We live in a society based on numbers, strategies, and logic. I do not think there is an upper limit on necessary math education in such a society.
What we really need to ask is, "Are we teaching the right things in math?" Perhaps instead of teaching students to memorize formulas and algorithms in basic algebra and calculus, we should teach students combinator logic, lambda calculus, etc....
But it's worth noting that when one actually attempts the exercise, one sees that the rest of the world has seen a remarkable rise in standard of living and wealth, not just for their wealth, but for the rest as well. Capitalism and global trade are far more likely to be responsible for that.
You must be referring to the way that capitalism led to a cure for polio. Except that it did not, and Jonas Salk gave the cure away to improve the world. In fact, the improvements in standards of living around the world have more to do with the hard work of scientists and engineers than with capitalism.
No, capitalism did not pay for that research; capitalism is bad at paying for long-term projects that have no clear or immediate profitability. Most of the major scientific breakthroughs that have really improved the standard of living in the world -- cures for diseases, better ways to grow food, etc. -- were paid for either with tax money or with some private endowment/gift money.
A lot of those effects from those who claim to be "fixing" capitalism.
No, those are the effects of unregulated capitalism. Unless you think that everyone is equally ruthless, intelligent, educated, and that they have equal amounts of capital, the "little guy" is going to be crushed by the "big guys" in an unregulated market. That is why, every so often, we break up monopolies (though lately we seem to be forgetting to do that): so that we can reset the market and start the competition again.
If there's a lot of regulation, then the big company that can navigate the regulation (say by have a huge legal staff for doing so) and bribe the right people, is going to fare better than the small company that can't.
If there is no regulation, the big company will crush the little company by selling its products/services at a loss until the little company has no customers left. The big company will also offer grossly inflated salaries to the most intelligent people at the little company. The big company will make deals with other big companies, to lock the little company out of the market.
That is what happens when one player has vastly more capital than the rest.
Until you can make those "crumbs" yourself, you'll always be subservient to those who can
That's funny, because in capitalism, the people who bake the bread are usually subservient to the people who own the oven. The winners in capitalism are those with capital, not the scientists and engineers who solve societies problems and not the workers who put those solutions into action.
Frankly, I think this is a disease that is mostly a result of the attempted cure rather than of capitalism.
Meanwhile, in the real world, the regulations we placed on businesses stopped child laborers from being killed and maimed, gave smaller, more innovative businesses an opportunity to compete, and raised our standard of living.
We are talking about the people who possess it, which is an entirely different matter. Possession of child pornography has never been evidence that a person is a child molester or even that the person is a threat to children. Production of child pornography involves raping a child, and so there can be little question that producers pose a threat to children.
Historically, child pornography possession was made illegal as part of an effort to fight its production. At the time, people had to buy their child pornography, through shady connections. The producers (i.e. child rapists) were hard to find, and for obvious reasons: they are organized, they are mobile, and they are secretive about what they do. As a strategy to protect children from these rapists, it made sense to put pressure on the distribution chain and ultimately on the people at the ends of that chain, the consumers, to reduce the demand for child pornography. The courts agreed with that when the law was challenged.
Today, things are substantially different. Child pornography is easily found on the Internet, just by entering some not-hard-to-guess code words (and those are not hard to learn, if you are not educated enough to guess them). It is easy to come into possession of child pornography without paying for it, trading for it, or creating it yourself. The situation has changed, and the strategy must change with that situation, assuming that goal is still to protect children (but chances are that the goal has shifted too).
So do not defend the producers of child pornography, because if you spent even one hour looking at the evidence in child pornography cases, you would be scarred for life. The people who produce that material are torturing children, plain and simple. They deserve no sympathy, and I hope they are caught and imprisoned. The people collecting that material are another story; only those who can be shown to have paid for it or otherwise directly encouraged its production should be charged with any crime.
So you are going to tell one of your biggest customers, "We told you over a year ago that you had to replace those hardware modules, so why did you not do it?"
