other languages try to save you from yourself.
But it wasn't really expected to at the time for the most part. "DOS" in those days often just meant a library of disk access routines (that's what Apple DOS 3.3 for Apple II is, as well as Commodore DOS and the rest of them).
I remember them being advertised heavily in Australia around 2002, around the same time Crazy Frog ringtones were selling. I don't remember anyone ever actually paying for them though. Just not worth the money.
That doesn't even make sense. FSF doesn't need copyright assignment to distribute LLVM under the terms of GPL. The BSD-like license allows for that already. They never promised to assign copyright of all of LLVM to the FSF. That wouldn't make sense because it would prevent anyone from distributing it under the less restrictive BSD-like license.
The patch to allow GCC to easily use LLVM as a backend needs to be GPL-licensed because it's a derivative of GCC which is itself GPL-licensed. Assigning copyright on that patch to FSF would have allowed integration into GCC. However, the FSF are opposed to making GCC modular in any way. They think this will be a gateway to integrating GCC with non-free (as in beards) development environments (yes this also makes it hard to integrate into FSF-approved development environments, but that's their choice).
But clang/LLVM has been the best thing for GCC in years. It's provided real competition in the compiler space, and without that GCC was just stagnating and getting buggier. C++ support, error messages, compile speed and plenty of other stuff has improved enormously since clang/LLVM has been on the scene.
There was a script in Steam for Linux that did just that and wiped out some home directories when the variable managed to be empty. I didn't hear about it happening to anyone on their first day of work.
I'm not saying they should use obsolete components, just that they should allow components to be replaced. With your analogy, I can replace the battery, spark plugs, injectors, ignition units and ECUs on my car pretty easily. I don't need to cut the bonnet open and glue it shut when I'm done.
So you've forgotten iMacs where you could just flip open the plastic door and replace the RAM, with no tools and no consumable, and no chance of fucking up the process of cutting the display out?
So the RAM/CPU are socketed and theoretically replaceable/upgradeable, but doing so requires CUTTING THE MONITOR AWAY. Great improvement there Apple. I remember when you could upgrade the RAM on an iMac by opening a plastic panel, with no tools or consumables required. On the LC575 (an all-in-one from the mid '90s) the back panel popped off for you to install/upgrade RAM, PDS and comm slot cards, and the internal SCSI hard disk was on a sled that just slid out.
Many high-performance 10Gbps and 40Gbps Ethernet cards can offload RDMA protocol. The high-level API uses Infiniband verbs for sending/receiving messages. It's a lot more efficient than dealing with TCP.
I work in finance, and I have big datasets to crunch. But for that I have multi-socket rack-mount servers running Linux. My desktop is a single-scoket Xeon running Windows, it doesn't do heavy lifting. Support for SMB over RDMA and a reliable filesystem look like nice things to have, but allowing four sockets isn't a big deal.
I can still catch an Australian domestic flight without showing ID, without taking my shoes off, and wife a six-pack of beer in my carry-on luggage. I can still travel to/from China, Hong Kong, Japan, Korea, Philippines, Viet Nam, hell the entire region, without this bullshit about entire travel histories and handing over social media details. How much of the world have you actually seen?
Your post-9/11 bullshit is already keeping people out of the US. How about you do something about it? Land of the free, home of the brave my arse. Home of the scared shitless. Do you even remember the '80s? US used to carry on about how the USSR was oppressive because you needed papers to travel. Travelling in the US is worse than that now. You need to present papers, and take your shoes off, and not lock your bag, and not take any liquids with you, and so on. If your name is similar to a name on a secret list, you're denied the right to fly and there's jack shit you can do about it. US used to carry on about DDR's mass surveillance where they were paying everyone to spy on everyone else as though it was some great evil as well. Yet you now wiretap all domestic communications. You've got your secret prisons where you disappear people, you've got your leaked torture incidents (although you seem to be getting better at covering that up), you punish people for embarrassing the government rather than actually addressing the issues that make the leaks embarrassing. It's just not worth visiting the US. You get treated like a criminal at the border, and you know you're visiting the most hypocritical country on earth.
No it wasn't. The original estimates were based on Soviet propaganda. Auschwitz officers were tried for the murder of more people than we now believe were processed by the camp. I'm not saying the holocaust didn't happen or it wasn't terrible. It's just that the initial numbers thrown around were very high estimates based largely on Soviet propaganda that were revised down.
Slashdot Mirror
NTFS supports a weird "copy-before-write" mode. It's not the same as copy-on-write as it copies the data to the shadow area before writing the new data to the file. This allows it to keep the shadow area and/or snapshots on separate media to the live copy.
What about all of us who live in older apartment buildings or terrace houses that don't have off-street parking? We have council parking permits allowing us to park on the street, we don't have a place to plug in and charge overnight.
