Exactly. Compare sales of fleshlights to sales of vibrators / dildos. Women are far more likely currently to be buying machines for sexual purposes. Why would the sexbot market be any different?
That depends a lot on what you mean by programming, and that's where this kind of discussion falls down. Anyone who has used an office suite has done, by hand, things that could be trivially automated. It's not a huge step to get from complete manual automation to writing some simple scripts that make life easier. It's a much bigger jump to get from there to being able to write complex and maintainable applications with sensible algorithms, but in a lot of cases that naive n^2 algorithm will do fine because you're only giving it small values of n and even a slow modern computer will do it a lot faster than the operator (which is the goal).
It's also worth remembering that most things that people will need to program are DSLs, not general-purpose languages. Before he retired, my stepfather was head groundskeeper at a golf course. The system that they used for controlling the irrigation system was a Turing-complete (visual) DSL. It had loops, conditionals, and maintained arbitrary state. It took input from a variety of sensors (motion sensors to track people, sunlight and moisture sensors around the place) and controlled a load of valves for the sprinklers. This is the kind of programming that someone in a blue-collar job has to do today. The programs are simple enough that maintainability is not a serious concern, but they do need debugging sometimes (get the conditions wrong and you accidentally get some quite damp people!).
Check again. Nothing in the natural sciences is ever proven. Theories are hypotheses that have been subject to attempts at falsification (which implies that they are falsifiable hypotheses), but which have so far survived all tests. Occasionally, long-lasting hypotheses are disproven. Newtons Laws are an example (and the main reason why the scientific establishment no longer uses the term 'law' for these hypotheses), where Arthur Eddington showed experimental evidence that falsified them. Of course, just because a hypothesis has been falsified doesn't mean that it's no longer useful. We still teach Newtonian physics because it's close enough to correct to be useful for things of a size and moving at speeds that people will regularly encounter (though not enough for GPS to work) and the mathematics behind it are a lot simpler than in theories that corrected for the observations.
There are several ways around this. The simplest is for the process to run setuid root, open the listening socket as the first thing, then setuid() to the correct user. That's not ideal, because it means config file parsing as root. The second option is to have a simple setuid root binary that opens the listening socket on behalf of the untrusted program and passes it back over a UNIX domain socket. This needs some careful design to avoid confused deputy attacks. As an administrator, you can set your firewall to forward connections on port 80 to another port. This is a bit dangerous on multi-user systems because, if a local user can crash apache, they can race it to listen on the forwarded port and then intercept all inbound HTTP traffic. inetd can also be configured to forward connections from port 80 to another process or pass the accepted socket over a UNIX-domain socket, which allows it to pass the connection only to a process that can open a UNIX domain socket in a specific location (e.g. a directory owned by the Apache user).
Finally, the simplest and least error-prone solution: MAC policies (e.g. SELinux, TrustedBSD, and so on) can be configured to permit a specific application, running as a specific user, to open listening sockets on specified privileged ports. The httpd binary running as the apache user may listen on port 80, but no other program running as the appache user and no other using running the httpd program may do so.
In short, while this is a problem with classic UNIX, it's one that was solved at least a decade ago.
What sort of degree does an expert in "thinking logically" have? What department of the University are they found in? And if your curriculum is heavy on Logic, what is your most likely major?
What makes you think that? Recent benchmarks show llvm/clang well behind across the board. At least it compiles faster
Recent benchmarks? They're comparing LLVM 3.5 (over a year old) with GCC 4.9 (over a year old). With the pace of current development (in both - having some competition has been very good for GCC), those are not recent, they're archaic. Oh, and don't overestimate the benefit of faster compilation when it comes to developer productivity, particularly for C++.
That is precisely the killer advantage of GPL over BSD license. There is no shame whatsoever in porting code if it is good. Rather, it is the professional thing to do. However it is a gross exaggeration to say the GCC team does only that.
The codebases are sufficiently different that it's not really feasible to port code over (unless you just lower GIMPLE to LLVM IR and plug LLVM in as a GCC backend, which is what LLVM used to do and how it was offered to the FSF originally. Fortunately for the rest of the world, the GCC community turned it down). Ideas and algorithms can't be copyrighted and these flow in both directions.