It is easy for Microsoft to phase out 768 bit keys; hardly anyone uses them these days. 1024 bit keys are a completely different story; they are widespread, popular, and it is going to be expensive to replace them all. For over a decade, 1024 bits has been the default, and during that time a lot of systems were deployed, including a lot of hardware modules. Some of those systems have the key-length set in stone, and some of those systems are hard to replace (imagine taking a mission critical system down to upgrade your key length -- try selling that one to management).
1024 bit is deprecated, but it is not going to be gone any time soon. There is just too much friction, and too little understanding of why key lengths should be increased.
The most common key length is 1024 bits, and that is not going to change for a while. Despite the fact that most popular software packages default to 2048 bits, the increased load on servers, the lack of any successful attack on 1024 bit RSA (except for implementation attacks), and the general effort required to replace a public key will mean that most people will not bother for many years to come. There is also a lot of hardware out there that was built with a 1024 bit maximum, and that hardware is going to be expensive to replace.
If you really want to get with the times, you should be talking about 3072 bit keys (roughly equivalent to AES-128), or you should be talking about ECC. If I had to guess, I would say that we will see elliptic curve systems become popular before 2048 bit RSA -- ECC has more promise in terms of efficiency and the security level you get per bit. Even the OpenPGP standard now includes ECC.
In what way is this going to be a pain in the butt? Are there really that many people out there using less than 1024 bit in this day and age?
You would think that after the successful factoring of a 768 bit RSA modulus, whoever was still using less than 1024 bit would have fixed that. Frankly, 1024 bit should be considered too short for any new applications going forward, but that is still built in to quite a lot of packages.
Aren't there existing protections limiting prosecution to knowingly and intentionally committing crimes? I can't see how legalizing possession completely will "fix" the "problem" of accidental prosecution in an effective way. Baby/bathwater and all that.
Unfortunately, possession of child pornography is a crime regardless of the reason you have it, regardless of whether or not you intended to possess it, regardless of whether or not you were even able to access it, regardless of whether or not you even knew you had it. Even cartoon depictions of child sex abuse are illegal in the United States.
As for prosecution, the point here is that possession alone should not be a crime at all -- the original concept was to attack the producers of child porn by attacking their customers, since the producers themselves were so hard to find. The Internet has changed things: people can just download child pornography without paying for it. The law should at least be revised to reflect that fact, perhaps by making it a crime to pay (by money or by barter) rather than to possess. Yes, that means the police will have to actually gather evidence that a person was paying i.e. encouraging child rape, otherwise known as doing their jobs.
Google did not kill Usenet; AOL killed Usenet in 1993, when they started sending millions of rude and uneducated users to Usenet without bothering to explain basic Usenet conventions or etiquette.
The Unix API was designed in the 70s and it still seems to work just fine.
I can think of a few things that are pretty bad about the Unix API:
The fork/exec model
File modes (as opposed to ACLs) and the lack of built-in mandatory access control/RBAC
The fact that files can only be streams (not records, not multidimensional arrays, not trees, etc.)
The "just give up" approach to error handling (don't even try to recover from certain signals)
The difficulty in propagating an error through the composition of several programs (if one is killed by SEGV, others will be killed by PIPE)
No standardized support for file versioning
A very primitive concept of object oriented programming (e.g. FILE, the sockets API, etc.)
...and there are many others. Note that Unix has actually changed significantly since the 70s -- we have standards for threads, for discretionary ACLs, for shared memory, etc. Still, the overarching design of Unix has not changed much with the times; it is still processes and files. This is not always the best approach, and it does not always work "just fine." Basic CGI is not very competitive against a JEE application server for a heavy load (yes, I know JEE servers are often run on Unix, but the JEE abstractions are almost entirely unrelated to the Unix abstraction).
My point is not that Unix is bad, but that for today's computing needs there are better approaches. Likewise, in 35 years, today's approach to desktop environments will probably not be up to the task; there will be new ideas, new technologies, and new ways to use computers that will require new approaches to UI. If we continue to use today's abstractions in 2047, we are going to be unable to rise to those challenges, just like CGI does not really scale well enough for today's websites.