I mostly remember the one in the film Dark City. Seems like a strange idea to be honest - the Japanese style of buying a ticket for the meal from a vending machine and then giving it to the person behind the counter makes more sense.
I don't think the GPL stops them doing this. They aren't stopping you from redistributing GPL software, they're just saying that if you redistribute the software, they won't give you future updates. GPL doesn't require supplying future updates, it just says that you must provide an offer of source with binaries, and can't restrict redistribution of source/binaries. It looks like they've found another way to follow the letter of the GPL without following the spirit of it.
So someone who buys some version of grsecurity can redistributes it, and the people they redistribute it from can also redistribute it. The vendor is free to refuse to do business with all these people. But it only takes one customer who no longer cares about receiving future updates to release all the versions they've received, or potentially one rogue employee who doesn't want their employer to receive future grsecurity updates.
The emergency call establishment doesn't have a number in in. You dial 112, 911 or whatever and the phone just says "get me emergency". It can't specify any more than that. The network decides how to route the emergency call. So yes, you can make a custom SIM that allows you to dial an arbitrary number for free, as long as you're happy for that number to just connect to emergency services rather than being routed as it usually would be.
Emergency numbers are actually recognised by the phone itself, and it initiates a different call establishment sequence for emergency calls. This allows the network to give emergency calls priority, and allow emergency calls without a SIM card, etc. All GSM-compliant phones (anything with WCDMA 3G or LTE) will recognise 112 as well as any additional emergency numbers programmed in the firmware or the SIM card. Phones sold in the US will recognise at least 112 and 911, phones sold in Australia recognise 112, 000 and usually 911 as well, phones sold in Hong Kong and UK recognise 112 and 999 and possibly other emergency numbers.
So while the story is wrong, you're wrong too. Unless you're using a Sprint-style CDMA phone, 112 will always work as an emergency number, and the other numbers that work depend on your phone and SIM card. If it's a US phone and/or SIM card, 911 will definitely work, but if you're using a phone and SIM card from overseas, that may not be the case. It has nothing to do with a courtesy to European visitors in cities.
The story probably got started early in US GSM rollout before the carriers thought to program 911 as an alternate emergency number in phones and SIM cards. If the phone/SIM card don't recognise 911 as an emergency number, it will be established as a normal call, and routed by the network. It will still go to the same destination US but it won't get priority the way an emergency call establishment sequence will. The same was true in Australia before 2000 - most phones/SIM cards didn't recognise 000 as an emergency number.
Yeah, it's an "electric multiple unit trainset" if you want to be specific, but that can be shortened to "EMU train" or just "EMU", and it's still a "trainset" or "train" if you don't want to specify the type.
What? He made money with investment banking and arguably won the lottery with his investments, but he didn't come from a privileged background. He grew up with a single dad below the poverty line.
That was Symantec issuing google.com certificates for their test environment, not WoSign. The thing everyone jumped on WoSign for was doing a customer a favour. Some significant Australian customer wasn't ready for SHA1 certificates being phased out and asked if WoSign could help them out. WoSign issued back-dated SHA1 certificates for the customer.
IMO, Symantec was worse yet they haven't been punished anywhere near as hard, and the timing seemed rather "convenient" in that it coincided with the launch of Let's Encrypt. Free StartCom certificates won't be trusted any more. Who else are you going to get free certificates from now, buddy?
That would be Bruce Breckets
you're after.
It's at the point where there's a lot to learn before you'll know what the right tool for the job is. You could be looking at a perfect use case for std::enable_if, or perhaps you could flatten that thing at compile time with a recursive variadic template. Did you know && is used for both universal references in templates and rvalue references? When should you use std::move vs std::forward? Can you effectively use std::promise? A lot of the new library features won't hurt you if you ignore them, you just might not be getting the most out of the tools you have available. But you do have to be very careful with rvalue references or you can end up inadvertently hurting performance. It's always the way though - if you give someone a gun, it's up to them not to shoot themselves in the foot.
I don't think you're quite right about C++ - I think we've finally got to the point where it's worth using C++. For years, C++ compilers were absolutely atrocious, and standard library implementations were even worse. C++ compilers finally became usable with GCC 4 and MSVC 2005. With these compilers, most of the useful C++ language features finally worked properly.
Then you had the problem with the standards committee arguing for years, but they've finally got a process that mostly works, and we're getting our updated standards on a three-year cycle. Implementations have mostly caught up with C++14, although support for things like variable templates is still weak. C++11 added a whole pile of stuff that was sorely needed. In particular, variadic templates, constexpr, lambdas, and the concurrency primitives have reduced the need for writing a lot of glue for every project.
Qt isn't the only extended C++ library. There's also Microsoft's MFC/ATL, although that's stagnated, and of course Boost. No sane person would say Boost is perfect, but it solves a lot of common problems and it's well-understood, so people understand its issues and know how to work around it.