The killer advantage of the BSDL is that people don't have to buy into your ideology to contribute. Microsoft and Azul, for example, are both contributing a lot to improving the garbage collection infrastructure in LLVM (GCC has nothing comparable and the GCC attempt at a JIT involved spitting out a stream of assembly that was run through gas, because GCC doesn't have an integrated assembler, so would be very hard to adapt). In Microsoft's case, they're integrating it into their MIT-licensed CLR implementation. Azul is integrating it into their proprietary JVM. Apple has contributed a lot in this area as well, and are using it in their LGPL'd JavaScriptCore. None of these licenses would permit incorporating GPL'd code, so no contributions from these vendors would appear in a GPL'd project.
The project that's suffering most from Clang is not GCC, it's EDG. They've been selling a proprietary C/C++ front end since 1988, which is used by a lot of vendors. Those vendors are increasingly realising that if they invested half of the EDG license fees in improving Clang, they could be completely independent. The BSDL project is killing proprietary competitors that thrived while the GPL project was dominant. What does that tell you about the relative merits of the two licenses for promoting free software?
For the companies that I've talked to, the issue isn't so much systemd (which does solve real problems), it's the attitude of vendors like RedHat to their customers. People with large deployments don't like core parts changing without consultation and without a migration path carefully laid out in advance. They also don't like seeing stuff pushed before it's fully baked, especially core infrastructure.
I don't have any figures, but from the FreeBSD side we see an influx of new users every time Pottering releases something new. The stream of new switchers escaping PulseAudio had just about dried up, but systemd has given a new set.
Most of the lack of environmental restriction in China does not impact the rest of the world, it just results in tracts of land and rivers within China becoming poisoned. The reason the US and Europe have regulations that (try to) prevent this is that people got fed up with factories destroying their locale for a quick profit. This kind of environmental regulation is easy to sell ('the government wants to stop factories polluting the river I live on? Great!'). Regulation of CFCs was a lot harder to push through and required a lot of education about the effects of the eroding ozone layer before it became possible ('the government wants to ban stuff I use that's causing some nebulous and invisible damage? Bah! Let the free market sort it out!'). Greenhouse gas emissions are still in the latter category and their effect is far harder to measure because it's even more diffuse - increasing the energy in a chaotic system does not give easily predictable outcomes, though the consensus generally is that a species that depends on a fragile equilibrium point in the system being maintained should probably care.
Coal burning was relatively rare until quite recently in history. Coal needed to be mined, wood was just there and charcoal was relatively easy to produce. Coal only made economic sense once you started to have steam engines requiring hotter fires and increased demand from factories (at which point industrialised coal mining pushed the cost of coal down and made it economically feasible for home heating).
Decay also produces a lot of methane as a byproduct, whereas combustion produces relatively little. This means that the total contribution to the greenhouse effect is less if you burn a tree than if you allow it to rot (though it will be condensed into a shorter time).
They've been coming for ages, but mostly we'd been keeping the best ones. Now we've introduced US-style visa stupidity so after we've educated them we promptly deport them before they can contribute money to the economy. It used to be relatively easy to transition from a student visa to a work visa (to the extent that some universities got into trouble for sponsoring people for student visas, who never turned up to lectures and then applied for a work visa a bit later). Now it's very hard and often requires returning home and applying for the work visa from there.
Yup, go to the USA, learn skills from US experts, but no contributing back to the US economy for you! You have to go back to wherever you came from and contribute to their economy with the skills that you've acquired. We've just instituted this kind of stupidity in the UK too.
I've not used Access since it ran on Windows 3.11, but I recently tried OpenOffice / LibreOffice Base and was amazed. Their form builder is incredible. Even simple joins in the UI require writing SQL by hand. How anyone can ship that as a GUI database tool, I have no idea.
GIMP and Photoshop both have pretty terrible UIs. If you've spent a lot of time and effort becoming comfortable with one, then having to switch to the other is painful. At least a bad UI that you're familiar with means that you've learned to work around its brokenness - an unfamiliar bad UI is far more painful than a familiar one.
You know that Bing Ads is one of the largest ad networks around, right?
Actually, I had no idea - I've never seen one. Apparently they have 15.6% of the US market share, though it's not clear how much they have worldwide. Thanks for the info.
I've found that just sticking the city name in the search works a lot better than enabling any kind of IP-based geolocation for that (especially if I'm in a hotel and using a VPN, but even without that, IP geolocation accuracy varies hugely).
That, and the fact that we only classified cancer as a thing quite recently. No one was diagnosed with cancer a couple of hundred years ago, but that doesn't mean that no one died of it.