Follow that logically, and what you see is that the RIAA and MPAA really do care about everything that runs on your computer, from the moment you boot up. Yes, at the end of the day, they only care about their specific copyrights; but to enforce those copyrights on your computer, they need to ensure that your kernel won't allow some program to read another program's memory, and they need to ensure that your bootloader won't allow you to run a modified kernel, and they need to ensure that your BIOS won't load a modified bootloader...
Essentially, they want to rewrite the rules, so that PCs are just fancy televisions, devices that are useful only for consuming entertainment but not for creating it. Sure, you can run whatever you want...as long as it is approved by someone who will make sure that your software is not going to violate the RIAA/MPAA/etc. copyright rules and edicts.
Claiming that someone only wants to ensure that some particular class of software is not run on a computer is just a complicated way to say that they want to control all software that runs on the computer.
A satisfied user doesn't help "spread the Linux base"? Why not, I ask seriously?
Maybe we do not want to get stuck doing tech support for our entire social circle. It's unfortunate, but it is true: we are still at the point where if we install GNU/Linux on someone's machine, we take on the responsibility of solving their problems (and that ultimately means solving problems that are unrelated to their OS -- an unplugged cable, an overheated router, a power outage, etc.). LUGs are dying and cannot provide useful community help, and online forums are full of bad, contradictory advice. We are still not bothering to educate anyone about computers (except how to use a speciifc company's product in specific ways), and so most computer users remain helpless.
This is not merely an uphill battle; it is more like an attempt to reach escape velocity.
Stop copying Apple and get back to being creative and innovative. Why make a (polished) copy of a (polished) copy? Let's do our own thing, let's do it well, and let's stand on the merits of our own ideas.
Of course, we have a lot of people in the user and developer communities that surround GNU/Linux who grew up thinking that the goal was to get away from Windows. Now everyone is seeing Windows in its decline...but the users are switching to Mac OS X and iOS, and so naturally, we need to give users something like OS X and iOS. We built a movement on the wrong premise (that the enemy was Microsoft) and now we are suffering for it.
Meh...the problem with that is that APIs are hard to change once they are so unified. Do we really want to cement ourselves with today's moed of thought?
Look at X11 -- that was designed decades ago, yet the core API and abstractions are still with us. Look at Wayland -- not even close to being popular. Once you standardize an API, you are stuck with it; I do not know about you, but today's DEs are not something I want to be stuck with.
A somewhat more accurate strawman would be "We need more *good* or *compelling* applications" -- that's challenging. Still only a part of the answer, but closer. It requires answering "What does 'good' or 'compelling' mean in this context?", etc.
You are referring to a "killer app." Spreadsheets were a killer app. The World Wide Web was a killer app. Someone needs to come up with some totally new reason to use computers, and if their code runs only on GNU/Linux, then we will see people moving in that direction.
Of course, why would anyone write a killer app that only ran on GNU/Linux? We do not have an OS that is so substantially different from Windows or Mac OS X that such a thing would make sense. What is needed to facilitate that is something so innovative that Apple and Microsoft are still busy trying to catch up by the time a killer app is written.
I know I'll get flamed for this since it goes against the Linux philosophy, but how about getting rid of competing Gnome and KDE (and now Unity) desktops and agree on one standard desktop with a single API for everyone to write to.
That is what KDE was originally supposed to be, but there were licensing problems with Qt, so we GNOME and GTK were created. By the time the Qt license problem was resolved, you had two communities with two different ideas on what the desktop should look like and where it should be going, and so there was no way we would get a standard.
I like the choice, and I like the fact that individual distros are choosing the DE they want to build on. That is how it should be, because the distros should not be groups together as a single OS anyway (any more than you can speak of "Unix" as a single OS). We do not want unification, because that would cement us (look at Windows -- unable to move on for many, many years because of legacy code and compatibility concerns) and would prevent us from trying new things.
