The flaw in this tactic is that it requires the person discussed on the billboard to be able to feel shame at the things they do in their official capacity.
I don't think that was the goal.
I think the goal is to sway the mind of voters. That doesn't necessarily mean changing a die hard republican into a die hard democrat, but it can be as simple as convincing a tepid republican voter to become more apathetic and just staying home, or convincing a tepid democrat voter to actually go and vote for something they already believe in rather than being lazy.
For me it's not about jumping ship. I already use linux everyday for work. I only have 2 machines that still use windows. My home desktop for gaming, and one work computer that is only used for outlook, and our IT department is already looking into ways to get rid of exchange.
They might very well do what you are saying and restrict users even further, I don't think it's a good idea or a bad idea, because I don't think it really matters what they do anymore.
I'm not some linux zealot. I don't think that the people who say windows is completely obsolete are being truthful either. They neglect that windows is currently the best solution for many situations. What I am saying is that the niche they are living in is getting smaller and smaller, and one day it will disappear.
Game developers are going to do whatever makes them the most money. For the last couple decades, that has meant treating windows as the primary game platform. I think Valve has the right idea in terms of starting to look for a way of this sinking ship. Tools for developing cross platform games are getting better and better. I think MS trying to over-exploit their past dominance has a very good chance of hastening their demise rather than prolonging it.
No doubt MS would love to "get young at a young age to not use ", but they have had limited and diminishing success at this.
Why would they change course?
That's like asking, why wouldn't they go from offering a product with 1 positive feature to a product with 0 positive features.
The only reason people like and use windows is because it is a convenient vessel through which they can easily run the programs they want to run. Take that away, and there is literally no more reason to use it at all.
If I can't play steam games, and run obscure utilities that were only built for windows, I will just use linux.
Windows is hanging on by a thread. They completely lost the mobile moarket, and the desktop market is shrinking. More and more things that you would do "on windows" are being done in the cloud.
Steam has decided windows is a dying platform for games, and has started supporting more and more games on linux.
Is windows going to die tomorrow? No. But there are just too many many good alternatives that are getting better everyday.
It's not like charismatic people need to be put *somewhere* in any given company. There are plenty of unemployed people. If they were actually so bad for these companies to warrant putting them in the positions of least harm potential, we may as well make all the humble unemployed people CEO's, and fire all the narcissistic CEO's to reduce harm even further.
There are different types of simplicity. There is simplicity involved in creating something and the simplicity of using it, and these 2 things are often opposed to each other. Raw pointers are simpler to create than smart pointers (the language feature, not an instance). They are easier to understand than smart pointers. They are harder to use correctly than smart pointers. How many bugs are a result of improper pointer use?
And if you look at code that uses raw pointers and the same code that has been modified to use smart pointers, the code that uses smart pointers will look simpler. It will look the same minus all the delete/free calls, and minus some NULL checks.
So yes, smart pointers are more complex in some sense. But if you consider the problem of enforcing the proper use of pointers, rather than just deferring that problem to the next programmer, smart pointers are a very simple solution to this problem (e.g. as opposed to something like garbage collection).
Not solving problems is simpler than solving them. I am all for not solving problems that don't need to be solved. But preventing pointer misuse seems like a problem worth solving, and it is solved elegantly with smart pointers.
This is just one example of something you can do with a more complex language like C++ that you can't do in C.
I don't think you can call yourself a C++ expert and be "totally unable to deal with" a C++ codebase that uses a different subset of C++ than what you would have used.
You can certainly be a C++ expert and be totally unable to deal with a very poorly designed codebase, but that is true for every language (even C).
Maybe there are more experts in the C language, because it is a simpler language and therefore an easier language to become an expert at. I would be totally in favor of using a simpler language for a project if there were no benefit to any of the more advanced features provided in C++, but that is rarely the case.
The 4 things I listed are insanely powerful tools for making safer and more maintainable code, and these things come with little to no performance overhead.
It also has good features, many of which provide you with an alternative to the bad features of C.
Constructors/destructors (RAII) > manual initialization/deinitialization.
Smart Pointers (made possible by RAII) > raw pointers
Polymorphism > function pointers
Templates > macros
If you are subscribing to a streaming movie service and you have the choice between netflix and a site that only allows you to watch "Armageddon (1998)", does it make sense to choose the latter because netflix has more bad movies? No of course not, because you don't have to watch the bad movies on netflix, and you can even choose only to watch movies better than Armageddon.
C++ is netflix. Nobody watches all the movies. Everyone watches the movies that are good from their point of view (even though many of those people are just wrong).