Exactly. With the exception of Microsoft (which sells software, yet still doesn't have a great track record, especially with the Windows 10 fiasco), all of the listed companies have business models that rely on collecting as much information as they possibly can from their users (not to be confused with their customers). If you want to resist surveillance, then don't buy into large centralised communication systems.
Is the connecting end resolvable? If it isn't, then you may be waiting for the lookup to time out (it has a low timeout, but it can add a noticeable amount of delay).
Actually, it isn't. The Linux Foundation is funding the effort, and it's mostly Debian people leading it, but they're working on a variety of projects (including FreeBSD!), not just Debian.
The problem with DuckDuckGo is that it doesn't take into account location
That's a configuration option (and you can look at the cookie contents to validate that it will be the same for anyone with that option). I generally leave it turned off, because most searches that I do are not geographical in nature (e.g. I want information about the C++ standard, not about the UK edition of the C++ standard).
at all times Google's result were superior to Altavista
That's a highly subjective judgement and likely to depend a lot on the search terms. Most of the time, either both or neither found what I was looking for. Towards the end, AltaVista largely succumbed to spam, but around 2000 there wasn't much in it.
Not only was the Google ranking algorithm superior, the Altavista crawl fell way behind the Google crawl, so often the page you were looking for wasn't even in the Altavista index.
Until Google moved away from MapReduce, their results also lagged a long time behind. Searching for anything in today's news wouldn't find any relevant results with Google until long after AltaVista's demise.
The 'fast' thing really can't be overstated. By the time Google launched, AltaVista's search page had become huge, to the extent that it took about 30 seconds to load on a 28.8Kb/s MODEM (the fastest that mine could connect at given the line quality, though on paper it could do 56Kb/s). Google took well under 5 seconds (not because Google devs were clever and actively aimed for this, quite the reverse: they didn't initially have anyone good at HTML/CSS stuff, so produced the simplest page that worked).
I remember the search results on Google being worse than AltaVista, but getting them so much faster that I could start loading the first 3-4 before AltaVista showed me anything. Occasionally I'd go back to AltaVista if Google failed. A few years later, Google fucked up their UI enough to make me switch to DuckDuckGo.
The question is why people should care about Apple being in the consortium. The answer is that people want to sell things (e.g. streaming services) to people with mobile devices, and these may depend on vendor support for the codecs that they use (on iOS, you can't easily get hardware acceleration unless you use Apple's AVFoundation framework). Whether this is actually important depends on whether or not that market is one that you care about.
Slashdot Mirror
Exactly. Compare sales of fleshlights to sales of vibrators / dildos. Women are far more likely currently to be buying machines for sexual purposes. Why would the sexbot market be any different?
That depends a lot on what you mean by programming, and that's where this kind of discussion falls down. Anyone who has used an office suite has done, by hand, things that could be trivially automated. It's not a huge step to get from complete manual automation to writing some simple scripts that make life easier. It's a much bigger jump to get from there to being able to write complex and maintainable applications with sensible algorithms, but in a lot of cases that naive n^2 algorithm will do fine because you're only giving it small values of n and even a slow modern computer will do it a lot faster than the operator (which is the goal).
It's also worth remembering that most things that people will need to program are DSLs, not general-purpose languages. Before he retired, my stepfather was head groundskeeper at a golf course. The system that they used for controlling the irrigation system was a Turing-complete (visual) DSL. It had loops, conditionals, and maintained arbitrary state. It took input from a variety of sensors (motion sensors to track people, sunlight and moisture sensors around the place) and controlled a load of valves for the sprinklers. This is the kind of programming that someone in a blue-collar job has to do today. The programs are simple enough that maintainability is not a serious concern, but they do need debugging sometimes (get the conditions wrong and you accidentally get some quite damp people!).
Check again. Nothing in the natural sciences is ever proven. Theories are hypotheses that have been subject to attempts at falsification (which implies that they are falsifiable hypotheses), but which have so far survived all tests. Occasionally, long-lasting hypotheses are disproven. Newtons Laws are an example (and the main reason why the scientific establishment no longer uses the term 'law' for these hypotheses), where Arthur Eddington showed experimental evidence that falsified them. Of course, just because a hypothesis has been falsified doesn't mean that it's no longer useful. We still teach Newtonian physics because it's close enough to correct to be useful for things of a size and moving at speeds that people will regularly encounter (though not enough for GPS to work) and the mathematics behind it are a lot simpler than in theories that corrected for the observations.