As soon as we accept the idea that we are not here to be a single OS where everyone sees applications the same way, we will be able to do things that will be innovative. We can try new things, we can try new approaches, we can build things that will actually change the state of software for the better. The "Ubuntu attitude" -- the idea that we need a unified Linux OS -- is holding us back.
The drivers work well enough in most cases; I do not have the problems you have (but my laptop is a bit older).
We have repeatedly asking hardware companies to release the specifications for their hardware, and so far we have gotten little from them. All we want to know is what a driver for the hardware should do, and then we can write it ourselves; they can issue proprietary drivers all they want, and we can even agree to specifically tell people that our drivers are not supported and come with no warranty whatsoever.
Also, some developer writing oodles of Python code is not going to write drivers. The kernel people is doing a bang-up job; it is not perfect, but it is good. The UI people are not doing a bang-up job, and in fact, they seem to be doing worse and worse with each passing year, trying harder and hard to copy Apple and Microsoft. We do not need copies, we need innovations -- not just new ways to display grids and linear menus, but new approaches entirely.
How about trying something that other people are not really doing? Let's get radial menus (there was one WM that had this, but I forget the name) instead of continuing to cling to inefficient linear menus. Let's find a way to make arbitrary compositions of GUI applications, the way we can arbitrarily compose applications in our terminal (KDE3 was a step towards this, but we could have done a whole lot more).
In other words, let's take a risk and try being innovative. What is the point of copying Apple? Let's do something that Apple will want to copy.
When I write://TODO: fix I usually include a little bit about what needs fixing
Which in my opinion is the right thing.
even the simple comment allows my code editor to build a "To Do" list automagically that includes the context (file, function, line number, method signature, etc).
Sure, but it does not tell you what needs fixing; that comment alone does not even tell you why the code needs fixing. You might have to spend an hour running tests or debugging to even figure out why the comment is here. Maybe the function returns the wrong value for one in a hundred test cases. Maybe you never run a test that breaks the function. Maybe the function works fine on 64 bit systems, but breaks on 32 bit systems, and your workstation is 64 bit.
A comment like that is no different from having your grandma say, "my computer is broken." Does that mean it won't boot up? It won't connect to the Internet? Is it a plugin? Is it plugged in? Did your 5 year old cousin spill soda on it? We can forgive our not-so-technical friends and family for leaving out context or being unable to articulate what is actually wrong; but surely a programmer can be expected to be able to explain why something needs to be fixed.
//TODO: Conditionally recycle near death particles if over system limit.
My research group will hate you, because this is going to be inserted somewhere in my code;)
True enough, although if you are not a prolog programmer, you really should not be reading through prolog code (unless you are just trying to learn). A comment might specify some behavior that the code does not exhibit -- you need to be able to read both.
As for the overflow, a comment could clarify things -- for example, if you know that the variable is constrained to some range of values, then just writing that in a comment (say, "0
for(i = 1; i < n; i++) res = res*i;
A quick comment clarifying that n will be in some range for which this will not wrap around could be really helpful -- if, say, you are debugging and seeing some strangeness, such a comment gives you something you can check (is n really in that range) or at least a clue about a possible issue (maybe the comment was written assuming 64 bits for res, but you are using a 32 bit type -- but the comment still provides a clue that an overflow can easily happen in this code).
The problem with writing clear code is that even seemingly simple code can have some non-obvious behavior; i++ is a perfect example. It looks like it only says, "increment i;" what it actually says is, "increment i modulo 2^n," where n is typically 32 or 64 but could be something else. That is not obvious, especially when you almost always see i++ being used as a way to increment integers. It is not enough to demand that programmers train themselves to think differently, and there are cases where just using a language with more natural definitions (e.g. one in which arbitrary precision arithmetic is the default, or perhaps one in which an overflow causes an exception to be thrown) is not really an option (if you have a million lines of C code, you need to write at least some C code).
At some point the code is so simple it comments itself.