That doesn't mean it's not hard. That just means it's a hard thing that no one needs to do anymore for any reason other than... educational purposes.
I'm not saying not to use the tool. Use the tool. Maybe running as fast as Usain Bolt is hard for some people. I use a tool called a car. Running is super easy for me.
Running isn't hard either. Most people can run. It's only hard when you say something like "You need to be able to run a 5 minute mile". Some math is not hard just like some running is not hard. If you lower your standards enough then you can make it so math is not hard for 95% of people. You can also raise your standards enough to make it hard for 99.99% of people.
Even seemingly simple things like algebra can be really hard. I have had many calculus classes where the actual calculus wasn't so difficult, but the algebra was a nightmare. There is also nightmare calculus as well.
I've been to a few theaters that are basically restaurants, with menus waiters, tables, and reclining chairs. I thought it would be annoying to have everybody ordering and eating food during the movie, but the seats were far enough apart to not be very distracted but what other people were doing, and the cost was probably high enough to preclude idiot teenagers and adults who act like idiot teenagers.
I don't really go to the movies anymore because I have 2 screaming babies, but once a year when we hire baby sitters and go out, I appreciate the more luxurious experience. I am already blowing at least $100 on a babysitting, trying to save money by being next to some middleschool shithead talking loudly through the whole movie seems like a waste.
I don't need super crazy sound. I don't need 3D. I just want some comfortable chairs, good food, low distraction.
I'm so sorry you have a problem with my crying babies, but I'm just doing what Christopher Nolan and Sofia Coppola urged me and my crying babies to do.
I guess it depends on what you consider an "error", which can depend on context. I am definitely more comfortable with the statement "Returning Null to indicate a catastrophic failure condition is unacceptable". I am currently unaware of any languages that do not allow you to abort ( or something equivalent).
The original article's problem with returning null, was the requirement to check the return value for NULL that is created. And yeah, I agree that's not ideal, but my point was that there is not *always* a much better alternative (i.e. *sometimes* it is acceptable).
Furthermore, there are ways of indicating an error other than returning NULL that still require a check equally as onerous(and easy to neglect) as a NULL check. So the problem isn't just returning NULL, but error-prone error-handling systems in general.
You can easily protect data with encryption. It's harder to protect meta data. For example: with proper encryption we may never know what Devin Nunes was actually watching on pornhub. To actually hide that Devin Nunes was on pornhub requires something like TOR or a VPN.
Some might abstain from selling customer data. But there are lots of politicians and lots of ISPs. Surely *some* family values congressmen will get disgraced as perverts. We may also catch some of them googling stuff like "How to launder money" or "How to discreetly ask for a bribe" buried deep in all the internet porn searches.
There certainly is some truth to it, but it is a bad argument against mechanisms that prevent mistakes.
Unfortunately, everyone makes mistakes. This is why such a huge amount of software development effort is put into mechanisms preventing human errors that would otherwise have occurred. It's not like these mechanisms are just for the bad programmers, and Knuth and Dijkstra don't need to use them. It also frees human time for real problem solving that would have otherwise been spent bug hunting.
My claim is that "null is acceptible" is not *always* a lie. *Sometimes* it is either the best solution or at least one of the better solutions, given the circumstance. I am not claiming that returning NULL especially in languages with better constructs to handle errors.
Unless you're talking about the performance implications in C++,
That is one of the cases, yes.
I did not mean to imply NULL be used as a hack to avoid proper error handling, and I will not infer that you are making this claim.
All I am saying is that in some very specific cases returning NULL may be the least (or one of the least) bad choices. You are still responsible for performing due diligence in exhausting the search for better options before settling on returning NULL.
The purpose of my comment was to maybe discourage people from choosing a worse option than returning NULL because of the "rule" that "returning NULL is always bad". It is important to understand the reasons for rules, so you have a better understanding of when they no longer apply.
I am not the OP. I can only infer that he/she meant median (because that's what most people mean in this context). I used the same terminology as the OP in an attempt to make it clearer how my claim was intended to relate to his/her claim. It should also be obvious that I was also referring to a median given the definition of a median. And you are free to infer that my claim is only valid for a median average.
I don't begrudge you for nitpicking, especially since that was what I was basically doing. I just think my "nitpick" may actually have some actual value, rather than only being a nitpick. My intention wasn't to nitpick even though it kind of ended up that way.
Obviously your function should be well documented and actually do what it says it will do. And if it's possible to ensure that a function can never fail, and can forcibly disallow NULL returns through some language construct, then by all means do that.