Finally, the simplest and least error-prone solution: MAC policies (e.g. SELinux, TrustedBSD, and so on) can be configured to permit a specific application, running as a specific user, to open listening sockets on specified privileged ports. The httpd binary running as the apache user may listen on port 80, but no other program running as the appache user and no other using running the httpd program may do so.
In short, while this is a problem with classic UNIX, it's one that was solved at least a decade ago.
What sort of degree does an expert in "thinking logically" have? What department of the University are they found in? And if your curriculum is heavy on Logic, what is your most likely major?
Philosophy.
What makes you think that? Recent benchmarks show llvm/clang well behind across the board. At least it compiles faster
Recent benchmarks? They're comparing LLVM 3.5 (over a year old) with GCC 4.9 (over a year old). With the pace of current development (in both - having some competition has been very good for GCC), those are not recent, they're archaic. Oh, and don't overestimate the benefit of faster compilation when it comes to developer productivity, particularly for C++.
That is precisely the killer advantage of GPL over BSD license. There is no shame whatsoever in porting code if it is good. Rather, it is the professional thing to do. However it is a gross exaggeration to say the GCC team does only that.
The codebases are sufficiently different that it's not really feasible to port code over (unless you just lower GIMPLE to LLVM IR and plug LLVM in as a GCC backend, which is what LLVM used to do and how it was offered to the FSF originally. Fortunately for the rest of the world, the GCC community turned it down). Ideas and algorithms can't be copyrighted and these flow in both directions.
The killer advantage of the BSDL is that people don't have to buy into your ideology to contribute. Microsoft and Azul, for example, are both contributing a lot to improving the garbage collection infrastructure in LLVM (GCC has nothing comparable and the GCC attempt at a JIT involved spitting out a stream of assembly that was run through gas, because GCC doesn't have an integrated assembler, so would be very hard to adapt). In Microsoft's case, they're integrating it into their MIT-licensed CLR implementation. Azul is integrating it into their proprietary JVM. Apple has contributed a lot in this area as well, and are using it in their LGPL'd JavaScriptCore. None of these licenses would permit incorporating GPL'd code, so no contributions from these vendors would appear in a GPL'd project.
The project that's suffering most from Clang is not GCC, it's EDG. They've been selling a proprietary C/C++ front end since 1988, which is used by a lot of vendors. Those vendors are increasingly realising that if they invested half of the EDG license fees in improving Clang, they could be completely independent. The BSDL project is killing proprietary competitors that thrived while the GPL project was dominant. What does that tell you about the relative merits of the two licenses for promoting free software?
For the companies that I've talked to, the issue isn't so much systemd (which does solve real problems), it's the attitude of vendors like RedHat to their customers. People with large deployments don't like core parts changing without consultation and without a migration path carefully laid out in advance. They also don't like seeing stuff pushed before it's fully baked, especially core infrastructure.
I don't have any figures, but from the FreeBSD side we see an influx of new users every time Pottering releases something new. The stream of new switchers escaping PulseAudio had just about dried up, but systemd has given a new set.
Most of the lack of environmental restriction in China does not impact the rest of the world, it just results in tracts of land and rivers within China becoming poisoned. The reason the US and Europe have regulations that (try to) prevent this is that people got fed up with factories destroying their locale for a quick profit. This kind of environmental regulation is easy to sell ('the government wants to stop factories polluting the river I live on? Great!'). Regulation of CFCs was a lot harder to push through and required a lot of education about the effects of the eroding ozone layer before it became possible ('the government wants to ban stuff I use that's causing some nebulous and invisible damage? Bah! Let the free market sort it out!'). Greenhouse gas emissions are still in the latter category and their effect is far harder to measure because it's even more diffuse - increasing the energy in a chaotic system does not give easily predictable outcomes, though the consensus generally is that a species that depends on a fragile equilibrium point in the system being maintained should probably care.
Coal burning was relatively rare until quite recently in history. Coal needed to be mined, wood was just there and charcoal was relatively easy to produce. Coal only made economic sense once you started to have steam engines requiring hotter fires and increased demand from factories (at which point industrialised coal mining pushed the cost of coal down and made it economically feasible for home heating).