That really depends on your language. Consider this EmacsLisp code:
(defun fact (x) (if (<= x 1) 1 (* x (fact (- x 1)))))
That sure looks like a definition of factorial, but on my system, that will only give you the right answer if x is less than 12. The reason, of course, is that in EmacsLisp numeric values have fixed precision, and you get an integer overflow. There are two possible answers to that situation:
Well duh! Just use an arbitrary precision type. Write one yourself if EmacsLisp does not have it built-in. That code can still be self documening!
Add a comment warning other programmers that the function will only work for some limited range of possible input values.
Or perhaps this third option: don't code in a language where such a thing can happen. I wonder how many C programmers would try pushing for that option...
Really, that's the example you chose? That is not at all clear. That might increment i (so that the new value of i is one greater than the old value) -- but that might also cause an integer overflow. If this is C++, you might have a type that overloads the ++ operator: maybe that is a type for saturation arithmetic, so i++ might not even change the value of i.
Really, if you want to see code that clearly specifies what it does, you need to look at higher level languages. Prolog comes to mind here, as do those languages Backus created for "function level programming." If you know Prolog syntax, then this code requires no comments:
Slashdot Mirror
How much math do most people really need?
We live in a society based on numbers, strategies, and logic. I do not think there is an upper limit on necessary math education in such a society.
What we really need to ask is, "Are we teaching the right things in math?" Perhaps instead of teaching students to memorize formulas and algorithms in basic algebra and calculus, we should teach students combinator logic, lambda calculus, etc....
I've had math professors who could barely speak English because they were foreign countries
Hm...
There's a lot of money at stake. If these big online courses catch on, the professoriate will be out on the street
No, they will be doing research, because that is what pays the bills for most professors.
But it's worth noting that when one actually attempts the exercise, one sees that the rest of the world has seen a remarkable rise in standard of living and wealth, not just for their wealth, but for the rest as well. Capitalism and global trade are far more likely to be responsible for that.
You must be referring to the way that capitalism led to a cure for polio. Except that it did not, and Jonas Salk gave the cure away to improve the world. In fact, the improvements in standards of living around the world have more to do with the hard work of scientists and engineers than with capitalism.
No, capitalism did not pay for that research; capitalism is bad at paying for long-term projects that have no clear or immediate profitability. Most of the major scientific breakthroughs that have really improved the standard of living in the world -- cures for diseases, better ways to grow food, etc. -- were paid for either with tax money or with some private endowment/gift money.
A lot of those effects from those who claim to be "fixing" capitalism.
No, those are the effects of unregulated capitalism. Unless you think that everyone is equally ruthless, intelligent, educated, and that they have equal amounts of capital, the "little guy" is going to be crushed by the "big guys" in an unregulated market. That is why, every so often, we break up monopolies (though lately we seem to be forgetting to do that): so that we can reset the market and start the competition again.
If there's a lot of regulation, then the big company that can navigate the regulation (say by have a huge legal staff for doing so) and bribe the right people, is going to fare better than the small company that can't.
If there is no regulation, the big company will crush the little company by selling its products/services at a loss until the little company has no customers left. The big company will also offer grossly inflated salaries to the most intelligent people at the little company. The big company will make deals with other big companies, to lock the little company out of the market.
That is what happens when one player has vastly more capital than the rest.
Until you can make those "crumbs" yourself, you'll always be subservient to those who can
That's funny, because in capitalism, the people who bake the bread are usually subservient to the people who own the oven. The winners in capitalism are those with capital, not the scientists and engineers who solve societies problems and not the workers who put those solutions into action.
Frankly, I think this is a disease that is mostly a result of the attempted cure rather than of capitalism.
Meanwhile, in the real world, the regulations we placed on businesses stopped child laborers from being killed and maimed, gave smaller, more innovative businesses an opportunity to compete, and raised our standard of living.
We are talking about the people who possess it, which is an entirely different matter. Possession of child pornography has never been evidence that a person is a child molester or even that the person is a threat to children. Production of child pornography involves raping a child, and so there can be little question that producers pose a threat to children.