What *should* your function do, in the event that something goes wrong enough that returning a "normal" object is not possible.
For example: you need a function that loads an image from a filename and returns a bitmap. There are lots of things that can go wrong. You can fail to allocate enough memory. You can discover that the image file is malformed. You can fail to read the file.
There are many other ways of dealing with this problem than returning a NULL pointer in place of an object pointer, but depending on the language those other ways may not actually be much or any better. So I'm not saying that returning NULLs is always good. I'm saying that it might need to be adequate given the alternatives.
Average is a general term that can refer to the mean or the median among other things. I would argue that phrase " is better than the average " implies a median average rather than a mean.
"If something seems stupid to me, it is, because I am definitely smart."
Example:
"Smart pointers are stupid. I know how to manage memory, and anyone who can't have zero memory leaks directly using malloc and free shouldn't be coding anyway."
This is not mathematically true.
A = set of all programmers. |A| = 100%
W = set of programmers that are worse than average. |W| = 50%
B = set of all programmers that believe they are better than average. |B| = ?%
(W intersect B) is a subset of W and therefore |W intersect B| <= |W|
Even if every programmer thinks they are better than average (|B| is 100%) then only half of programmers are lying (not most) to themselves.
I, however, 100% agree with this sentiment. From my experience, I would guess that |W intersect B| is shockingly close to |W|, 50% (the theoretical maximum). Or to put it another way most bad coders believe the lie that they are better than average.
"Null is acceptable"
Yes null is problematic. What's the solution? Some languages (like java) allow you to use notnull annotations, etc, but what do you do when something went wrong in a function and you don't have an object to return? Sometimes you can just abort (I'd rather have a nice core dump than a messy memory bug), but this is not always appropriate. You can use exceptions, but exceptions come with their own problems that are worse than NULL pointers in many cases. So I would say NULL is acceptable (not good), when the alternatives are something worse or nothing.
"Human relationships can be codified"
They absolutely can be codified. Geneology software is pretty good at doing just this. It is not a hard problem. Even the article suggests that the rel problem is that the software may not properly reflect local laws due to ambiguities in the laws. If the title were "Ambiguous laws written by politicians can be codified", it would be a well accepted fact rather than a lie.
Slashdot Mirror
The flaw in this tactic is that it requires the person discussed on the billboard to be able to feel shame at the things they do in their official capacity.
I don't think that was the goal.
I think the goal is to sway the mind of voters. That doesn't necessarily mean changing a die hard republican into a die hard democrat, but it can be as simple as convincing a tepid republican voter to become more apathetic and just staying home, or convincing a tepid democrat voter to actually go and vote for something they already believe in rather than being lazy.
For me it's not about jumping ship. I already use linux everyday for work. I only have 2 machines that still use windows. My home desktop for gaming, and one work computer that is only used for outlook, and our IT department is already looking into ways to get rid of exchange.
They might very well do what you are saying and restrict users even further, I don't think it's a good idea or a bad idea, because I don't think it really matters what they do anymore.
I'm not some linux zealot. I don't think that the people who say windows is completely obsolete are being truthful either. They neglect that windows is currently the best solution for many situations. What I am saying is that the niche they are living in is getting smaller and smaller, and one day it will disappear.
Game developers are going to do whatever makes them the most money. For the last couple decades, that has meant treating windows as the primary game platform. I think Valve has the right idea in terms of starting to look for a way of this sinking ship. Tools for developing cross platform games are getting better and better. I think MS trying to over-exploit their past dominance has a very good chance of hastening their demise rather than prolonging it.
No doubt MS would love to "get young at a young age to not use ", but they have had limited and diminishing success at this.
Why would they change course?
That's like asking, why wouldn't they go from offering a product with 1 positive feature to a product with 0 positive features.
The only reason people like and use windows is because it is a convenient vessel through which they can easily run the programs they want to run. Take that away, and there is literally no more reason to use it at all.
If I can't play steam games, and run obscure utilities that were only built for windows, I will just use linux.
Windows is hanging on by a thread. They completely lost the mobile moarket, and the desktop market is shrinking. More and more things that you would do "on windows" are being done in the cloud.
Steam has decided windows is a dying platform for games, and has started supporting more and more games on linux.
Is windows going to die tomorrow? No. But there are just too many many good alternatives that are getting better everyday.
It's not like charismatic people need to be put *somewhere* in any given company. There are plenty of unemployed people. If they were actually so bad for these companies to warrant putting them in the positions of least harm potential, we may as well make all the humble unemployed people CEO's, and fire all the narcissistic CEO's to reduce harm even further.