Decay also produces a lot of methane as a byproduct, whereas combustion produces relatively little. This means that the total contribution to the greenhouse effect is less if you burn a tree than if you allow it to rot (though it will be condensed into a shorter time).
They've been coming for ages, but mostly we'd been keeping the best ones. Now we've introduced US-style visa stupidity so after we've educated them we promptly deport them before they can contribute money to the economy. It used to be relatively easy to transition from a student visa to a work visa (to the extent that some universities got into trouble for sponsoring people for student visas, who never turned up to lectures and then applied for a work visa a bit later). Now it's very hard and often requires returning home and applying for the work visa from there.
Yup, go to the USA, learn skills from US experts, but no contributing back to the US economy for you! You have to go back to wherever you came from and contribute to their economy with the skills that you've acquired. We've just instituted this kind of stupidity in the UK too.
I've not used Access since it ran on Windows 3.11, but I recently tried OpenOffice / LibreOffice Base and was amazed. Their form builder is incredible. Even simple joins in the UI require writing SQL by hand. How anyone can ship that as a GUI database tool, I have no idea.
GIMP and Photoshop both have pretty terrible UIs. If you've spent a lot of time and effort becoming comfortable with one, then having to switch to the other is painful. At least a bad UI that you're familiar with means that you've learned to work around its brokenness - an unfamiliar bad UI is far more painful than a familiar one.
You know that Bing Ads is one of the largest ad networks around, right?
Actually, I had no idea - I've never seen one. Apparently they have 15.6% of the US market share, though it's not clear how much they have worldwide. Thanks for the info.
I've found that just sticking the city name in the search works a lot better than enabling any kind of IP-based geolocation for that (especially if I'm in a hotel and using a VPN, but even without that, IP geolocation accuracy varies hugely).
That, and the fact that we only classified cancer as a thing quite recently. No one was diagnosed with cancer a couple of hundred years ago, but that doesn't mean that no one died of it.
Exactly. With the exception of Microsoft (which sells software, yet still doesn't have a great track record, especially with the Windows 10 fiasco), all of the listed companies have business models that rely on collecting as much information as they possibly can from their users (not to be confused with their customers). If you want to resist surveillance, then don't buy into large centralised communication systems.
Is the connecting end resolvable? If it isn't, then you may be waiting for the lookup to time out (it has a low timeout, but it can add a noticeable amount of delay).
but that's of course just Debian
Actually, it isn't. The Linux Foundation is funding the effort, and it's mostly Debian people leading it, but they're working on a variety of projects (including FreeBSD!), not just Debian.
The problem with DuckDuckGo is that it doesn't take into account location
That's a configuration option (and you can look at the cookie contents to validate that it will be the same for anyone with that option). I generally leave it turned off, because most searches that I do are not geographical in nature (e.g. I want information about the C++ standard, not about the UK edition of the C++ standard).
at all times Google's result were superior to Altavista
That's a highly subjective judgement and likely to depend a lot on the search terms. Most of the time, either both or neither found what I was looking for. Towards the end, AltaVista largely succumbed to spam, but around 2000 there wasn't much in it.
Not only was the Google ranking algorithm superior, the Altavista crawl fell way behind the Google crawl, so often the page you were looking for wasn't even in the Altavista index.
Until Google moved away from MapReduce, their results also lagged a long time behind. Searching for anything in today's news wouldn't find any relevant results with Google until long after AltaVista's demise.
The 'fast' thing really can't be overstated. By the time Google launched, AltaVista's search page had become huge, to the extent that it took about 30 seconds to load on a 28.8Kb/s MODEM (the fastest that mine could connect at given the line quality, though on paper it could do 56Kb/s). Google took well under 5 seconds (not because Google devs were clever and actively aimed for this, quite the reverse: they didn't initially have anyone good at HTML/CSS stuff, so produced the simplest page that worked).
I remember the search results on Google being worse than AltaVista, but getting them so much faster that I could start loading the first 3-4 before AltaVista showed me anything. Occasionally I'd go back to AltaVista if Google failed. A few years later, Google fucked up their UI enough to make me switch to DuckDuckGo.
The question is why people should care about Apple being in the consortium. The answer is that people want to sell things (e.g. streaming services) to people with mobile devices, and these may depend on vendor support for the codecs that they use (on iOS, you can't easily get hardware acceleration unless you use Apple's AVFoundation framework). Whether this is actually important depends on whether or not that market is one that you care about.