Historically, child pornography possession was made illegal as part of an effort to fight its production. At the time, people had to buy their child pornography, through shady connections. The producers (i.e. child rapists) were hard to find, and for obvious reasons: they are organized, they are mobile, and they are secretive about what they do. As a strategy to protect children from these rapists, it made sense to put pressure on the distribution chain and ultimately on the people at the ends of that chain, the consumers, to reduce the demand for child pornography. The courts agreed with that when the law was challenged.
Today, things are substantially different. Child pornography is easily found on the Internet, just by entering some not-hard-to-guess code words (and those are not hard to learn, if you are not educated enough to guess them). It is easy to come into possession of child pornography without paying for it, trading for it, or creating it yourself. The situation has changed, and the strategy must change with that situation, assuming that goal is still to protect children (but chances are that the goal has shifted too).
So do not defend the producers of child pornography, because if you spent even one hour looking at the evidence in child pornography cases, you would be scarred for life. The people who produce that material are torturing children, plain and simple. They deserve no sympathy, and I hope they are caught and imprisoned. The people collecting that material are another story; only those who can be shown to have paid for it or otherwise directly encouraged its production should be charged with any crime.
So you are going to tell one of your biggest customers, "We told you over a year ago that you had to replace those hardware modules, so why did you not do it?"
It is easy for Microsoft to phase out 768 bit keys; hardly anyone uses them these days. 1024 bit keys are a completely different story; they are widespread, popular, and it is going to be expensive to replace them all. For over a decade, 1024 bits has been the default, and during that time a lot of systems were deployed, including a lot of hardware modules. Some of those systems have the key-length set in stone, and some of those systems are hard to replace (imagine taking a mission critical system down to upgrade your key length -- try selling that one to management).
1024 bit is deprecated, but it is not going to be gone any time soon. There is just too much friction, and too little understanding of why key lengths should be increased.
As everyone moves to 2048 bit keys
In what way is this going to be a pain in the butt? Are there really that many people out there using less than 1024 bit in this day and age?
You would think that after the successful factoring of a 768 bit RSA modulus, whoever was still using less than 1024 bit would have fixed that. Frankly, 1024 bit should be considered too short for any new applications going forward, but that is still built in to quite a lot of packages.
Aren't there existing protections limiting prosecution to knowingly and intentionally committing crimes? I can't see how legalizing possession completely will "fix" the "problem" of accidental prosecution in an effective way. Baby/bathwater and all that.
Unfortunately, possession of child pornography is a crime regardless of the reason you have it, regardless of whether or not you intended to possess it, regardless of whether or not you were even able to access it, regardless of whether or not you even knew you had it. Even cartoon depictions of child sex abuse are illegal in the United States.
As for prosecution, the point here is that possession alone should not be a crime at all -- the original concept was to attack the producers of child porn by attacking their customers, since the producers themselves were so hard to find. The Internet has changed things: people can just download child pornography without paying for it. The law should at least be revised to reflect that fact, perhaps by making it a crime to pay (by money or by barter) rather than to possess. Yes, that means the police will have to actually gather evidence that a person was paying i.e. encouraging child rape, otherwise known as doing their jobs.
The advent of ads has allowed websites to gather money & use that money to develop content I want to see.
At the cost of turning the Internet into the world's biggest adversarial game, where you need various software filters just to survive.
free online magazines to read
Hm...
https://en.wikipedia.org/wiki/Phrack
I guess that there were free magazines to read online before the adversarial Internet...
Google did not kill Usenet; AOL killed Usenet in 1993, when they started sending millions of rude and uneducated users to Usenet without bothering to explain basic Usenet conventions or etiquette.
This hasn't gone down well with ad networks
To quote Firefly: "Do we care? Is this something we are caring about?"
The Unix API was designed in the 70s and it still seems to work just fine.
I can think of a few things that are pretty bad about the Unix API:
My point is not that Unix is bad, but that for today's computing needs there are better approaches. Likewise, in 35 years, today's approach to desktop environments will probably not be up to the task; there will be new ideas, new technologies, and new ways to use computers that will require new approaches to UI. If we continue to use today's abstractions in 2047, we are going to be unable to rise to those challenges, just like CGI does not really scale well enough for today's websites.