I guess it depends on your standards. If your standards are low enough, then all of humanity is really smart.
There are different types of simplicity. There is simplicity involved in creating something and the simplicity of using it, and these 2 things are often opposed to each other. Raw pointers are simpler to create than smart pointers (the language feature, not an instance). They are easier to understand than smart pointers. They are harder to use correctly than smart pointers. How many bugs are a result of improper pointer use?
And if you look at code that uses raw pointers and the same code that has been modified to use smart pointers, the code that uses smart pointers will look simpler. It will look the same minus all the delete/free calls, and minus some NULL checks.
So yes, smart pointers are more complex in some sense. But if you consider the problem of enforcing the proper use of pointers, rather than just deferring that problem to the next programmer, smart pointers are a very simple solution to this problem (e.g. as opposed to something like garbage collection).
Not solving problems is simpler than solving them. I am all for not solving problems that don't need to be solved. But preventing pointer misuse seems like a problem worth solving, and it is solved elegantly with smart pointers.
This is just one example of something you can do with a more complex language like C++ that you can't do in C.
I don't think you can call yourself a C++ expert and be "totally unable to deal with" a C++ codebase that uses a different subset of C++ than what you would have used.
You can certainly be a C++ expert and be totally unable to deal with a very poorly designed codebase, but that is true for every language (even C).
Maybe there are more experts in the C language, because it is a simpler language and therefore an easier language to become an expert at. I would be totally in favor of using a simpler language for a project if there were no benefit to any of the more advanced features provided in C++, but that is rarely the case.
The 4 things I listed are insanely powerful tools for making safer and more maintainable code, and these things come with little to no performance overhead.
It also has good features, many of which provide you with an alternative to the bad features of C.
Constructors/destructors (RAII) > manual initialization/deinitialization.
Smart Pointers (made possible by RAII) > raw pointers
Polymorphism > function pointers
Templates > macros
If you are subscribing to a streaming movie service and you have the choice between netflix and a site that only allows you to watch "Armageddon (1998)", does it make sense to choose the latter because netflix has more bad movies? No of course not, because you don't have to watch the bad movies on netflix, and you can even choose only to watch movies better than Armageddon.
C++ is netflix. Nobody watches all the movies. Everyone watches the movies that are good from their point of view (even though many of those people are just wrong).
That doesn't mean it's not hard. That just means it's a hard thing that no one needs to do anymore for any reason other than... educational purposes.
I'm not saying not to use the tool. Use the tool. Maybe running as fast as Usain Bolt is hard for some people. I use a tool called a car. Running is super easy for me.
Running isn't hard either. Most people can run. It's only hard when you say something like "You need to be able to run a 5 minute mile". Some math is not hard just like some running is not hard. If you lower your standards enough then you can make it so math is not hard for 95% of people. You can also raise your standards enough to make it hard for 99.99% of people.
Even seemingly simple things like algebra can be really hard. I have had many calculus classes where the actual calculus wasn't so difficult, but the algebra was a nightmare. There is also nightmare calculus as well.
Only if some famous celebrities urged you to do it.
I've been to a few theaters that are basically restaurants, with menus waiters, tables, and reclining chairs. I thought it would be annoying to have everybody ordering and eating food during the movie, but the seats were far enough apart to not be very distracted but what other people were doing, and the cost was probably high enough to preclude idiot teenagers and adults who act like idiot teenagers.
I don't really go to the movies anymore because I have 2 screaming babies, but once a year when we hire baby sitters and go out, I appreciate the more luxurious experience. I am already blowing at least $100 on a babysitting, trying to save money by being next to some middleschool shithead talking loudly through the whole movie seems like a waste.
I don't need super crazy sound. I don't need 3D. I just want some comfortable chairs, good food, low distraction.
I'm so sorry you have a problem with my crying babies, but I'm just doing what Christopher Nolan and Sofia Coppola urged me and my crying babies to do.
I don't think we have different opinions on this.
I don't think we do either.
I guess it depends on what you consider an "error", which can depend on context. I am definitely more comfortable with the statement "Returning Null to indicate a catastrophic failure condition is unacceptable". I am currently unaware of any languages that do not allow you to abort ( or something equivalent).
The original article's problem with returning null, was the requirement to check the return value for NULL that is created. And yeah, I agree that's not ideal, but my point was that there is not *always* a much better alternative (i.e. *sometimes* it is acceptable).
Furthermore, there are ways of indicating an error other than returning NULL that still require a check equally as onerous(and easy to neglect) as a NULL check. So the problem isn't just returning NULL, but error-prone error-handling systems in general.