The biggest problem with C++ now is that the size of it makes it a steep learning curve. There's a lot to get your head around if you want to use it effectively, and it gives you lots of ways to shoot yourself in the foot. Java has similar issues with the size of its standard library, but C++ allows you to do all sorts of things where other languages try to save you from yourself.
But it wasn't really expected to at the time for the most part. "DOS" in those days often just meant a library of disk access routines (that's what Apple DOS 3.3 for Apple II is, as well as Commodore DOS and the rest of them).
I remember them being advertised heavily in Australia around 2002, around the same time Crazy Frog ringtones were selling. I don't remember anyone ever actually paying for them though. Just not worth the money.
That doesn't even make sense. FSF doesn't need copyright assignment to distribute LLVM under the terms of GPL. The BSD-like license allows for that already. They never promised to assign copyright of all of LLVM to the FSF. That wouldn't make sense because it would prevent anyone from distributing it under the less restrictive BSD-like license.
The patch to allow GCC to easily use LLVM as a backend needs to be GPL-licensed because it's a derivative of GCC which is itself GPL-licensed. Assigning copyright on that patch to FSF would have allowed integration into GCC. However, the FSF are opposed to making GCC modular in any way. They think this will be a gateway to integrating GCC with non-free (as in beards) development environments (yes this also makes it hard to integrate into FSF-approved development environments, but that's their choice).
But clang/LLVM has been the best thing for GCC in years. It's provided real competition in the compiler space, and without that GCC was just stagnating and getting buggier. C++ support, error messages, compile speed and plenty of other stuff has improved enormously since clang/LLVM has been on the scene.
There was a script in Steam for Linux that did just that and wiped out some home directories when the variable managed to be empty. I didn't hear about it happening to anyone on their first day of work.
I'm not saying they should use obsolete components, just that they should allow components to be replaced. With your analogy, I can replace the battery, spark plugs, injectors, ignition units and ECUs on my car pretty easily. I don't need to cut the bonnet open and glue it shut when I'm done.
So you've forgotten iMacs where you could just flip open the plastic door and replace the RAM, with no tools and no consumable, and no chance of fucking up the process of cutting the display out?
So the RAM/CPU are socketed and theoretically replaceable/upgradeable, but doing so requires CUTTING THE MONITOR AWAY. Great improvement there Apple. I remember when you could upgrade the RAM on an iMac by opening a plastic panel, with no tools or consumables required. On the LC575 (an all-in-one from the mid '90s) the back panel popped off for you to install/upgrade RAM, PDS and comm slot cards, and the internal SCSI hard disk was on a sled that just slid out.
Suspension in the Prius is horrible. If you want a Toyota hybrid, you're better off with the Camry or one of the Lexus hybrids.
Many high-performance 10Gbps and 40Gbps Ethernet cards can offload RDMA protocol. The high-level API uses Infiniband verbs for sending/receiving messages. It's a lot more efficient than dealing with TCP.
I work in finance, and I have big datasets to crunch. But for that I have multi-socket rack-mount servers running Linux. My desktop is a single-scoket Xeon running Windows, it doesn't do heavy lifting. Support for SMB over RDMA and a reliable filesystem look like nice things to have, but allowing four sockets isn't a big deal.
I can still catch an Australian domestic flight without showing ID, without taking my shoes off, and wife a six-pack of beer in my carry-on luggage. I can still travel to/from China, Hong Kong, Japan, Korea, Philippines, Viet Nam, hell the entire region, without this bullshit about entire travel histories and handing over social media details. How much of the world have you actually seen?
Your post-9/11 bullshit is already keeping people out of the US. How about you do something about it? Land of the free, home of the brave my arse. Home of the scared shitless. Do you even remember the '80s? US used to carry on about how the USSR was oppressive because you needed papers to travel. Travelling in the US is worse than that now. You need to present papers, and take your shoes off, and not lock your bag, and not take any liquids with you, and so on. If your name is similar to a name on a secret list, you're denied the right to fly and there's jack shit you can do about it. US used to carry on about DDR's mass surveillance where they were paying everyone to spy on everyone else as though it was some great evil as well. Yet you now wiretap all domestic communications. You've got your secret prisons where you disappear people, you've got your leaked torture incidents (although you seem to be getting better at covering that up), you punish people for embarrassing the government rather than actually addressing the issues that make the leaks embarrassing. It's just not worth visiting the US. You get treated like a criminal at the border, and you know you're visiting the most hypocritical country on earth.
No it wasn't. The original estimates were based on Soviet propaganda. Auschwitz officers were tried for the murder of more people than we now believe were processed by the camp. I'm not saying the holocaust didn't happen or it wasn't terrible. It's just that the initial numbers thrown around were very high estimates based largely on Soviet propaganda that were revised down.