Eh...
Follow that logically, and what you see is that the RIAA and MPAA really do care about everything that runs on your computer, from the moment you boot up. Yes, at the end of the day, they only care about their specific copyrights; but to enforce those copyrights on your computer, they need to ensure that your kernel won't allow some program to read another program's memory, and they need to ensure that your bootloader won't allow you to run a modified kernel, and they need to ensure that your BIOS won't load a modified bootloader...
Essentially, they want to rewrite the rules, so that PCs are just fancy televisions, devices that are useful only for consuming entertainment but not for creating it. Sure, you can run whatever you want...as long as it is approved by someone who will make sure that your software is not going to violate the RIAA/MPAA/etc. copyright rules and edicts.
Claiming that someone only wants to ensure that some particular class of software is not run on a computer is just a complicated way to say that they want to control all software that runs on the computer.
A satisfied user doesn't help "spread the Linux base"? Why not, I ask seriously?
Maybe we do not want to get stuck doing tech support for our entire social circle. It's unfortunate, but it is true: we are still at the point where if we install GNU/Linux on someone's machine, we take on the responsibility of solving their problems (and that ultimately means solving problems that are unrelated to their OS -- an unplugged cable, an overheated router, a power outage, etc.). LUGs are dying and cannot provide useful community help, and online forums are full of bad, contradictory advice. We are still not bothering to educate anyone about computers (except how to use a speciifc company's product in specific ways), and so most computer users remain helpless.
This is not merely an uphill battle; it is more like an attempt to reach escape velocity.
Stop copying Apple and get back to being creative and innovative. Why make a (polished) copy of a (polished) copy? Let's do our own thing, let's do it well, and let's stand on the merits of our own ideas.
Of course, we have a lot of people in the user and developer communities that surround GNU/Linux who grew up thinking that the goal was to get away from Windows. Now everyone is seeing Windows in its decline...but the users are switching to Mac OS X and iOS, and so naturally, we need to give users something like OS X and iOS. We built a movement on the wrong premise (that the enemy was Microsoft) and now we are suffering for it.
Meh...the problem with that is that APIs are hard to change once they are so unified. Do we really want to cement ourselves with today's moed of thought?
Look at X11 -- that was designed decades ago, yet the core API and abstractions are still with us. Look at Wayland -- not even close to being popular. Once you standardize an API, you are stuck with it; I do not know about you, but today's DEs are not something I want to be stuck with.
A somewhat more accurate strawman would be "We need more *good* or *compelling* applications" -- that's challenging. Still only a part of the answer, but closer. It requires answering "What does 'good' or 'compelling' mean in this context?", etc.
You are referring to a "killer app." Spreadsheets were a killer app. The World Wide Web was a killer app. Someone needs to come up with some totally new reason to use computers, and if their code runs only on GNU/Linux, then we will see people moving in that direction.
Of course, why would anyone write a killer app that only ran on GNU/Linux? We do not have an OS that is so substantially different from Windows or Mac OS X that such a thing would make sense. What is needed to facilitate that is something so innovative that Apple and Microsoft are still busy trying to catch up by the time a killer app is written.
I know I'll get flamed for this since it goes against the Linux philosophy, but how about getting rid of competing Gnome and KDE (and now Unity) desktops and agree on one standard desktop with a single API for everyone to write to.
That is what KDE was originally supposed to be, but there were licensing problems with Qt, so we GNOME and GTK were created. By the time the Qt license problem was resolved, you had two communities with two different ideas on what the desktop should look like and where it should be going, and so there was no way we would get a standard.
I like the choice, and I like the fact that individual distros are choosing the DE they want to build on. That is how it should be, because the distros should not be groups together as a single OS anyway (any more than you can speak of "Unix" as a single OS). We do not want unification, because that would cement us (look at Windows -- unable to move on for many, many years because of legacy code and compatibility concerns) and would prevent us from trying new things.