You can easily protect data with encryption. It's harder to protect meta data. For example: with proper encryption we may never know what Devin Nunes was actually watching on pornhub. To actually hide that Devin Nunes was on pornhub requires something like TOR or a VPN.
Some might abstain from selling customer data. But there are lots of politicians and lots of ISPs. Surely *some* family values congressmen will get disgraced as perverts. We may also catch some of them googling stuff like "How to launder money" or "How to discreetly ask for a bribe" buried deep in all the internet porn searches.
There certainly is some truth to it, but it is a bad argument against mechanisms that prevent mistakes.
Unfortunately, everyone makes mistakes. This is why such a huge amount of software development effort is put into mechanisms preventing human errors that would otherwise have occurred. It's not like these mechanisms are just for the bad programmers, and Knuth and Dijkstra don't need to use them. It also frees human time for real problem solving that would have otherwise been spent bug hunting.
My claim is that "null is acceptible" is not *always* a lie. *Sometimes* it is either the best solution or at least one of the better solutions, given the circumstance. I am not claiming that returning NULL especially in languages with better constructs to handle errors.
Unless you're talking about the performance implications in C++,
That is one of the cases, yes.
I did not mean to imply NULL be used as a hack to avoid proper error handling, and I will not infer that you are making this claim.
All I am saying is that in some very specific cases returning NULL may be the least (or one of the least) bad choices. You are still responsible for performing due diligence in exhausting the search for better options before settling on returning NULL.
The purpose of my comment was to maybe discourage people from choosing a worse option than returning NULL because of the "rule" that "returning NULL is always bad". It is important to understand the reasons for rules, so you have a better understanding of when they no longer apply.
I am not the OP. I can only infer that he/she meant median (because that's what most people mean in this context). I used the same terminology as the OP in an attempt to make it clearer how my claim was intended to relate to his/her claim. It should also be obvious that I was also referring to a median given the definition of a median. And you are free to infer that my claim is only valid for a median average.
I don't begrudge you for nitpicking, especially since that was what I was basically doing. I just think my "nitpick" may actually have some actual value, rather than only being a nitpick. My intention wasn't to nitpick even though it kind of ended up that way.
Obviously your function should be well documented and actually do what it says it will do. And if it's possible to ensure that a function can never fail, and can forcibly disallow NULL returns through some language construct, then by all means do that.
What *should* your function do, in the event that something goes wrong enough that returning a "normal" object is not possible.
For example: you need a function that loads an image from a filename and returns a bitmap. There are lots of things that can go wrong. You can fail to allocate enough memory. You can discover that the image file is malformed. You can fail to read the file.
There are many other ways of dealing with this problem than returning a NULL pointer in place of an object pointer, but depending on the language those other ways may not actually be much or any better. So I'm not saying that returning NULLs is always good. I'm saying that it might need to be adequate given the alternatives.
Average is a general term that can refer to the mean or the median among other things. I would argue that phrase " is better than the average " implies a median average rather than a mean.
"If something seems stupid to me, it is, because I am definitely smart."
Example:
"Smart pointers are stupid. I know how to manage memory, and anyone who can't have zero memory leaks directly using malloc and free shouldn't be coding anyway."
A = set of all programmers. |A| = 100%
W = set of programmers that are worse than average. |W| = 50%
B = set of all programmers that believe they are better than average. |B| = ?%
(W intersect B) is a subset of W and therefore |W intersect B| <= |W|
Even if every programmer thinks they are better than average (|B| is 100%) then only half of programmers are lying (not most) to themselves.
I, however, 100% agree with this sentiment. From my experience, I would guess that |W intersect B| is shockingly close to |W|, 50% (the theoretical maximum). Or to put it another way most bad coders believe the lie that they are better than average.
"Null is acceptable"
Yes null is problematic. What's the solution? Some languages (like java) allow you to use notnull annotations, etc, but what do you do when something went wrong in a function and you don't have an object to return? Sometimes you can just abort (I'd rather have a nice core dump than a messy memory bug), but this is not always appropriate. You can use exceptions, but exceptions come with their own problems that are worse than NULL pointers in many cases. So I would say NULL is acceptable (not good), when the alternatives are something worse or nothing.
"Human relationships can be codified"
They absolutely can be codified. Geneology software is pretty good at doing just this. It is not a hard problem. Even the article suggests that the rel problem is that the software may not properly reflect local laws due to ambiguities in the laws. If the title were "Ambiguous laws written by politicians can be codified", it would be a well accepted fact rather than a lie.