As soon as we accept the idea that we are not here to be a single OS where everyone sees applications the same way, we will be able to do things that will be innovative. We can try new things, we can try new approaches, we can build things that will actually change the state of software for the better. The "Ubuntu attitude" -- the idea that we need a unified Linux OS -- is holding us back.
Also, some developer writing oodles of Python code is not going to write drivers. The kernel people is doing a bang-up job; it is not perfect, but it is good. The UI people are not doing a bang-up job, and in fact, they seem to be doing worse and worse with each passing year, trying harder and hard to copy Apple and Microsoft. We do not need copies, we need innovations -- not just new ways to display grids and linear menus, but new approaches entirely.
How about trying something that other people are not really doing? Let's get radial menus (there was one WM that had this, but I forget the name) instead of continuing to cling to inefficient linear menus. Let's find a way to make arbitrary compositions of GUI applications, the way we can arbitrarily compose applications in our terminal (KDE3 was a step towards this, but we could have done a whole lot more).
In other words, let's take a risk and try being innovative. What is the point of copying Apple? Let's do something that Apple will want to copy.
When I write: //TODO: fix I usually include a little bit about what needs fixing
Which in my opinion is the right thing.
even the simple comment allows my code editor to build a "To Do" list automagically that includes the context (file, function, line number, method signature, etc).
Sure, but it does not tell you what needs fixing; that comment alone does not even tell you why the code needs fixing. You might have to spend an hour running tests or debugging to even figure out why the comment is here. Maybe the function returns the wrong value for one in a hundred test cases. Maybe you never run a test that breaks the function. Maybe the function works fine on 64 bit systems, but breaks on 32 bit systems, and your workstation is 64 bit.
A comment like that is no different from having your grandma say, "my computer is broken." Does that mean it won't boot up? It won't connect to the Internet? Is it a plugin? Is it plugged in? Did your 5 year old cousin spill soda on it? We can forgive our not-so-technical friends and family for leaving out context or being unable to articulate what is actually wrong; but surely a programmer can be expected to be able to explain why something needs to be fixed.
//TODO: Conditionally recycle near death particles if over system limit.
My research group will hate you, because this is going to be inserted somewhere in my code ;)
Speaking of non-obvious problems...slashdot ate a few < characters in there...
As for the overflow, a comment could clarify things -- for example, if you know that the variable is constrained to some range of values, then just writing that in a comment (say, "0
A quick comment clarifying that n will be in some range for which this will not wrap around could be really helpful -- if, say, you are debugging and seeing some strangeness, such a comment gives you something you can check (is n really in that range) or at least a clue about a possible issue (maybe the comment was written assuming 64 bits for res, but you are using a 32 bit type -- but the comment still provides a clue that an overflow can easily happen in this code).
The problem with writing clear code is that even seemingly simple code can have some non-obvious behavior; i++ is a perfect example. It looks like it only says, "increment i;" what it actually says is, "increment i modulo 2^n," where n is typically 32 or 64 but could be something else. That is not obvious, especially when you almost always see i++ being used as a way to increment integers. It is not enough to demand that programmers train themselves to think differently, and there are cases where just using a language with more natural definitions (e.g. one in which arbitrary precision arithmetic is the default, or perhaps one in which an overflow causes an exception to be thrown) is not really an option (if you have a million lines of C code, you need to write at least some C code).
At some point the code is so simple it comments itself.
That really depends on your language. Consider this EmacsLisp code:
That sure looks like a definition of factorial, but on my system, that will only give you the right answer if x is less than 12. The reason, of course, is that in EmacsLisp numeric values have fixed precision, and you get an integer overflow. There are two possible answers to that situation:
Or perhaps this third option: don't code in a language where such a thing can happen. I wonder how many C programmers would try pushing for that option...
Really, if you want to see code that clearly specifies what it does, you need to look at higher level languages. Prolog comes to mind here, as do those languages Backus created for "function level programming." If you know Prolog syntax, then this code requires no comments: